Hello
I'm Tan Shaopeng from Fujitsu Limited.
I$B!G(Bm trying to implement Fujitsu A64FX$B!G(Bs cache related features.
It is a cache partitioning function we called sector cache function
that using the value of the tag that is upper 8 bits of the 64bit
address and the value of the sector cache register to control virtual
cache capacity of the L1D&L2 cache.
A few days ago, when I sent a driver that realizes this function to
ARM64 kernel community, Will Deacon and Arnd Bergmann suggested
an idea to add the sector cache function of A64FX into resctrl.
https://lore.kernel.org/linux-arm-kernel/CAK8P3a2pFcNTw9NpRtQfYr7A5OcZ=As2kM0D_sbfFcGQ_J2Q+Q@mail.gmail.com/
Based on my study, I think the sector cache function of A64FX can be
added into the allocation features of resctrl after James' resctrl
rework has finished. But, in order to implement this function,
more interfaces for resctrl are need. The details are as follow,
and could you give me some advice?
[Sector cache function]
The sector cache function split cache into multiple sectors and
control them separately. It is implemented on the L1D cache and
L2 cache in the A64FX processor and can be controlled individually
for L1D cache and L2 cache. A64FX has no L3 cache. Each L1D cache
and L2 cache has 4 sectors. Which L1D sector is used is specified
by the value of [57:56] bits of address, how many ways of sector
are specified by the value of register (IMP_SCCR_L1_EL0).
Which L2 sector is used is specified by the value of [56] bits of
address, and how many ways of sector are specified by value of register
(IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1, IMP_SCCR_SET1_L2_EL1).
For more details of sector cache function,
see A64FX HPC extension specification (1.2. Sector cache) in
https://github.com/fujitsu/A64FX
[Difference between resctrl(CAT) and this sector cache function]
L2/L3 CAT (Cache Allocation Technology) enables the user to specify
some physical partition of cache space that an application can fill.
A64FX's L1D/L2 cache has 4 sectors and 16ways. This sector function
enables a user to specify number of ways each sector uses.
Therefore, for CAT it is enough to specify a cache portion for
each cache_id (socket). On the other hand, sector cache needs to
specify cache portion of each sector for each cache_id, and following
extension to resctrl interface is needed to support sector cache.
[Idear for A64FX sector cache function control interface (schemata file details)]
L1:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
L2:<cache_id0>=>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
$B!&(BL1: Add a new interface to control the L1D cache.
$B!&(B<cwbm>,<cwbm>,<cwbm>,<cwbm>$B!'(BSpecify the number of ways for each sector.
$B!&(Bcwbm$B!'(BSpecify the number of ways in each sector as a bitmap (percentage),
but the bitmap does not indicate the location of the cache.
* In the sector cache function, L2 sector cache way setting register is
shared among PEs (Processor Element) in shared domain. If two PEs
which share L2 cache belongs to different resource groups, one resource
group's L2 setting will affect to other resource group's L2 setting.
* Since A64FX does not support MPAM, it is not necessary to consider
how to switch between MPAM and sector cache function now.
Some questions:
1.I'm still studying about RDT, could you tell me whether RDT has
the similar mechanism with sector cache function?
2.In RDT, L3 cache is shared among cores in socket. If two cores which
share L3 cache belongs to different resource groups, one resource
group's L3 setting will affect to other resource group's L3 setting?
3.Is this approach acceptable? could you give me some advice?
Best regards
Tan Shaopeng
Hi,
Ping... any comments&advice about add an A64FX cache control function into resctrl?
Best regards
Tan Shaopeng
> Hello
>
>
> I'm Tan Shaopeng from Fujitsu Limited.
>
> I$B!G(Bm trying to implement Fujitsu A64FX$B!G(Bs cache related features.
> It is a cache partitioning function we called sector cache function that using
> the value of the tag that is upper 8 bits of the 64bit address and the value of the
> sector cache register to control virtual cache capacity of the L1D&L2 cache.
>
> A few days ago, when I sent a driver that realizes this function to
> ARM64 kernel community, Will Deacon and Arnd Bergmann suggested an idea
> to add the sector cache function of A64FX into resctrl.
> https://lore.kernel.org/linux-arm-kernel/CAK8P3a2pFcNTw9NpRtQfYr7A5Oc
> [email protected]/
>
> Based on my study, I think the sector cache function of A64FX can be added
> into the allocation features of resctrl after James' resctrl rework has finished.
> But, in order to implement this function, more interfaces for resctrl are need.
> The details are as follow, and could you give me some advice?
>
> [Sector cache function]
> The sector cache function split cache into multiple sectors and control them
> separately. It is implemented on the L1D cache and
> L2 cache in the A64FX processor and can be controlled individually for L1D
> cache and L2 cache. A64FX has no L3 cache. Each L1D cache and L2 cache
> has 4 sectors. Which L1D sector is used is specified by the value of [57:56] bits
> of address, how many ways of sector are specified by the value of register
> (IMP_SCCR_L1_EL0).
> Which L2 sector is used is specified by the value of [56] bits of address, and
> how many ways of sector are specified by value of register
> (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
> IMP_SCCR_SET1_L2_EL1).
>
> For more details of sector cache function, see A64FX HPC extension
> specification (1.2. Sector cache) in https://github.com/fujitsu/A64FX
>
> [Difference between resctrl(CAT) and this sector cache function]
> L2/L3 CAT (Cache Allocation Technology) enables the user to specify some
> physical partition of cache space that an application can fill.
> A64FX's L1D/L2 cache has 4 sectors and 16ways. This sector function enables
> a user to specify number of ways each sector uses.
> Therefore, for CAT it is enough to specify a cache portion for each cache_id
> (socket). On the other hand, sector cache needs to specify cache portion of
> each sector for each cache_id, and following extension to resctrl interface is
> needed to support sector cache.
>
> [Idear for A64FX sector cache function control interface (schemata file
> details)]
> L1:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cw
> bm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
> L2:<cache_id0>=>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=
> <cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
>
> $B!&(BL1: Add a new interface to control the L1D cache.
> $B!&(B<cwbm>,<cwbm>,<cwbm>,<cwbm>$B!'(BSpecify the number of ways for each
> sector.
> $B!&(Bcwbm$B!'(BSpecify the number of ways in each sector as a bitmap (percentage),
> but the bitmap does not indicate the location of the cache.
> * In the sector cache function, L2 sector cache way setting register is
> shared among PEs (Processor Element) in shared domain. If two PEs
> which share L2 cache belongs to different resource groups, one resource
> group's L2 setting will affect to other resource group's L2 setting.
> * Since A64FX does not support MPAM, it is not necessary to consider
> how to switch between MPAM and sector cache function now.
>
> Some questions:
> 1.I'm still studying about RDT, could you tell me whether RDT has
> the similar mechanism with sector cache function?
> 2.In RDT, L3 cache is shared among cores in socket. If two cores which
> share L3 cache belongs to different resource groups, one resource
> group's L3 setting will affect to other resource group's L3 setting?
> 3.Is this approach acceptable? could you give me some advice?
>
>
> Best regards
> Tan Shaopeng
Hi Tan Shaopeng,
On 4/21/2021 1:37 AM, [email protected] wrote:
> Hi,
>
> Ping... any comments&advice about add an A64FX cache control function into resctrl?
My apologies for the delay.
>
> Best regards
> Tan Shaopeng
>
>> Hello
>>
>>
>> I'm Tan Shaopeng from Fujitsu Limited.
>>
>> I$B!G(Bm trying to implement Fujitsu A64FX$B!G(Bs cache related features.
>> It is a cache partitioning function we called sector cache function that using
>> the value of the tag that is upper 8 bits of the 64bit address and the value of the
>> sector cache register to control virtual cache capacity of the L1D&L2 cache.
>>
>> A few days ago, when I sent a driver that realizes this function to
>> ARM64 kernel community, Will Deacon and Arnd Bergmann suggested an idea
>> to add the sector cache function of A64FX into resctrl.
>> https://lore.kernel.org/linux-arm-kernel/CAK8P3a2pFcNTw9NpRtQfYr7A5Oc
>> [email protected]/
>>
>> Based on my study, I think the sector cache function of A64FX can be added
>> into the allocation features of resctrl after James' resctrl rework has finished.
>> But, in order to implement this function, more interfaces for resctrl are need.
>> The details are as follow, and could you give me some advice?
>>
>> [Sector cache function]
>> The sector cache function split cache into multiple sectors and control them
>> separately. It is implemented on the L1D cache and
>> L2 cache in the A64FX processor and can be controlled individually for L1D
>> cache and L2 cache. A64FX has no L3 cache. Each L1D cache and L2 cache
>> has 4 sectors. Which L1D sector is used is specified by the value of [57:56] bits
>> of address, how many ways of sector are specified by the value of register
>> (IMP_SCCR_L1_EL0).
>> Which L2 sector is used is specified by the value of [56] bits of address, and
>> how many ways of sector are specified by value of register
>> (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
>> IMP_SCCR_SET1_L2_EL1).
>>
>> For more details of sector cache function, see A64FX HPC extension
>> specification (1.2. Sector cache) in https://github.com/fujitsu/A64FX
The overview in section 12 was informative but very high level.
I was not able to find any instance of "IMP_SCCR" in this document to
explore how this cache allocation works.
Are these cache sectors exposed to the OS in any way? For example, when
the OS discovers the cache, does it learn about these sectors and expose
the details to user space (/sys/devices/system/cpuX/cache)?
The overview of Sector Cache in that document provides details of how
the size of the sector itself is dynamically adjusted to usage. That
description is quite cryptic but it seems like a sector, since the
number of ways associated with it can dynamically change, is more
equivalent to a class of service or resource group in the resctrl
environment.
I really may be interpreting things wrong here, could you perhaps point
me to where I can obtain more details?
>> [Difference between resctrl(CAT) and this sector cache function]
>> L2/L3 CAT (Cache Allocation Technology) enables the user to specify some
>> physical partition of cache space that an application can fill.
>> A64FX's L1D/L2 cache has 4 sectors and 16ways. This sector function enables
>> a user to specify number of ways each sector uses.
>> Therefore, for CAT it is enough to specify a cache portion for each cache_id
>> (socket). On the other hand, sector cache needs to specify cache portion of
>> each sector for each cache_id, and following extension to resctrl interface is
>> needed to support sector cache.
>>
>> [Idear for A64FX sector cache function control interface (schemata file
>> details)]
>> L1:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cw
>> bm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
>> L2:<cache_id0>=>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=
>> <cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
>>
>> $B!&(BL1: Add a new interface to control the L1D cache.
>> $B!&(B<cwbm>,<cwbm>,<cwbm>,<cwbm>$B!'(BSpecify the number of ways for each
>> sector.
>> $B!&(Bcwbm$B!'(BSpecify the number of ways in each sector as a bitmap (percentage),
>> but the bitmap does not indicate the location of the cache.
>> * In the sector cache function, L2 sector cache way setting register is
>> shared among PEs (Processor Element) in shared domain. If two PEs
>> which share L2 cache belongs to different resource groups, one resource
>> group's L2 setting will affect to other resource group's L2 setting.
In resctrl a "resource group" can be viewed as a class of service.
>> * Since A64FX does not support MPAM, it is not necessary to consider
>> how to switch between MPAM and sector cache function now.
>>
>> Some questions:
>> 1.I'm still studying about RDT, could you tell me whether RDT has
>> the similar mechanism with sector cache function?
This is not clear to me yet. One thing to keep in mind is that a bit in
the capacity bitmask could correspond to some number of ways in a cache,
but it does not have to. It is essentially a hint to hardware on how
much cache space needs to be allocated while also indicating overlap and
isolation from other allocations.
resctrl already supports the bitmask being interpreted differently
between architectures and with the MPAM support there will be even more
support for different interpretations.
>> 2.In RDT, L3 cache is shared among cores in socket. If two cores which
>> share L3 cache belongs to different resource groups, one resource
>> group's L3 setting will affect to other resource group's L3 setting?
This question is not entirely clear to me. Are you referring to the
hardware layout or configuration changes via the resctrl "cpus" file?
Each resource group is a class of service (CLOS) that is supported by
all cache instances. By default each resource group would thus contain
all cache instances on the system (even if some cache instances do not
support the same number of CLOS resctrl would only support the CLOS
supported by all resources).
Reinette
Hi Reinette,
> On 4/21/2021 1:37 AM, [email protected] wrote:
> > Hi,
> >
> > Ping... any comments&advice about add an A64FX cache control function
> into resctrl?
>
> My apologies for the delay.
>
> >
> > Best regards
> > Tan Shaopeng
> >
> >> Hello
> >>
> >>
> >> I'm Tan Shaopeng from Fujitsu Limited.
> >>
> >> I$B!G(Bm trying to implement Fujitsu A64FX$B!G(Bs cache related features.
> >> It is a cache partitioning function we called sector cache function
> >> that using the value of the tag that is upper 8 bits of the 64bit
> >> address and the value of the sector cache register to control virtual cache
> capacity of the L1D&L2 cache.
> >>
> >> A few days ago, when I sent a driver that realizes this function to
> >> ARM64 kernel community, Will Deacon and Arnd Bergmann suggested an
> >> idea to add the sector cache function of A64FX into resctrl.
> >>
> https://lore.kernel.org/linux-arm-kernel/CAK8P3a2pFcNTw9NpRtQfYr7A5Oc
> >> [email protected]/
> >>
> >> Based on my study, I think the sector cache function of A64FX can be
> >> added into the allocation features of resctrl after James' resctrl rework has
> finished.
> >> But, in order to implement this function, more interfaces for resctrl are
> need.
> >> The details are as follow, and could you give me some advice?
> >>
> >> [Sector cache function]
> >> The sector cache function split cache into multiple sectors and
> >> control them separately. It is implemented on the L1D cache and
> >> L2 cache in the A64FX processor and can be controlled individually
> >> for L1D cache and L2 cache. A64FX has no L3 cache. Each L1D cache and
> >> L2 cache has 4 sectors. Which L1D sector is used is specified by the
> >> value of [57:56] bits of address, how many ways of sector are
> >> specified by the value of register (IMP_SCCR_L1_EL0).
> >> Which L2 sector is used is specified by the value of [56] bits of
> >> address, and how many ways of sector are specified by value of
> >> register (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
> >> IMP_SCCR_SET1_L2_EL1).
> >>
> >> For more details of sector cache function, see A64FX HPC extension
> >> specification (1.2. Sector cache) in https://github.com/fujitsu/A64FX
>
> The overview in section 12 was informative but very high level.
> I was not able to find any instance of "IMP_SCCR" in this document to explore
> how this cache allocation works.
>
> Are these cache sectors exposed to the OS in any way? For example, when the
> OS discovers the cache, does it learn about these sectors and expose the
> details to user space (/sys/devices/system/cpuX/cache)?
>
> The overview of Sector Cache in that document provides details of how the size
> of the sector itself is dynamically adjusted to usage. That description is quite
> cryptic but it seems like a sector, since the number of ways associated with it
> can dynamically change, is more equivalent to a class of service or resource
> group in the resctrl environment.
>
> I really may be interpreting things wrong here, could you perhaps point me to
> where I can obtain more details?
>
>
> >> [Difference between resctrl(CAT) and this sector cache function]
> >> L2/L3 CAT (Cache Allocation Technology) enables the user to specify
> >> some physical partition of cache space that an application can fill.
> >> A64FX's L1D/L2 cache has 4 sectors and 16ways. This sector function
> >> enables a user to specify number of ways each sector uses.
> >> Therefore, for CAT it is enough to specify a cache portion for each
> >> cache_id (socket). On the other hand, sector cache needs to specify
> >> cache portion of each sector for each cache_id, and following
> >> extension to resctrl interface is needed to support sector cache.
> >>
> >> [Idear for A64FX sector cache function control interface (schemata
> >> file details)]
> >>
> L1:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cw
> >> bm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
> >>
> L2:<cache_id0>=>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=
> >> <cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
> >>
> >> $B!&(BL1: Add a new interface to control the L1D cache.
> >> $B!&(B<cwbm>,<cwbm>,<cwbm>,<cwbm>$B!'(BSpecify the number of ways for
> each
> >> sector.
> >> $B!&(Bcwbm$B!'(BSpecify the number of ways in each sector as a bitmap
> (percentage),
> >> but the bitmap does not indicate the location of the cache.
> >> * In the sector cache function, L2 sector cache way setting register is
> >> shared among PEs (Processor Element) in shared domain. If two PEs
> >> which share L2 cache belongs to different resource groups, one
> resource
> >> group's L2 setting will affect to other resource group's L2 setting.
>
> In resctrl a "resource group" can be viewed as a class of service.
>
> >> * Since A64FX does not support MPAM, it is not necessary to consider
> >> how to switch between MPAM and sector cache function now.
> >>
> >> Some questions:
> >> 1.I'm still studying about RDT, could you tell me whether RDT has
> >> the similar mechanism with sector cache function?
>
> This is not clear to me yet. One thing to keep in mind is that a bit in the capacity
> bitmask could correspond to some number of ways in a cache, but it does not
> have to. It is essentially a hint to hardware on how much cache space needs to
> be allocated while also indicating overlap and isolation from other allocations.
>
> resctrl already supports the bitmask being interpreted differently between
> architectures and with the MPAM support there will be even more support for
> different interpretations.
>
> >> 2.In RDT, L3 cache is shared among cores in socket. If two cores which
> >> share L3 cache belongs to different resource groups, one resource
> >> group's L3 setting will affect to other resource group's L3 setting?
>
> This question is not entirely clear to me. Are you referring to the hardware layout
> or configuration changes via the resctrl "cpus" file?
>
> Each resource group is a class of service (CLOS) that is supported by all cache
> instances. By default each resource group would thus contain all cache
> instances on the system (even if some cache instances do not support the
> same number of CLOS resctrl would only support the CLOS supported by all
> resources).
Thanks for your comment.
I am sorry that the description about the sector cache function was
difficult to understand. Since all public specifications were shown
in the URL, please give me some time, I will organize the contents of
64FX cache control function.
Best regards,
Tan Shaopeng
Hi Reinette,
> On 4/21/2021 1:37 AM, [email protected] wrote:
> > Hi,
> >
> > Ping... any comments&advice about add an A64FX cache control function
> into resctrl?
>
> My apologies for the delay.
>
> >
> > Best regards
> > Tan Shaopeng
> >
> >> Hello
> >>
> >>
> >> I'm Tan Shaopeng from Fujitsu Limited.
> >>
> >> I$B!G(Bm trying to implement Fujitsu A64FX$B!G(Bs cache related features.
> >> It is a cache partitioning function we called sector cache function
> >> that using the value of the tag that is upper 8 bits of the 64bit
> >> address and the value of the sector cache register to control virtual cache
> capacity of the L1D&L2 cache.
> >>
> >> A few days ago, when I sent a driver that realizes this function to
> >> ARM64 kernel community, Will Deacon and Arnd Bergmann suggested an
> >> idea to add the sector cache function of A64FX into resctrl.
> >>
> https://lore.kernel.org/linux-arm-kernel/CAK8P3a2pFcNTw9NpRtQfYr7A5Oc
> >> [email protected]/
> >>
> >> Based on my study, I think the sector cache function of A64FX can be
> >> added into the allocation features of resctrl after James' resctrl rework has
> finished.
> >> But, in order to implement this function, more interfaces for resctrl are
> need.
> >> The details are as follow, and could you give me some advice?
> >>
> >> [Sector cache function]
> >> The sector cache function split cache into multiple sectors and
> >> control them separately. It is implemented on the L1D cache and
> >> L2 cache in the A64FX processor and can be controlled individually
> >> for L1D cache and L2 cache. A64FX has no L3 cache. Each L1D cache and
> >> L2 cache has 4 sectors. Which L1D sector is used is specified by the
> >> value of [57:56] bits of address, how many ways of sector are
> >> specified by the value of register (IMP_SCCR_L1_EL0).
> >> Which L2 sector is used is specified by the value of [56] bits of
> >> address, and how many ways of sector are specified by value of
> >> register (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
> >> IMP_SCCR_SET1_L2_EL1).
> >>
> >> For more details of sector cache function, see A64FX HPC extension
> >> specification (1.2. Sector cache) in https://github.com/fujitsu/A64FX
>
> The overview in section 12 was informative but very high level.
I'm considering how to answer your questions from your email which
I received before, when I check the email again, I am sorry that
the information I provided before are insufficient.
To understand the sector cache function of A64FX, could you please see
A64FX_Microarchitecture_Manual - section 12. Sector Cache
https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Microarchitecture_Manual_en_1.4.pdf
and,
A64FX_Specification_HPC_Extension ? section 1.2. Sector Cache
https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Specification_HPC_Extension_v1_EN.pdf
In addition, Japan will be on a long holiday about one week from
April 29th, I will answer your other questions after the holidays.
> I was not able to find any instance of "IMP_SCCR" in this document to explore
> how this cache allocation works.
>
> Are these cache sectors exposed to the OS in any way? For example, when the
> OS discovers the cache, does it learn about these sectors and expose the
> details to user space (/sys/devices/system/cpuX/cache)?
>
> The overview of Sector Cache in that document provides details of how the size
> of the sector itself is dynamically adjusted to usage. That description is quite
> cryptic but it seems like a sector, since the number of ways associated with it
> can dynamically change, is more equivalent to a class of service or resource
> group in the resctrl environment.
>
> I really may be interpreting things wrong here, could you perhaps point me to
> where I can obtain more details?
>
>
> >> [Difference between resctrl(CAT) and this sector cache function]
> >> L2/L3 CAT (Cache Allocation Technology) enables the user to specify
> >> some physical partition of cache space that an application can fill.
> >> A64FX's L1D/L2 cache has 4 sectors and 16ways. This sector function
> >> enables a user to specify number of ways each sector uses.
> >> Therefore, for CAT it is enough to specify a cache portion for each
> >> cache_id (socket). On the other hand, sector cache needs to specify
> >> cache portion of each sector for each cache_id, and following
> >> extension to resctrl interface is needed to support sector cache.
> >>
> >> [Idear for A64FX sector cache function control interface (schemata
> >> file details)]
> >>
> L1:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cw
> >> bm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
> >>
> L2:<cache_id0>=>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=
> >> <cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
> >>
> >> $B!&(BL1: Add a new interface to control the L1D cache.
> >> $B!&(B<cwbm>,<cwbm>,<cwbm>,<cwbm>$B!'(BSpecify the number of ways for
> each
> >> sector.
> >> $B!&(Bcwbm$B!'(BSpecify the number of ways in each sector as a bitmap
> (percentage),
> >> but the bitmap does not indicate the location of the cache.
> >> * In the sector cache function, L2 sector cache way setting register is
> >> shared among PEs (Processor Element) in shared domain. If two PEs
> >> which share L2 cache belongs to different resource groups, one
> resource
> >> group's L2 setting will affect to other resource group's L2 setting.
>
> In resctrl a "resource group" can be viewed as a class of service.
>
> >> * Since A64FX does not support MPAM, it is not necessary to consider
> >> how to switch between MPAM and sector cache function now.
> >>
> >> Some questions:
> >> 1.I'm still studying about RDT, could you tell me whether RDT has
> >> the similar mechanism with sector cache function?
>
> This is not clear to me yet. One thing to keep in mind is that a bit in the capacity
> bitmask could correspond to some number of ways in a cache, but it does not
> have to. It is essentially a hint to hardware on how much cache space needs to
> be allocated while also indicating overlap and isolation from other allocations.
>
> resctrl already supports the bitmask being interpreted differently between
> architectures and with the MPAM support there will be even more support for
> different interpretations.
>
> >> 2.In RDT, L3 cache is shared among cores in socket. If two cores which
> >> share L3 cache belongs to different resource groups, one resource
> >> group's L3 setting will affect to other resource group's L3 setting?
>
> This question is not entirely clear to me. Are you referring to the hardware layout
> or configuration changes via the resctrl "cpus" file?
>
> Each resource group is a class of service (CLOS) that is supported by all cache
> instances. By default each resource group would thus contain all cache
> instances on the system (even if some cache instances do not support the
> same number of CLOS resctrl would only support the CLOS supported by all
> resources).
Best regards
Tan Shaopeng
Hi Tan Shaopeng,
On 4/28/2021 1:16 AM, [email protected] wrote:
> Hi Reinette,
>
>> On 4/21/2021 1:37 AM, [email protected] wrote:
>>> Hi,
>>>
>>> Ping... any comments&advice about add an A64FX cache control function
>> into resctrl?
>>
>> My apologies for the delay.
>>
>>>
>>> Best regards
>>> Tan Shaopeng
>>>
>>>> Hello
>>>>
>>>>
>>>> I'm Tan Shaopeng from Fujitsu Limited.
>>>>
>>>> I$B!G(Bm trying to implement Fujitsu A64FX$B!G(Bs cache related features.
>>>> It is a cache partitioning function we called sector cache function
>>>> that using the value of the tag that is upper 8 bits of the 64bit
>>>> address and the value of the sector cache register to control virtual cache
>> capacity of the L1D&L2 cache.
>>>>
>>>> A few days ago, when I sent a driver that realizes this function to
>>>> ARM64 kernel community, Will Deacon and Arnd Bergmann suggested an
>>>> idea to add the sector cache function of A64FX into resctrl.
>>>>
>> https://lore.kernel.org/linux-arm-kernel/CAK8P3a2pFcNTw9NpRtQfYr7A5Oc
>>>> [email protected]/
>>>>
>>>> Based on my study, I think the sector cache function of A64FX can be
>>>> added into the allocation features of resctrl after James' resctrl rework has
>> finished.
>>>> But, in order to implement this function, more interfaces for resctrl are
>> need.
>>>> The details are as follow, and could you give me some advice?
>>>>
>>>> [Sector cache function]
>>>> The sector cache function split cache into multiple sectors and
>>>> control them separately. It is implemented on the L1D cache and
>>>> L2 cache in the A64FX processor and can be controlled individually
>>>> for L1D cache and L2 cache. A64FX has no L3 cache. Each L1D cache and
>>>> L2 cache has 4 sectors. Which L1D sector is used is specified by the
>>>> value of [57:56] bits of address, how many ways of sector are
>>>> specified by the value of register (IMP_SCCR_L1_EL0).
>>>> Which L2 sector is used is specified by the value of [56] bits of
>>>> address, and how many ways of sector are specified by value of
>>>> register (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
>>>> IMP_SCCR_SET1_L2_EL1).
>>>>
>>>> For more details of sector cache function, see A64FX HPC extension
>>>> specification (1.2. Sector cache) in https://github.com/fujitsu/A64FX
>>
>> The overview in section 12 was informative but very high level.
>
> I'm considering how to answer your questions from your email which
> I received before, when I check the email again, I am sorry that
> the information I provided before are insufficient.
>
> To understand the sector cache function of A64FX, could you please see
> A64FX_Microarchitecture_Manual - section 12. Sector Cache
> https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Microarchitecture_Manual_en_1.4.pdf
> and,
> A64FX_Specification_HPC_Extension ? section 1.2. Sector Cache
> https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Specification_HPC_Extension_v1_EN.pdf
Thank you for the direct links - I missed that there are two documents
available.
After reading the spec portion it does seem to me even more as though
"sectors" could be considered the same as the resctrl "classes of
service". The Fujitsu hardware supports four sectors that can be
configured with different number of ways using the registers you mention
above. In resctrl this could be considered as hardware that supports
four classes of service and each class of service can be allocated a
different number of ways.
The other part is how hardware knows which sector is being used at any
moment in time. In resctrl that is programmed by writing the active
class of service into needed register at the time the application is
context switched (resctrl_sched_in()). This seems different here since
as you describe the sector is chosen by bits in the address. Even so,
which bits to set in the address needs to be programmed also and I also
understand that there is a "default" sector that can be programmed via
register. Could these be equivalent to what is done currently in resctrl?
(Could you please also consider my original questions?)
>
> In addition, Japan will be on a long holiday about one week from
> April 29th, I will answer your other questions after the holidays.
>
>> I was not able to find any instance of "IMP_SCCR" in this document to explore
>> how this cache allocation works.
>>
>> Are these cache sectors exposed to the OS in any way? For example, when the
>> OS discovers the cache, does it learn about these sectors and expose the
>> details to user space (/sys/devices/system/cpuX/cache)?
>>
>> The overview of Sector Cache in that document provides details of how the size
>> of the sector itself is dynamically adjusted to usage. That description is quite
>> cryptic but it seems like a sector, since the number of ways associated with it
>> can dynamically change, is more equivalent to a class of service or resource
>> group in the resctrl environment.
>>
>> I really may be interpreting things wrong here, could you perhaps point me to
>> where I can obtain more details?
>>
>>
>>>> [Difference between resctrl(CAT) and this sector cache function]
>>>> L2/L3 CAT (Cache Allocation Technology) enables the user to specify
>>>> some physical partition of cache space that an application can fill.
>>>> A64FX's L1D/L2 cache has 4 sectors and 16ways. This sector function
>>>> enables a user to specify number of ways each sector uses.
>>>> Therefore, for CAT it is enough to specify a cache portion for each
>>>> cache_id (socket). On the other hand, sector cache needs to specify
>>>> cache portion of each sector for each cache_id, and following
>>>> extension to resctrl interface is needed to support sector cache.
>>>>
>>>> [Idear for A64FX sector cache function control interface (schemata
>>>> file details)]
>>>>
>> L1:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cw
>>>> bm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
>>>>
>> L2:<cache_id0>=>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=
>>>> <cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
>>>>
>>>> $B!&(BL1: Add a new interface to control the L1D cache.
>>>> $B!&(B<cwbm>,<cwbm>,<cwbm>,<cwbm>$B!'(BSpecify the number of ways for
>> each
>>>> sector.
>>>> $B!&(Bcwbm$B!'(BSpecify the number of ways in each sector as a bitmap
>> (percentage),
>>>> but the bitmap does not indicate the location of the cache.
>>>> * In the sector cache function, L2 sector cache way setting register is
>>>> shared among PEs (Processor Element) in shared domain. If two PEs
>>>> which share L2 cache belongs to different resource groups, one
>> resource
>>>> group's L2 setting will affect to other resource group's L2 setting.
>>
>> In resctrl a "resource group" can be viewed as a class of service.
>>
>>>> * Since A64FX does not support MPAM, it is not necessary to consider
>>>> how to switch between MPAM and sector cache function now.
>>>>
>>>> Some questions:
>>>> 1.I'm still studying about RDT, could you tell me whether RDT has
>>>> the similar mechanism with sector cache function?
>>
>> This is not clear to me yet. One thing to keep in mind is that a bit in the capacity
>> bitmask could correspond to some number of ways in a cache, but it does not
>> have to. It is essentially a hint to hardware on how much cache space needs to
>> be allocated while also indicating overlap and isolation from other allocations.
>>
>> resctrl already supports the bitmask being interpreted differently between
>> architectures and with the MPAM support there will be even more support for
>> different interpretations.
>>
>>>> 2.In RDT, L3 cache is shared among cores in socket. If two cores which
>>>> share L3 cache belongs to different resource groups, one resource
>>>> group's L3 setting will affect to other resource group's L3 setting?
>>
>> This question is not entirely clear to me. Are you referring to the hardware layout
>> or configuration changes via the resctrl "cpus" file?
>>
>> Each resource group is a class of service (CLOS) that is supported by all cache
>> instances. By default each resource group would thus contain all cache
>> instances on the system (even if some cache instances do not support the
>> same number of CLOS resctrl would only support the CLOS supported by all
>> resources).
>
Reinette
>>>> [Sector cache function]
>>>> The sector cache function split cache into multiple sectors and
>>>> control them separately. It is implemented on the L1D cache and
>>>> L2 cache in the A64FX processor and can be controlled individually
>>>> for L1D cache and L2 cache. A64FX has no L3 cache. Each L1D cache and
>>>> L2 cache has 4 sectors. Which L1D sector is used is specified by the
>>>> value of [57:56] bits of address, how many ways of sector are
>>>> specified by the value of register (IMP_SCCR_L1_EL0).
>>>> Which L2 sector is used is specified by the value of [56] bits of
>>>> address, and how many ways of sector are specified by value of
>>>> register (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
>>>> IMP_SCCR_SET1_L2_EL1).
Are A64FX binaries position independent? I.e. could the OS reassign
a running task to a different sector by remapping it to different virtual
addresses during a context switch?
Or is this a static property at task launch?
-Tony
On Thu, Apr 29, 2021 at 05:50:20PM +0000, Luck, Tony wrote:
> >>>> [Sector cache function]
> >>>> The sector cache function split cache into multiple sectors and
> >>>> control them separately. It is implemented on the L1D cache and
> >>>> L2 cache in the A64FX processor and can be controlled individually
> >>>> for L1D cache and L2 cache. A64FX has no L3 cache. Each L1D cache and
> >>>> L2 cache has 4 sectors. Which L1D sector is used is specified by the
> >>>> value of [57:56] bits of address, how many ways of sector are
> >>>> specified by the value of register (IMP_SCCR_L1_EL0).
> >>>> Which L2 sector is used is specified by the value of [56] bits of
> >>>> address, and how many ways of sector are specified by value of
> >>>> register (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
> >>>> IMP_SCCR_SET1_L2_EL1).
>
> Are A64FX binaries position independent? I.e. could the OS reassign
> a running task to a different sector by remapping it to different virtual
> addresses during a context switch?
Arm64 supports a maximum of 52-bit of virtual or physical addresses. The
maximum the MMU would produce would be a 52-bit output address. I
presume bits 56, 57 of the address bus are used for some cache affinity
(sector selection) but they don't influence the memory addressing, nor
could the MMU set them.
--
Catalin
Hi, Tony, Catalin
> On Thu, Apr 29, 2021 at 05:50:20PM +0000, Luck, Tony wrote:
> > >>>> [Sector cache function]
> > >>>> The sector cache function split cache into multiple sectors and
> > >>>> control them separately. It is implemented on the L1D cache and
> > >>>> L2 cache in the A64FX processor and can be controlled
> > >>>> individually for L1D cache and L2 cache. A64FX has no L3 cache.
> > >>>> Each L1D cache and
> > >>>> L2 cache has 4 sectors. Which L1D sector is used is specified by
> > >>>> the value of [57:56] bits of address, how many ways of sector are
> > >>>> specified by the value of register (IMP_SCCR_L1_EL0).
> > >>>> Which L2 sector is used is specified by the value of [56] bits of
> > >>>> address, and how many ways of sector are specified by value of
> > >>>> register (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
> > >>>> IMP_SCCR_SET1_L2_EL1).
> >
> > Are A64FX binaries position independent? I.e. could the OS reassign a
> > running task to a different sector by remapping it to different
> > virtual addresses during a context switch?
>
> Arm64 supports a maximum of 52-bit of virtual or physical addresses. The
> maximum the MMU would produce would be a 52-bit output address. I
> presume bits 56, 57 of the address bus are used for some cache affinity (sector
> selection) but they don't influence the memory addressing, nor could the MMU
> set them.
Yes, A64FX binaries are position independent. Arm64 supports
a maximum of 52-bit of virtual or physical address. On A64FX,
the [56:57] bits of virtual addresses are used for some cache
affinity (sector selection) and set by user program instead of MMU.
Best regards,
Tan Shaopeng
Hi Reinette,
I$B!G(Bm sorry for the late reply.
I think I could not explain A64FX$B!G(Bs sector cache function well in
my first mail. While answering the question, I will also explain
this function in more detail. Though maybe you have already learned
more about this function by reading specification and manual,
in order to better understand this function, some contents may have
duplicate explanations.
> >> The overview in section 12 was informative but very high level.
> >
> > I'm considering how to answer your questions from your email which I
> > received before, when I check the email again, I am sorry that the
> > information I provided before are insufficient.
> >
> > To understand the sector cache function of A64FX, could you please see
> > A64FX_Microarchitecture_Manual - section 12. Sector Cache
> >
> https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Microarchitectu
> > re_Manual_en_1.4.pdf
> > and,
> > A64FX_Specification_HPC_Extension ? section 1.2. Sector Cache
> >
> https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Specification_H
> > PC_Extension_v1_EN.pdf
>
> Thank you for the direct links - I missed that there are two documents available.
>
> After reading the spec portion it does seem to me even more as though
> "sectors" could be considered the same as the resctrl "classes of service". The
> Fujitsu hardware supports four sectors that can be configured with different
> number of ways using the registers you mention above. In resctrl this could be
> considered as hardware that supports four classes of service and each class of
> service can be allocated a different number of ways.
Fujitsu hardware supports four sectors that can be configured with
different number of ways by using "IMP_SCCR" registers, and when this
function is added into resctrl, the maximum ways of each sector are
indicated by bitmap.
However, A64FX's L2 cache setting registers are shared among PEs
(Processor Element) in NUMA. If two PEs in the same NUMA are assigned
to different resource groups, changing one PE's L2 setting on one
resource group, the other PE's L2 setting on other resource groups
will be influenced. So, adding this function into resctrl, we will
assign NUMA to the resource group. (On F64FX, each NUMA has 12 PEs,
and each PE has L1 cache setting registers, but these registers are
not shared.) There are 4 NUMAs on A64FX, 4 NUMAs could be considered
as hardware that supports four classes of service at most, and each
class of service has 4 sectors (4 L1 sectors& 4 L2 sectors),
and each sector can be allocated a different number of ways.
And, when a running task on resource group, the [56:57] bits of
virtual address are used for sector selection (cache affinity).
> The other part is how hardware knows which sector is being used at any
> moment in time. In resctrl that is programmed by writing the active class of
> service into needed register at the time the application is context switched
> (resctrl_sched_in()). This seems different here since as you describe the
> sector is chosen by bits in the address. Even so, which bits to set in the
> address needs to be programmed also and I also understand that there is a
> "default" sector that can be programmed via register. Could these be equivalent
> to what is done currently in resctrl?
Adding this function into resctrl, there is no need to write active
class of service into needed register. When running a task, the sector
id is decided by [56:57] bits of virtual address, and these bits are
programed by users. When creating a resource group, the maximum number
of ways of each sector are set by "IMP_SCCR" setting registers.
As long as the task is running in a certain resource group, the sector
and the maximum number of ways of sectors are used will not be changed.
Therefore, we need not consider context switches on A64FX.
> (Could you please also consider my original questions?)
I will reply to the original questions mail.
Best regards,
Tan Shaopeng
Hi Reinette,
> On 4/21/2021 1:37 AM, [email protected] wrote:
> > Hi,
> >
> > Ping... any comments&advice about add an A64FX cache control function
> into resctrl?
>
> My apologies for the delay.
>
> >
> > Best regards
> > Tan Shaopeng
> >
> >> Hello
> >>
> >>
> >> I'm Tan Shaopeng from Fujitsu Limited.
> >>
> >> I$B!G(Bm trying to implement Fujitsu A64FX$B!G(Bs cache related features.
> >> It is a cache partitioning function we called sector cache function
> >> that using the value of the tag that is upper 8 bits of the 64bit
> >> address and the value of the sector cache register to control virtual cache
> capacity of the L1D&L2 cache.
> >>
> >> A few days ago, when I sent a driver that realizes this function to
> >> ARM64 kernel community, Will Deacon and Arnd Bergmann suggested an
> >> idea to add the sector cache function of A64FX into resctrl.
> >>
> https://lore.kernel.org/linux-arm-kernel/CAK8P3a2pFcNTw9NpRtQfYr7A5Oc
> >> [email protected]/
> >>
> >> Based on my study, I think the sector cache function of A64FX can be
> >> added into the allocation features of resctrl after James' resctrl rework has
> finished.
> >> But, in order to implement this function, more interfaces for resctrl are
> need.
> >> The details are as follow, and could you give me some advice?
> >>
> >> [Sector cache function]
> >> The sector cache function split cache into multiple sectors and
> >> control them separately. It is implemented on the L1D cache and
> >> L2 cache in the A64FX processor and can be controlled individually
> >> for L1D cache and L2 cache. A64FX has no L3 cache. Each L1D cache and
> >> L2 cache has 4 sectors. Which L1D sector is used is specified by the
> >> value of [57:56] bits of address, how many ways of sector are
> >> specified by the value of register (IMP_SCCR_L1_EL0).
> >> Which L2 sector is used is specified by the value of [56] bits of
> >> address, and how many ways of sector are specified by value of
> >> register (IMP_SCCR_ASSIGN_EL1, IMP_SCCR_SET0_L2_EL1,
> >> IMP_SCCR_SET1_L2_EL1).
> >>
> >> For more details of sector cache function, see A64FX HPC extension
> >> specification (1.2. Sector cache) in https://github.com/fujitsu/A64FX
>
> The overview in section 12 was informative but very high level.
> I was not able to find any instance of "IMP_SCCR" in this document to explore
> how this cache allocation works.
Maybe you have already known, the sector cache works as follows.
- Set the maximum number of ways for sector (id = 0/1/2/3) of L1&L2
by setting the "IMP_SCCR" register.
- When running a task, the sector id is specified by [56:57] bits of
the virtual address. If the sector id is not specified, the sector
id specified in IMP_SCCR_ASSIGN_EL1.default_sector will be used.
> Are these cache sectors exposed to the OS in any way? For example, when the
> OS discovers the cache, does it learn about these sectors and expose the
> details to user space (/sys/devices/system/cpuX/cache)?
These cache sectors are not exposed to the OS in any way.
> The overview of Sector Cache in that document provides details of how the size
> of the sector itself is dynamically adjusted to usage. That description is quite
> cryptic but it seems like a sector, since the number of ways associated with it
> can dynamically change, is more equivalent to a class of service or resource
> group in the resctrl environment.
I explained the difference between "sector" and "class of service"
in another email.
> I really may be interpreting things wrong here, could you perhaps point me to
> where I can obtain more details?
I'm sorry, there is no documentation other than the manual and
specifications. More details about how sector cache function works
as follows.
(1) By setting the access control register IMP_SCCR_CTRL_EL1, cache
capacity setting registers (IMP_SCCR_CTRL_EL1, IMP_SCCR_ASSIGN_EL1,
IMP_SCCR_L1_EL0, IMP_SCCR_SET0_L2_EL1, IMP_SCCR_SET1_L2_EL1,
IMP_SCCR_VSCCR_L2_EL0) can be set from user space or kernel space.
(2) Set L1 sector cache capacity register from kernel space.
By setting the register IMP_SCCR_L1_EL0, set the maximum number
of ways of sector (id = 0/1/2/3) of L1.
(3) Set L2 sector cache capacity register.
(one of cases) By setting IMP_SCCR_ASSIGN_EL1.assign = 0 from
kernel space, IMP_SCCR_VSCCR_L2_EL0 becomes alias of
IMP_SCCR_SET0_L2_EL1. By setting IMP_SCCR_VSCCR_L2_EL0 from user
space, set the maximum number of ways of sector (id = 0/1) of L2.
(4) When running a task, the sector ID of L1 is decided by [56:57]
bits of virtual address, and the sector ID of L2 is decided by
[56] bit of the address. These bits are programed by the users.
> >> [Difference between resctrl(CAT) and this sector cache function]
> >> L2/L3 CAT (Cache Allocation Technology) enables the user to specify
> >> some physical partition of cache space that an application can fill.
> >> A64FX's L1D/L2 cache has 4 sectors and 16ways. This sector function
> >> enables a user to specify number of ways each sector uses.
> >> Therefore, for CAT it is enough to specify a cache portion for each
> >> cache_id (socket). On the other hand, sector cache needs to specify
> >> cache portion of each sector for each cache_id, and following
> >> extension to resctrl interface is needed to support sector cache.
> >>
> >> [Idear for A64FX sector cache function control interface (schemata
> >> file details)]
> >>
> L1:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cw
> >> bm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
> >>
> L2:<cache_id0>=>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=
> >> <cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
> >>
> >> $B!&(BL1: Add a new interface to control the L1D cache.
> >> $B!&(B<cwbm>,<cwbm>,<cwbm>,<cwbm>$B!'(BSpecify the number of ways for
> each
> >> sector.
> >> $B!&(Bcwbm$B!'(BSpecify the number of ways in each sector as a bitmap
> (percentage),
> >> but the bitmap does not indicate the location of the cache.
> >> * In the sector cache function, L2 sector cache way setting register is
> >> shared among PEs (Processor Element) in shared domain. If two PEs
> >> which share L2 cache belongs to different resource groups, one
> resource
> >> group's L2 setting will affect to other resource group's L2 setting.
>
> In resctrl a "resource group" can be viewed as a class of service.
Thanks for your explanation. Adding sector cache function into resctrl,
I will use this mechanism of resctrl as it is.
> >> * Since A64FX does not support MPAM, it is not necessary to consider
> >> how to switch between MPAM and sector cache function now.
> >>
> >> Some questions:
> >> 1.I'm still studying about RDT, could you tell me whether RDT has
> >> the similar mechanism with sector cache function?
>
> This is not clear to me yet. One thing to keep in mind is that a bit in the capacity
> bitmask could correspond to some number of ways in a cache, but it does not
> have to. It is essentially a hint to hardware on how much cache space needs to
> be allocated while also indicating overlap and isolation from other allocations.
>
> resctrl already supports the bitmask being interpreted differently between
> architectures and with the MPAM support there will be even more support for
> different interpretations.
when adding sector cache function into resctrl,
the bitmap will only show the maximum number of ways of sector
and does not indicate cache position like in RDT.
Sector is a group of cache ways, and one cache line cannot be assigned
to different sectors at the same time. Different sectors have different
cache space. When different tasks use different sectors,
the cache space used can be isolated.
> >> 2.In RDT, L3 cache is shared among cores in socket. If two cores which
> >> share L3 cache belongs to different resource groups, one resource
> >> group's L3 setting will affect to other resource group's L3 setting?
>
> This question is not entirely clear to me. Are you referring to the hardware layout
> or configuration changes via the resctrl "cpus" file?
>
> Each resource group is a class of service (CLOS) that is supported by all cache
> instances. By default each resource group would thus contain all cache
> instances on the system (even if some cache instances do not support the
> same number of CLOS resctrl would only support the CLOS supported by all
> resources).
[Idea for A64FX sector cache function control]
An example of using the sector function when working on resctrl
as follows.
# mount -t resctrl resctrl /sys/fs/resctrl
# cd /sys/fs/resctrl
# mkdir p0
# echo XXXX > /sys/fs/resctrl/p0/cpus *1
# echo $B!H(BL1:0=000F,000F,000F,000F;1=000F,000F,000F,000F$B!I(B > /sys/fs/resctrl/p0/schemata*2
# echo $B!H(BL2:0=000F,000F,0,0;1=0,0,000F,000F$B!I(B > /sys/fs/resctrl/p0/schemata$B"((B2
# echo PID > sys/fs/resctrl/p0/tasks
*1
Since the A64FX L2 settings are shared by NUMA, all PEs (cores)
on the same NUMA should be specified at the same time.
In other words, we want to specify NUMAs instead of PEs to the
resource group. Maybe it is better to change the interface to
numas(/sys/fs/resctrl/p0/numas). Could you give me some advice?
*2
L1:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
L2:<cache_id0>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;<cache_id1>=<cwbm>,<cwbm>,<cwbm>,<cwbm>;$B!D(B
$B!&(B L1:
Add a new interface to control the L1D cache.
$B!&(B <cwbm>,<cwbm>,<cwbm>,<cwbm>:
Specify the number of ways for each sector.
Each L1/L2 cache has 4 sectors.
And it is needed to specify the number of ways for each sector.
$B!&(B cwbm$B!'(B
Specify the number of ways in each sector as a bitmap (percentage),
but the bitmap does not indicate the position of the cache.
The range is from 0 way to 16 ways.
When creating a resource group, the number of ways for 4 sectors of
L1&L2 is set. When running a task, the [56:57] bits of virtual
address are used for sector selection. Even different tasks running
in the same resource group can use different sector caches. Therefore,
when running a task that handles a large amount of infrequently used
data and a task that handles a large amount of frequently used data
at the same time, cache size limitation and cache space isolation can
be performed, and cache thrashing also can be reduced.
Is this approach acceptable? Could you give me some advice?
Best regards,
Tan Shaopeng
Hi Tan Shaopeng,
On 5/17/2021 1:31 AM, [email protected] wrote:
> Hi Reinette,
>
> I$B!G(Bm sorry for the late reply.
> I think I could not explain A64FX$B!G(Bs sector cache function well in
> my first mail. While answering the question, I will also explain
> this function in more detail. Though maybe you have already learned
> more about this function by reading specification and manual,
> in order to better understand this function, some contents may have
> duplicate explanations.
>
>>>> The overview in section 12 was informative but very high level.
>>>
>>> I'm considering how to answer your questions from your email which I
>>> received before, when I check the email again, I am sorry that the
>>> information I provided before are insufficient.
>>>
>>> To understand the sector cache function of A64FX, could you please see
>>> A64FX_Microarchitecture_Manual - section 12. Sector Cache
>>>
>> https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Microarchitectu
>>> re_Manual_en_1.4.pdf
>>> and,
>>> A64FX_Specification_HPC_Extension ? section 1.2. Sector Cache
>>>
>> https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Specification_H
>>> PC_Extension_v1_EN.pdf
>>
>> Thank you for the direct links - I missed that there are two documents available.
>>
>> After reading the spec portion it does seem to me even more as though
>> "sectors" could be considered the same as the resctrl "classes of service". The
>> Fujitsu hardware supports four sectors that can be configured with different
>> number of ways using the registers you mention above. In resctrl this could be
>> considered as hardware that supports four classes of service and each class of
>> service can be allocated a different number of ways.
>
> Fujitsu hardware supports four sectors that can be configured with
> different number of ways by using "IMP_SCCR" registers, and when this
> function is added into resctrl, the maximum ways of each sector are
> indicated by bitmap.
>
> However, A64FX's L2 cache setting registers are shared among PEs
> (Processor Element) in NUMA. If two PEs in the same NUMA are assigned
> to different resource groups, changing one PE's L2 setting on one
> resource group, the other PE's L2 setting on other resource groups
> will be influenced. So, adding this function into resctrl, we will
> assign NUMA to the resource group. (On F64FX, each NUMA has 12 PEs,
> and each PE has L1 cache setting registers, but these registers are
> not shared.) There are 4 NUMAs on A64FX, 4 NUMAs could be considered
> as hardware that supports four classes of service at most, and each
> class of service has 4 sectors (4 L1 sectors& 4 L2 sectors),
> and each sector can be allocated a different number of ways.
> And, when a running task on resource group, the [56:57] bits of
> virtual address are used for sector selection (cache affinity).
It is not clear to me why NUMA needs to be involved.
Processors sharing a cache, either L2 or L3 cache, is familiar and well
supported by resctrl.
My understanding of the sector cache feature is that each cache can be
split into multiple (4) sectors. It thus seems to me something specific
to the cache itself.
Let me try and give an example of my understanding based on the cache
architecture described in the A64FX Microarchitecture Manual.
I see in Figure 9-2 that each processor has an L1D as well as L1I Cache,
and twelve processors share an L2 cache. The L1D cache has 4 ways (0xF
bitmask) and L2 cache has 16 (0xFFFF bitmask) ways. From what I
understand the sector cache function is supported on L1D and L2.
First, the goal would be to discover all the caches on the system -
since it is the sectors need to be programmed on each cache. On the
system with 48 cores there would thus be 48 L1D caches, and 4 L2 caches.
Let's start by assigning the caches IDs: the L1D caches are numbered
from 0 to 47 and the L2 caches numbered from 0 to 3.
My understanding is that the goal is to program these sectors using
resctrl. Each cache instance can have maximum four sectors, they cannot
overlap. (I do not know if each sector has to have some portion of cache
associated with it or if a sector is allowed to be "empty").
So, what is needed is, for example, to have a way to say: "sector 0 on
cache L1D with id X is assigned Y ways", "sector 1 on cache L2 with id Z
is assigned XX ways". Is this correct?
If my understanding is correct then you can do this with resctrl as
follows (I am making many assumptions on behavior here, especially
regarding how many ways a sector is required to have, but I hope this
could be a baseline to evaluate and correct my understanding and build
on how this could be supported):
On boot all cache ways on all cache instances belong to sector 0:
# cd /sys/fs/resctrl/
# cat schemata
L1D:0=0xf;1=0xf;2=0xf;.....;47=0xf
L2:0=0xffff;1=0xffff;2=0xffff;3=0xffff
Create sector2 and assign half of all cache ways to it:
(In support of this it would be required that resctrl resource groups
are exclusive. Exclusive resource groups are already supported but not
the default as it needed here.)
First, to provide cache ways to sector 1, the cache ways needs to be
removed from sector 0:
(I am not sure if specific ways can be assigned to a sector or just a
number of ways, both could be supported)
# echo 'L1D:0=0x3;1=0x3;...;47=0x3' > /sys/fs/resctrl/schemata
# echo 'L2:0=0xff;1=0xff;2=0xff;3=0xff'> /sys/fs/resctrl/schemata
Now create sector2 (alternatively all sectors could exist on boot for
this system):
# mkdir /sys/fs/resctrl/sector2
# echo 'L1D:0=0x3;1=0x3;...;47=0x3' > /sys/fs/resctrl/sector2/schemata
# echo 'L2:0=0xff;1=0xff;2=0xff;3=0xff'> /sys/fs/resctrl/sector2/schemata
At this point there are two sectors configured. Configuration of sector0
can be found in /sys/fs/resctrl/schemata and configuration of sector1 in
/sys/fs/resctrl/sector1/schemata
>> The other part is how hardware knows which sector is being used at any
>> moment in time. In resctrl that is programmed by writing the active class of
>> service into needed register at the time the application is context switched
>> (resctrl_sched_in()). This seems different here since as you describe the
>> sector is chosen by bits in the address. Even so, which bits to set in the
>> address needs to be programmed also and I also understand that there is a
>> "default" sector that can be programmed via register. Could these be equivalent
>> to what is done currently in resctrl?
>
> Adding this function into resctrl, there is no need to write active
> class of service into needed register. When running a task, the sector
> id is decided by [56:57] bits of virtual address, and these bits are
> programed by users. When creating a resource group, the maximum number
> of ways of each sector are set by "IMP_SCCR" setting registers.
> As long as the task is running in a certain resource group, the sector
> and the maximum number of ways of sectors are used will not be changed.
> Therefore, we need not consider context switches on A64FX.
>
The current interface would associate a "tasks" file with each sector to
indicate which tasks run with the particular sector id. I thought there
was a way to program the default sector id in a register, which is
something that could be done when a task is context switched in.
Otherwise there would need to be some re-architecting to remove the
"tasks" association. This would be a significant change.
Reinette
Hi Reinette,
Sorry, I have not explained A64FX's sector cache function well yet.
I think I need explain this function from different perspective.
> On 5/17/2021 1:31 AM, [email protected] wrote:
> > Hi Reinette,
> >
> > I$B!G(Bm sorry for the late reply.
> > I think I could not explain A64FX$B!G(Bs sector cache function well in my
> > first mail. While answering the question, I will also explain this
> > function in more detail. Though maybe you have already learned more
> > about this function by reading specification and manual, in order to
> > better understand this function, some contents may have duplicate
> > explanations.
> >
> >>>> The overview in section 12 was informative but very high level.
> >>>
> >>> I'm considering how to answer your questions from your email which I
> >>> received before, when I check the email again, I am sorry that the
> >>> information I provided before are insufficient.
> >>>
> >>> To understand the sector cache function of A64FX, could you please
> >>> see A64FX_Microarchitecture_Manual - section 12. Sector Cache
> >>>
> >>
> https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Microarchitect
> >> u
> >>> re_Manual_en_1.4.pdf
> >>> and,
> >>> A64FX_Specification_HPC_Extension ? section 1.2. Sector Cache
> >>>
> >>
> https://github.com/fujitsu/A64FX/blob/master/doc/A64FX_Specification_
> >> H
> >>> PC_Extension_v1_EN.pdf
> >>
> >> Thank you for the direct links - I missed that there are two documents
> available.
> >>
> >> After reading the spec portion it does seem to me even more as though
> >> "sectors" could be considered the same as the resctrl "classes of
> >> service". The Fujitsu hardware supports four sectors that can be
> >> configured with different number of ways using the registers you
> >> mention above. In resctrl this could be considered as hardware that
> >> supports four classes of service and each class of service can be allocated
> a different number of ways.
> >
> > Fujitsu hardware supports four sectors that can be configured with
> > different number of ways by using "IMP_SCCR" registers, and when this
> > function is added into resctrl, the maximum ways of each sector are
> > indicated by bitmap.
> >
> > However, A64FX's L2 cache setting registers are shared among PEs
> > (Processor Element) in NUMA. If two PEs in the same NUMA are assigned
> > to different resource groups, changing one PE's L2 setting on one
> > resource group, the other PE's L2 setting on other resource groups
> > will be influenced. So, adding this function into resctrl, we will
> > assign NUMA to the resource group. (On F64FX, each NUMA has 12 PEs,
> > and each PE has L1 cache setting registers, but these registers are
> > not shared.) There are 4 NUMAs on A64FX, 4 NUMAs could be considered
> > as hardware that supports four classes of service at most, and each
> > class of service has 4 sectors (4 L1 sectors& 4 L2 sectors), and each
> > sector can be allocated a different number of ways.
> > And, when a running task on resource group, the [56:57] bits of
> > virtual address are used for sector selection (cache affinity).
>
> It is not clear to me why NUMA needs to be involved.
>
> Processors sharing a cache, either L2 or L3 cache, is familiar and well
> supported by resctrl.
>
> My understanding of the sector cache feature is that each cache can be split
> into multiple (4) sectors. It thus seems to me something specific to the cache
> itself.
>
> Let me try and give an example of my understanding based on the cache
> architecture described in the A64FX Microarchitecture Manual.
>
> I see in Figure 9-2 that each processor has an L1D as well as L1I Cache, and
> twelve processors share an L2 cache. The L1D cache has 4 ways (0xF
> bitmask) and L2 cache has 16 (0xFFFF bitmask) ways. From what I understand
> the sector cache function is supported on L1D and L2.
>
> First, the goal would be to discover all the caches on the system - since it is the
> sectors need to be programmed on each cache. On the system with 48 cores
> there would thus be 48 L1D caches, and 4 L2 caches.
>
> Let's start by assigning the caches IDs: the L1D caches are numbered from 0 to
> 47 and the L2 caches numbered from 0 to 3.
>
> My understanding is that the goal is to program these sectors using resctrl.
> Each cache instance can have maximum four sectors, they cannot overlap. (I do
> not know if each sector has to have some portion of cache associated with it or
> if a sector is allowed to be "empty").
>
> So, what is needed is, for example, to have a way to say: "sector 0 on cache L1D
> with id X is assigned Y ways", "sector 1 on cache L2 with id Z is assigned XX
> ways". Is this correct?
>
> If my understanding is correct then you can do this with resctrl as follows (I am
> making many assumptions on behavior here, especially regarding how many
> ways a sector is required to have, but I hope this could be a baseline to evaluate
> and correct my understanding and build on how this could be supported):
>
> On boot all cache ways on all cache instances belong to sector 0:
>
> # cd /sys/fs/resctrl/
> # cat schemata
> L1D:0=0xf;1=0xf;2=0xf;.....;47=0xf
> L2:0=0xffff;1=0xffff;2=0xffff;3=0xffff
>
> Create sector2 and assign half of all cache ways to it:
> (In support of this it would be required that resctrl resource groups are
> exclusive. Exclusive resource groups are already supported but not the default
> as it needed here.)
>
> First, to provide cache ways to sector 1, the cache ways needs to be removed
> from sector 0:
> (I am not sure if specific ways can be assigned to a sector or just a number of
> ways, both could be supported) # echo 'L1D:0=0x3;1=0x3;...;47=0x3' >
> /sys/fs/resctrl/schemata # echo 'L2:0=0xff;1=0xff;2=0xff;3=0xff'>
> /sys/fs/resctrl/schemata
>
> Now create sector2 (alternatively all sectors could exist on boot for this
> system):
> # mkdir /sys/fs/resctrl/sector2
> # echo 'L1D:0=0x3;1=0x3;...;47=0x3' > /sys/fs/resctrl/sector2/schemata #
> echo 'L2:0=0xff;1=0xff;2=0xff;3=0xff'> /sys/fs/resctrl/sector2/schemata
>
> At this point there are two sectors configured. Configuration of sector0 can be
> found in /sys/fs/resctrl/schemata and configuration of sector1 in
> /sys/fs/resctrl/sector1/schemata
>
> >> The other part is how hardware knows which sector is being used at
> >> any moment in time. In resctrl that is programmed by writing the
> >> active class of service into needed register at the time the
> >> application is context switched (resctrl_sched_in()). This seems
> >> different here since as you describe the sector is chosen by bits in
> >> the address. Even so, which bits to set in the address needs to be
> >> programmed also and I also understand that there is a "default"
> >> sector that can be programmed via register. Could these be equivalent to
> what is done currently in resctrl?
> >
> > Adding this function into resctrl, there is no need to write active
> > class of service into needed register. When running a task, the sector
> > id is decided by [56:57] bits of virtual address, and these bits are
> > programed by users. When creating a resource group, the maximum number
> > of ways of each sector are set by "IMP_SCCR" setting registers.
> > As long as the task is running in a certain resource group, the sector
> > and the maximum number of ways of sectors are used will not be changed.
> > Therefore, we need not consider context switches on A64FX.
> >
>
> The current interface would associate a "tasks" file with each sector to indicate
> which tasks run with the particular sector id. I thought there was a way to
> program the default sector id in a register, which is something that could be
> done when a task is context switched in.
> Otherwise there would need to be some re-architecting to remove the "tasks"
> association. This would be a significant change.
--------
A64FX NUMA-PE-Cache Architecture:
NUMA0:
PE0:
L1sector0,L1sector1,L1sector2,L1sector3
PE1:
L1sector0,L1sector1,L1sector2,L1sector3
...
PE11:
L1sector0,L1sector1,L1sector2,L1sector3
L2sector0,1/L2sector2,3
NUMA1:
PE0:
L1sector0,L1sector1,L1sector2,L1sector3
...
PE11:
L1sector0,L1sector1,L1sector2,L1sector3
L2sector0,1/L2sector2,3
NUMA2:
...
NUMA3:
...
--------
In A64FX processor, one L1 sector cache capacity setting register is
only for one PE and not shared among PEs. L2 sector cache maximum
capacity setting registers are shared among PEs in same NUMA, and it is
to be noted that changing these registers in one PE influences other PE.
The number of ways for L2 Sector ID (0,1 or 2,3) can be set through
any PEs in same NUMA. The sector ID 0,1 and 2,3 are not available at
the same time in same NUMA.
I think, in your idea, a resource group will be created for each sector ID.
(> "sectors" could be considered the same as the resctrl "classes of service")
Then, an example of resource group is created as follows.
$B!&(B L1: NUMAX-PEY-L1sector0 (X = 0,1,2,3.Y = 0,1,2 ... 11),
$B!&(B L2: NUMAX-L2sector0 (X = 0,1,2,3)
In this example, sector with same ID(0) of all PEs is allocated to
resource group. The L1D caches are numbered from NUMA0_PE0-L1sector0(0)
to NUMA4_PE11-L1sector0(47) and the L2 caches numbered from
NUMA0-L2sector0(0) to NUM4-L2sector0(3).
(NUMA number X is from 0-4, PE number Y is from 0-11)
(1) The number of ways of NUMAX-PEY-L1sector0 can be set independently
for each PEs (0-47). When run a task on this resource group,
we cannot control on which PE the task is running on and how many
cache ways the task is using.
(2) Since L2 can only use 2 sectors at a time, when creating more than
2 resource groups, L2setctor0 will have to be allocated to a
different resource group. If the L2sector0 is shared by different
resource groups, the L2 sector settings on resource group will be
influenced by each other.
etc... there are various problems, and no merit to using resctrl.
In my idea, in order to allocate the L1 and L2 cache to a resource
group, allocate NUMA to the resource group.
An example of resource group is as follows.
$B!&(B NUMA0-PEY-L1sectorZ (Y = 0,1,2...11. Z = 0,1,2,3)
$B!&(B NUMA0-L2sectorZZ (ZZ = 0,1,2,3)
#cat /sys/fs/resctrl/p0/cpus
0-11 *1
#cat /sys/fs/resctrl/p0/schemata
L1:0=0xF,0x3,0x1,x0x0 *2
L2:0=0xFFF,0xF,0,0 *3
*1: PEs belong one NUMA. (Of course, multiple NUMAs can also be
specified in one resource group)
*2: The number of ways for L1sector0,1,2,3. On this resource group
the number of ways of all sector0 is the same(0xF). If 0 way is
specified for one sector, this sector cannot be used. If 4(0xF)
ways are specified for one sector, this sector can use cache fully.
If 4 ways are specified for each sector, there will be no
restriction for using cache.
*3: The number of ways for L2 sector 0,1. If L2sector0,1 is used,
the number of ways of L2sector2,3 must be set to 0.
All sectors with the same ID on the same resource group were set to
the same number of ways, and when running a task on A64FX, the sector
ID used by task is determined by [56:57] bits of virtual address.
By specifying the PID to /sys/fs/resctrl/tasks, the task will be bound
to the resource group, and then, the cache size used by task will not
be changed never.
Best regards,
Tan Shaopeng
Hi Tan Shaopeng,
On 5/25/2021 1:45 AM, [email protected] wrote:
> Hi Reinette,
>
> Sorry, I have not explained A64FX's sector cache function well yet.
> I think I need explain this function from different perspective.
You have explained the A64FX's sector cache function well. I have also
read both specs to understand it better. It appears to me that you are
not considering the resctrl architecture as part of your solution but
instead just forcing your architecture onto the resctrl filesystem. For
example, in resctrl the resource groups are not just a directory
structure but has significance in what is being represented within the
directory (a class of service). The files within a resource group's
directory build on that. From your side I have not seen any effort in
aligning the sector cache function with the resctrl architecture but
instead you are just changing resctrl interface to match the A64FX
architecture.
Could you please take a moment to understand what resctrl is and how it
could be mapped to A64FX in a coherent way?
>
>> On 5/17/2021 1:31 AM, [email protected] wrote:
> --------
> A64FX NUMA-PE-Cache Architecture:
> NUMA0:
> PE0:
> L1sector0,L1sector1,L1sector2,L1sector3
> PE1:
> L1sector0,L1sector1,L1sector2,L1sector3
> ...
> PE11:
> L1sector0,L1sector1,L1sector2,L1sector3
>
> L2sector0,1/L2sector2,3
> NUMA1:
> PE0:
> L1sector0,L1sector1,L1sector2,L1sector3
> ...
> PE11:
> L1sector0,L1sector1,L1sector2,L1sector3
>
> L2sector0,1/L2sector2,3
> NUMA2:
> ...
> NUMA3:
> ...
> --------
> In A64FX processor, one L1 sector cache capacity setting register is
> only for one PE and not shared among PEs. L2 sector cache maximum
> capacity setting registers are shared among PEs in same NUMA, and it is
> to be noted that changing these registers in one PE influences other PE.
Understood. cache affinity is familiar to resctrl. When a CPU becomes
online it is discovered which caches/resources it has affinity to.
Resources then have CPU mask associated with them to indicate on which
CPU a register could be changed to configure the resource/cache. See
domain_add_cpu() and struct rdt_domain.
> The number of ways for L2 Sector ID (0,1 or 2,3) can be set through
> any PEs in same NUMA. The sector ID 0,1 and 2,3 are not available at
> the same time in same NUMA.
>
>
> I think, in your idea, a resource group will be created for each sector ID.
> (> "sectors" could be considered the same as the resctrl "classes of service")
> Then, an example of resource group is created as follows.
> $B!&(B L1: NUMAX-PEY-L1sector0 (X = 0,1,2,3.Y = 0,1,2 ... 11),
> $B!&(B L2: NUMAX-L2sector0 (X = 0,1,2,3)
>
> In this example, sector with same ID(0) of all PEs is allocated to
> resource group. The L1D caches are numbered from NUMA0_PE0-L1sector0(0)
> to NUMA4_PE11-L1sector0(47) and the L2 caches numbered from
> NUMA0-L2sector0(0) to NUM4-L2sector0(3).
> (NUMA number X is from 0-4, PE number Y is from 0-11)
> (1) The number of ways of NUMAX-PEY-L1sector0 can be set independently
> for each PEs (0-47). When run a task on this resource group,
> we cannot control on which PE the task is running on and how many
> cache ways the task is using.
resctrl does not control the affinity on which PE/CPU a task is run.
resctrl is an interface with which to configure how resources are
allocated on the system. resctrl could thus provide interface with which
each sector of each cache instance is assigned a number of cache ways.
resctrl also provides an interface to assign a task with a class of
service (sector id?). Through this the task obtains access to all
resources that is allocated to the particular class of service (sector
id?). Depending on which CPU the task is running it may indeed
experience different performance if the sector id it is running with
does not have the same allocations on all cache instances. The affinity
of the task needs to be managed separately using for example taskset.
Please see Documentation/x86/resctrl.rst "Examples for RDT allocation usage"
> (2) Since L2 can only use 2 sectors at a time, when creating more than
> 2 resource groups, L2setctor0 will have to be allocated to a
> different resource group. If the L2sector0 is shared by different
> resource groups, the L2 sector settings on resource group will be
> influenced by each other.
> etc... there are various problems, and no merit to using resctrl.
>
>
> In my idea, in order to allocate the L1 and L2 cache to a resource
> group, allocate NUMA to the resource group.
> An example of resource group is as follows.
> $B!&(B NUMA0-PEY-L1sectorZ (Y = 0,1,2...11. Z = 0,1,2,3)
> $B!&(B NUMA0-L2sectorZZ (ZZ = 0,1,2,3)
>
> #cat /sys/fs/resctrl/p0/cpus
> 0-11 *1
> #cat /sys/fs/resctrl/p0/schemata
> L1:0=0xF,0x3,0x1,x0x0 *2
> L2:0=0xFFF,0xF,0,0 *3
>
> *1: PEs belong one NUMA. (Of course, multiple NUMAs can also be
> specified in one resource group)
> *2: The number of ways for L1sector0,1,2,3. On this resource group
> the number of ways of all sector0 is the same(0xF). If 0 way is
> specified for one sector, this sector cannot be used. If 4(0xF)
> ways are specified for one sector, this sector can use cache fully.
> If 4 ways are specified for each sector, there will be no
> restriction for using cache.
> *3: The number of ways for L2 sector 0,1. If L2sector0,1 is used,
> the number of ways of L2sector2,3 must be set to 0.
>
> All sectors with the same ID on the same resource group were set to
> the same number of ways, and when running a task on A64FX, the sector
> ID used by task is determined by [56:57] bits of virtual address.
> By specifying the PID to /sys/fs/resctrl/tasks, the task will be bound
> to the resource group, and then, the cache size used by task will not
> be changed never.
This completely ignores how this directory and files are currently used.
What is missing how this implementation maps to the current resctrl
architecture.
Reinette
Hi Reinette,
> On 5/25/2021 1:45 AM, [email protected] wrote:
> > Hi Reinette,
> >
> > Sorry, I have not explained A64FX's sector cache function well yet.
> > I think I need explain this function from different perspective.
>
> You have explained the A64FX's sector cache function well. I have also read
> both specs to understand it better. It appears to me that you are not considering
> the resctrl architecture as part of your solution but instead just forcing your
> architecture onto the resctrl filesystem. For example, in resctrl the resource
> groups are not just a directory structure but has significance in what is being
> represented within the directory (a class of service). The files within a resource
> group's directory build on that. From your side I have not seen any effort in
> aligning the sector cache function with the resctrl architecture but instead you
> are just changing resctrl interface to match the A64FX architecture.
>
> Could you please take a moment to understand what resctrl is and how it could
> be mapped to A64FX in a coherent way?
Thanks for your mail.
Sorry, I$B!G(Bm wrong in understanding how to use
/sys/fs/resctrl/p0/cpus and /sys/fs/resctrl/tasks.
I think I have not understood resctrl yet, and I will learn more about it.
If I have questions, please allow me to mail you.
> >> On 5/17/2021 1:31 AM, [email protected] wrote:
>
> > --------
> > A64FX NUMA-PE-Cache Architecture:
> > NUMA0:
> > PE0:
> > L1sector0,L1sector1,L1sector2,L1sector3
> > PE1:
> > L1sector0,L1sector1,L1sector2,L1sector3
> > ...
> > PE11:
> > L1sector0,L1sector1,L1sector2,L1sector3
> >
> > L2sector0,1/L2sector2,3
> > NUMA1:
> > PE0:
> > L1sector0,L1sector1,L1sector2,L1sector3
> > ...
> > PE11:
> > L1sector0,L1sector1,L1sector2,L1sector3
> >
> > L2sector0,1/L2sector2,3
> > NUMA2:
> > ...
> > NUMA3:
> > ...
> > --------
> > In A64FX processor, one L1 sector cache capacity setting register is
> > only for one PE and not shared among PEs. L2 sector cache maximum
> > capacity setting registers are shared among PEs in same NUMA, and it
> > is to be noted that changing these registers in one PE influences other PE.
>
> Understood. cache affinity is familiar to resctrl. When a CPU becomes online it
> is discovered which caches/resources it has affinity to.
> Resources then have CPU mask associated with them to indicate on which
> CPU a register could be changed to configure the resource/cache. See
> domain_add_cpu() and struct rdt_domain.
>
> > The number of ways for L2 Sector ID (0,1 or 2,3) can be set through
> > any PEs in same NUMA. The sector ID 0,1 and 2,3 are not available at
> > the same time in same NUMA.
> >
> >
> > I think, in your idea, a resource group will be created for each sector ID.
> > (> "sectors" could be considered the same as the resctrl "classes of
> > service") Then, an example of resource group is created as follows.
> > $B!&(B L1: NUMAX-PEY-L1sector0 (X = 0,1,2,3.Y = 0,1,2 ... 11),
> > $B!&(B L2: NUMAX-L2sector0 (X = 0,1,2,3)
> >
> > In this example, sector with same ID(0) of all PEs is allocated to
> > resource group. The L1D caches are numbered from
> > NUMA0_PE0-L1sector0(0) to NUMA4_PE11-L1sector0(47) and the L2
> caches
> > numbered from
> > NUMA0-L2sector0(0) to NUM4-L2sector0(3).
> > (NUMA number X is from 0-4, PE number Y is from 0-11)
> > (1) The number of ways of NUMAX-PEY-L1sector0 can be set independently
> > for each PEs (0-47). When run a task on this resource group,
> > we cannot control on which PE the task is running on and how many
> > cache ways the task is using.
>
> resctrl does not control the affinity on which PE/CPU a task is run.
> resctrl is an interface with which to configure how resources are allocated on
> the system. resctrl could thus provide interface with which each sector of each
> cache instance is assigned a number of cache ways.
> resctrl also provides an interface to assign a task with a class of service (sector
> id?). Through this the task obtains access to all resources that is allocated to
> the particular class of service (sector id?). Depending on which CPU the task is
> running it may indeed experience different performance if the sector id it is
> running with does not have the same allocations on all cache instances. The
> affinity of the task needs to be managed separately using for example taskset.
> Please see Documentation/x86/resctrl.rst "Examples for RDT allocation usage"
>
> > (2) Since L2 can only use 2 sectors at a time, when creating more than
> > 2 resource groups, L2setctor0 will have to be allocated to a
> > different resource group. If the L2sector0 is shared by different
> > resource groups, the L2 sector settings on resource group will be
> > influenced by each other.
> > etc... there are various problems, and no merit to using resctrl.
> >
> >
> > In my idea, in order to allocate the L1 and L2 cache to a resource
> > group, allocate NUMA to the resource group.
> > An example of resource group is as follows.
> > $B!&(B NUMA0-PEY-L1sectorZ (Y = 0,1,2...11. Z = 0,1,2,3)
> > $B!&(B NUMA0-L2sectorZZ (ZZ = 0,1,2,3)
> >
> > #cat /sys/fs/resctrl/p0/cpus
> > 0-11 *1
> > #cat /sys/fs/resctrl/p0/schemata
> > L1:0=0xF,0x3,0x1,x0x0 *2
> > L2:0=0xFFF,0xF,0,0 *3
> >
> > *1: PEs belong one NUMA. (Of course, multiple NUMAs can also be
> > specified in one resource group)
> > *2: The number of ways for L1sector0,1,2,3. On this resource group
> > the number of ways of all sector0 is the same(0xF). If 0 way is
> > specified for one sector, this sector cannot be used. If 4(0xF)
> > ways are specified for one sector, this sector can use cache fully.
> > If 4 ways are specified for each sector, there will be no
> > restriction for using cache.
> > *3: The number of ways for L2 sector 0,1. If L2sector0,1 is used,
> > the number of ways of L2sector2,3 must be set to 0.
> >
> > All sectors with the same ID on the same resource group were set to
> > the same number of ways, and when running a task on A64FX, the sector
> > ID used by task is determined by [56:57] bits of virtual address.
> > By specifying the PID to /sys/fs/resctrl/tasks, the task will be bound
> > to the resource group, and then, the cache size used by task will not
> > be changed never.
>
> This completely ignores how this directory and files are currently used.
> What is missing how this implementation maps to the current resctrl
> architecture.
Best regards,
Tan Shaopeng
Hi Reinette,
> > Sorry, I have not explained A64FX's sector cache function well yet.
> > I think I need explain this function from different perspective.
>
> You have explained the A64FX's sector cache function well. I have also read
> both specs to understand it better. It appears to me that you are not considering
> the resctrl architecture as part of your solution but instead just forcing your
> architecture onto the resctrl filesystem. For example, in resctrl the resource
> groups are not just a directory structure but has significance in what is being
> represented within the directory (a class of service). The files within a resource
> group's directory build on that. From your side I have not seen any effort in
> aligning the sector cache function with the resctrl architecture but instead you
> are just changing resctrl interface to match the A64FX architecture.
>
> Could you please take a moment to understand what resctrl is and how it could
> be mapped to A64FX in a coherent way?
Previously, my idea is based on how to make instructions use different
sectors in one task. After I studied resctrl, to utilize resctrl
architecture on A64FX, I think it$B!G(Bs better to assign one sector to
one task. Thanks for your idea that "sectors" could be considered the
same as the resctrl "classes of service".
Based on your idea, I am considering the implementation details.
In this email, I will explain the outline of new proposal, and then
please allow me to confirm a few technologies about resctrl.
The outline of my proposal is as follows.
- Add a sector function equivalent to Intel's CAT function into resctrl.
(divide shared L2 cache into multiple partitions for multiple cores use)
- Allocate one sector to one resource group (one CLOSID). Since one
core can only be assigned to one resource group, on A64FX each core
only uses one sector at a time.
- Disable A64FX's HPC tag address override function. We only set each
core's default sector value according to closid(default sector ID=CLOSID).
- No L1 cache control since L1 cache is not shared for cores. It is not
necessary to add L1 cache interface for schemata file.
- No need to update schemata interface. Resctrl's L2 cache interface
(L2: <cache_id0> = <cbm>; <cache_id1> = <cbm>; ...)
will be used as it is. However, on A64FX, <cbm> does not indicate
the position of cache partition, only indicate the number of
cache ways (size).
This is the smallest start of incorporating sector cache function into
resctrl. I will consider if we could add more sector cache features
into resctrl (e.g. selecting different sectors from one task) after
finishing this.
(some questions are below)
> >
> >> On 5/17/2021 1:31 AM, [email protected] wrote:
>
> > --------
> > A64FX NUMA-PE-Cache Architecture:
> > NUMA0:
> > PE0:
> > L1sector0,L1sector1,L1sector2,L1sector3
> > PE1:
> > L1sector0,L1sector1,L1sector2,L1sector3
> > ...
> > PE11:
> > L1sector0,L1sector1,L1sector2,L1sector3
> >
> > L2sector0,1/L2sector2,3
> > NUMA1:
> > PE0:
> > L1sector0,L1sector1,L1sector2,L1sector3
> > ...
> > PE11:
> > L1sector0,L1sector1,L1sector2,L1sector3
> >
> > L2sector0,1/L2sector2,3
> > NUMA2:
> > ...
> > NUMA3:
> > ...
> > --------
> > In A64FX processor, one L1 sector cache capacity setting register is
> > only for one PE and not shared among PEs. L2 sector cache maximum
> > capacity setting registers are shared among PEs in same NUMA, and it
> > is to be noted that changing these registers in one PE influences other PE.
>
> Understood. cache affinity is familiar to resctrl. When a CPU becomes online it
> is discovered which caches/resources it has affinity to.
> Resources then have CPU mask associated with them to indicate on which
> CPU a register could be changed to configure the resource/cache. See
> domain_add_cpu() and struct rdt_domain.
Is the following understanding correct?
Struct rdt_domain is a group of online CPUs that share a same cache
instance. When a CPU is online(resctrl initialization),
the domain_add_cpu() function add the online cpu to corresponding
rdt_domain (in rdt_resource:domains list). For example, if there are
4 L2 cache instances, then there will be 4 rdt_domain in the list and
each CPU is assigned to corresponding rdt_domain.
The set values of cache/memory are stored in the *ctrl_val array
(indexed by CLOSID) of struct rdt_domain. For example, in CAT function,
the CBM value of CLOSID=x is stored in ctrl_val [x].
When we create a resource group and write set values of cache into
the schemata file, the update_domains() function updates the CBM value
to ctrl_val [CLOSID = resource group ID] in rdt_domain and updates the
CBM value to CBM register(MSR_IA32_Lx_CBM_BASE).
> > The number of ways for L2 Sector ID (0,1 or 2,3) can be set through
> > any PEs in same NUMA. The sector ID 0,1 and 2,3 are not available at
> > the same time in same NUMA.
> >
> >
> > I think, in your idea, a resource group will be created for each sector ID.
> > (> "sectors" could be considered the same as the resctrl "classes of
> > service") Then, an example of resource group is created as follows.
> > $B!&(B L1: NUMAX-PEY-L1sector0 (X = 0,1,2,3.Y = 0,1,2 ... 11),
> > $B!&(B L2: NUMAX-L2sector0 (X = 0,1,2,3)
> >
> > In this example, sector with same ID(0) of all PEs is allocated to
> > resource group. The L1D caches are numbered from
> > NUMA0_PE0-L1sector0(0) to NUMA4_PE11-L1sector0(47) and the L2
> caches
> > numbered from
> > NUMA0-L2sector0(0) to NUM4-L2sector0(3).
> > (NUMA number X is from 0-4, PE number Y is from 0-11)
> > (1) The number of ways of NUMAX-PEY-L1sector0 can be set independently
> > for each PEs (0-47). When run a task on this resource group,
> > we cannot control on which PE the task is running on and how many
> > cache ways the task is using.
>
> resctrl does not control the affinity on which PE/CPU a task is run.
> resctrl is an interface with which to configure how resources are allocated on
> the system. resctrl could thus provide interface with which each sector of each
> cache instance is assigned a number of cache ways.
> resctrl also provides an interface to assign a task with a class of service (sector
> id?). Through this the task obtains access to all resources that is allocated to
> the particular class of service (sector id?). Depending on which CPU the task is
> running it may indeed experience different performance if the sector id it is
> running with does not have the same allocations on all cache instances. The
> affinity of the task needs to be managed separately using for example taskset.
> Please see Documentation/x86/resctrl.rst "Examples for RDT allocation usage"
In resctrl_sched_in(), there are comments as follow:
/*
* If this task has a closid/rmid assigned, use it.
* Else use the closid/rmid assigned to this cpu.
*/
I thought when we write PID to tasks file, this task (PID) will only
run on the CPUs which are specified in cpus file in the same resource
group. So, the task_struct's closid and cpu's closid is the same.
When task's closid is different from cpu's closid?
Best regards,
Tan Shaopeng
Hi Reinette,
> > > Sorry, I have not explained A64FX's sector cache function well yet.
> > > I think I need explain this function from different perspective.
> >
> > You have explained the A64FX's sector cache function well. I have also
> > read both specs to understand it better. It appears to me that you are
> > not considering the resctrl architecture as part of your solution but
> > instead just forcing your architecture onto the resctrl filesystem.
> > For example, in resctrl the resource groups are not just a directory
> > structure but has significance in what is being represented within the
> > directory (a class of service). The files within a resource group's
> > directory build on that. From your side I have not seen any effort in
> > aligning the sector cache function with the resctrl architecture but instead
> you are just changing resctrl interface to match the A64FX architecture.
> >
> > Could you please take a moment to understand what resctrl is and how
> > it could be mapped to A64FX in a coherent way?
>
> Previously, my idea is based on how to make instructions use different sectors
> in one task. After I studied resctrl, to utilize resctrl architecture on A64FX, I
> think it$B!G(Bs better to assign one sector to one task. Thanks for your idea that
> "sectors" could be considered the same as the resctrl "classes of service".
>
> Based on your idea, I am considering the implementation details.
> In this email, I will explain the outline of new proposal, and then please allow
> me to confirm a few technologies about resctrl.
Could you give me some comments & advices?
Best regards,
Tan Shaopeng
> The outline of my proposal is as follows.
> - Add a sector function equivalent to Intel's CAT function into resctrl.
> (divide shared L2 cache into multiple partitions for multiple cores use)
> - Allocate one sector to one resource group (one CLOSID). Since one
> core can only be assigned to one resource group, on A64FX each core
> only uses one sector at a time.
> - Disable A64FX's HPC tag address override function. We only set each
> core's default sector value according to closid(default sector ID=CLOSID).
> - No L1 cache control since L1 cache is not shared for cores. It is not
> necessary to add L1 cache interface for schemata file.
> - No need to update schemata interface. Resctrl's L2 cache interface
> (L2: <cache_id0> = <cbm>; <cache_id1> = <cbm>; ...)
> will be used as it is. However, on A64FX, <cbm> does not indicate
> the position of cache partition, only indicate the number of
> cache ways (size).
>
> This is the smallest start of incorporating sector cache function into resctrl. I
> will consider if we could add more sector cache features into resctrl (e.g.
> selecting different sectors from one task) after finishing this.
>
> (some questions are below)
>
> > >
> > >> On 5/17/2021 1:31 AM, [email protected] wrote:
> >
> > > --------
> > > A64FX NUMA-PE-Cache Architecture:
> > > NUMA0:
> > > PE0:
> > > L1sector0,L1sector1,L1sector2,L1sector3
> > > PE1:
> > > L1sector0,L1sector1,L1sector2,L1sector3
> > > ...
> > > PE11:
> > > L1sector0,L1sector1,L1sector2,L1sector3
> > >
> > > L2sector0,1/L2sector2,3
> > > NUMA1:
> > > PE0:
> > > L1sector0,L1sector1,L1sector2,L1sector3
> > > ...
> > > PE11:
> > > L1sector0,L1sector1,L1sector2,L1sector3
> > >
> > > L2sector0,1/L2sector2,3
> > > NUMA2:
> > > ...
> > > NUMA3:
> > > ...
> > > --------
> > > In A64FX processor, one L1 sector cache capacity setting register is
> > > only for one PE and not shared among PEs. L2 sector cache maximum
> > > capacity setting registers are shared among PEs in same NUMA, and it
> > > is to be noted that changing these registers in one PE influences other PE.
> >
> > Understood. cache affinity is familiar to resctrl. When a CPU becomes
> > online it is discovered which caches/resources it has affinity to.
> > Resources then have CPU mask associated with them to indicate on which
> > CPU a register could be changed to configure the resource/cache. See
> > domain_add_cpu() and struct rdt_domain.
>
> Is the following understanding correct?
> Struct rdt_domain is a group of online CPUs that share a same cache instance.
> When a CPU is online(resctrl initialization), the domain_add_cpu() function
> add the online cpu to corresponding rdt_domain (in rdt_resource:domains list).
> For example, if there are
> 4 L2 cache instances, then there will be 4 rdt_domain in the list and each CPU
> is assigned to corresponding rdt_domain.
>
> The set values of cache/memory are stored in the *ctrl_val array (indexed by
> CLOSID) of struct rdt_domain. For example, in CAT function, the CBM value of
> CLOSID=x is stored in ctrl_val [x].
> When we create a resource group and write set values of cache into the
> schemata file, the update_domains() function updates the CBM value to
> ctrl_val [CLOSID = resource group ID] in rdt_domain and updates the CBM
> value to CBM register(MSR_IA32_Lx_CBM_BASE).
>
> > > The number of ways for L2 Sector ID (0,1 or 2,3) can be set through
> > > any PEs in same NUMA. The sector ID 0,1 and 2,3 are not available at
> > > the same time in same NUMA.
> > >
> > >
> > > I think, in your idea, a resource group will be created for each sector ID.
> > > (> "sectors" could be considered the same as the resctrl "classes of
> > > service") Then, an example of resource group is created as follows.
> > > $B!&(B L1: NUMAX-PEY-L1sector0 (X = 0,1,2,3.Y = 0,1,2 ... 11),
> > > $B!&(B L2: NUMAX-L2sector0 (X = 0,1,2,3)
> > >
> > > In this example, sector with same ID(0) of all PEs is allocated to
> > > resource group. The L1D caches are numbered from
> > > NUMA0_PE0-L1sector0(0) to NUMA4_PE11-L1sector0(47) and the L2
> > caches
> > > numbered from
> > > NUMA0-L2sector0(0) to NUM4-L2sector0(3).
> > > (NUMA number X is from 0-4, PE number Y is from 0-11)
> > > (1) The number of ways of NUMAX-PEY-L1sector0 can be set
> independently
> > > for each PEs (0-47). When run a task on this resource group,
> > > we cannot control on which PE the task is running on and how many
> > > cache ways the task is using.
> >
> > resctrl does not control the affinity on which PE/CPU a task is run.
> > resctrl is an interface with which to configure how resources are
> > allocated on the system. resctrl could thus provide interface with
> > which each sector of each cache instance is assigned a number of cache
> ways.
> > resctrl also provides an interface to assign a task with a class of
> > service (sector id?). Through this the task obtains access to all
> > resources that is allocated to the particular class of service (sector
> > id?). Depending on which CPU the task is running it may indeed
> > experience different performance if the sector id it is running with
> > does not have the same allocations on all cache instances. The affinity of the
> task needs to be managed separately using for example taskset.
> > Please see Documentation/x86/resctrl.rst "Examples for RDT allocation
> usage"
>
> In resctrl_sched_in(), there are comments as follow:
> /*
> * If this task has a closid/rmid assigned, use it.
> * Else use the closid/rmid assigned to this cpu.
> */
> I thought when we write PID to tasks file, this task (PID) will only run on the
> CPUs which are specified in cpus file in the same resource group. So, the
> task_struct's closid and cpu's closid is the same.
> When task's closid is different from cpu's closid?
>
>
> Best regards,
> Tan Shaopeng
Hi Tan Shaopeng,
On 7/7/2021 4:26 AM, [email protected] wrote:
>>> Sorry, I have not explained A64FX's sector cache function well yet.
>>> I think I need explain this function from different perspective.
>>
>> You have explained the A64FX's sector cache function well. I have also read
>> both specs to understand it better. It appears to me that you are not considering
>> the resctrl architecture as part of your solution but instead just forcing your
>> architecture onto the resctrl filesystem. For example, in resctrl the resource
>> groups are not just a directory structure but has significance in what is being
>> represented within the directory (a class of service). The files within a resource
>> group's directory build on that. From your side I have not seen any effort in
>> aligning the sector cache function with the resctrl architecture but instead you
>> are just changing resctrl interface to match the A64FX architecture.
>>
>> Could you please take a moment to understand what resctrl is and how it could
>> be mapped to A64FX in a coherent way?
>
> Previously, my idea is based on how to make instructions use different
> sectors in one task. After I studied resctrl, to utilize resctrl
> architecture on A64FX, I think it’s better to assign one sector to
> one task. Thanks for your idea that "sectors" could be considered the
> same as the resctrl "classes of service".
>
> Based on your idea, I am considering the implementation details.
> In this email, I will explain the outline of new proposal, and then
> please allow me to confirm a few technologies about resctrl.
>
> The outline of my proposal is as follows.
> - Add a sector function equivalent to Intel's CAT function into resctrl.
> (divide shared L2 cache into multiple partitions for multiple cores use)
> - Allocate one sector to one resource group (one CLOSID). Since one
> core can only be assigned to one resource group, on A64FX each core
> only uses one sector at a time.
ok, so a sector is a portion of cache and matches with what can be
represented with a resource group.
The second part of your comment is not clear to me. In the first part
you mention: "one core can only be assigned to one resource group" -
this seems to indicate some static assignment between cores and sectors
and if this is the case this needs more thinking since the current
implementation assumes that any core that can access the cache can
access all resource groups associated with that cache. On the other
hand, you mention "on A64FX each core only uses one sector at a time" -
this now sounds dynamic and is how resctrl works since the CPU is
assigned a single class of service to indicate all resources accessible
to it.
> - Disable A64FX's HPC tag address override function. We only set each
> core's default sector value according to closid(default sector ID=CLOSID).
> - No L1 cache control since L1 cache is not shared for cores. It is not
> necessary to add L1 cache interface for schemata file.
> - No need to update schemata interface. Resctrl's L2 cache interface
> (L2: <cache_id0> = <cbm>; <cache_id1> = <cbm>; ...)
> will be used as it is. However, on A64FX, <cbm> does not indicate
> the position of cache partition, only indicate the number of
> cache ways (size).
From what I understand the upcoming MPAM support would make this easier
to do.
>
> This is the smallest start of incorporating sector cache function into
> resctrl. I will consider if we could add more sector cache features
> into resctrl (e.g. selecting different sectors from one task) after
> finishing this.
>
> (some questions are below)
>
>>>
>>>> On 5/17/2021 1:31 AM, [email protected] wrote:
>>
>>> --------
>>> A64FX NUMA-PE-Cache Architecture:
>>> NUMA0:
>>> PE0:
>>> L1sector0,L1sector1,L1sector2,L1sector3
>>> PE1:
>>> L1sector0,L1sector1,L1sector2,L1sector3
>>> ...
>>> PE11:
>>> L1sector0,L1sector1,L1sector2,L1sector3
>>>
>>> L2sector0,1/L2sector2,3
>>> NUMA1:
>>> PE0:
>>> L1sector0,L1sector1,L1sector2,L1sector3
>>> ...
>>> PE11:
>>> L1sector0,L1sector1,L1sector2,L1sector3
>>>
>>> L2sector0,1/L2sector2,3
>>> NUMA2:
>>> ...
>>> NUMA3:
>>> ...
>>> --------
>>> In A64FX processor, one L1 sector cache capacity setting register is
>>> only for one PE and not shared among PEs. L2 sector cache maximum
>>> capacity setting registers are shared among PEs in same NUMA, and it
>>> is to be noted that changing these registers in one PE influences other PE.
>>
>> Understood. cache affinity is familiar to resctrl. When a CPU becomes online it
>> is discovered which caches/resources it has affinity to.
>> Resources then have CPU mask associated with them to indicate on which
>> CPU a register could be changed to configure the resource/cache. See
>> domain_add_cpu() and struct rdt_domain.
>
> Is the following understanding correct?
> Struct rdt_domain is a group of online CPUs that share a same cache
> instance. When a CPU is online(resctrl initialization),
> the domain_add_cpu() function add the online cpu to corresponding
> rdt_domain (in rdt_resource:domains list). For example, if there are
> 4 L2 cache instances, then there will be 4 rdt_domain in the list and
> each CPU is assigned to corresponding rdt_domain.
Correct.
>
> The set values of cache/memory are stored in the *ctrl_val array
> (indexed by CLOSID) of struct rdt_domain. For example, in CAT function,
> the CBM value of CLOSID=x is stored in ctrl_val [x].
> When we create a resource group and write set values of cache into
> the schemata file, the update_domains() function updates the CBM value
> to ctrl_val [CLOSID = resource group ID] in rdt_domain and updates the
> CBM value to CBM register(MSR_IA32_Lx_CBM_BASE).
For the most part, yes. The only part that I would like to clarify is
that each CLOSID is represented by a different register, which register
is updated depends on which CLOSID is changed. Could be written as
MSR_IA32_L2_CBM_CLOSID/MSR_IA32_L3_CBM_CLOSID. The "BASE" register is
CLOSID 0, the default, and the other registers are determined as offset
from it.
Also, the registers have the scope of the resource/cache. So, for
example, if CPU 0 and CPU 1 share a L2 cache then it is only necessary
to update the register on one of these CPUs.
>
>>> The number of ways for L2 Sector ID (0,1 or 2,3) can be set through
>>> any PEs in same NUMA. The sector ID 0,1 and 2,3 are not available at
>>> the same time in same NUMA.
>>>
>>>
>>> I think, in your idea, a resource group will be created for each sector ID.
>>> (> "sectors" could be considered the same as the resctrl "classes of
>>> service") Then, an example of resource group is created as follows.
>>> ・ L1: NUMAX-PEY-L1sector0 (X = 0,1,2,3.Y = 0,1,2 ... 11),
>>> ・ L2: NUMAX-L2sector0 (X = 0,1,2,3)
>>>
>>> In this example, sector with same ID(0) of all PEs is allocated to
>>> resource group. The L1D caches are numbered from
>>> NUMA0_PE0-L1sector0(0) to NUMA4_PE11-L1sector0(47) and the L2
>> caches
>>> numbered from
>>> NUMA0-L2sector0(0) to NUM4-L2sector0(3).
>>> (NUMA number X is from 0-4, PE number Y is from 0-11)
>>> (1) The number of ways of NUMAX-PEY-L1sector0 can be set independently
>>> for each PEs (0-47). When run a task on this resource group,
>>> we cannot control on which PE the task is running on and how many
>>> cache ways the task is using.
>>
>> resctrl does not control the affinity on which PE/CPU a task is run.
>> resctrl is an interface with which to configure how resources are allocated on
>> the system. resctrl could thus provide interface with which each sector of each
>> cache instance is assigned a number of cache ways.
>> resctrl also provides an interface to assign a task with a class of service (sector
>> id?). Through this the task obtains access to all resources that is allocated to
>> the particular class of service (sector id?). Depending on which CPU the task is
>> running it may indeed experience different performance if the sector id it is
>> running with does not have the same allocations on all cache instances. The
>> affinity of the task needs to be managed separately using for example taskset.
>> Please see Documentation/x86/resctrl.rst "Examples for RDT allocation usage"
>
> In resctrl_sched_in(), there are comments as follow:
> /*
> * If this task has a closid/rmid assigned, use it.
> * Else use the closid/rmid assigned to this cpu.
> */
> I thought when we write PID to tasks file, this task (PID) will only
> run on the CPUs which are specified in cpus file in the same resource
> group. So, the task_struct's closid and cpu's closid is the same.
> When task's closid is different from cpu's closid?
resctrl does not manage the affinity of tasks.
Tony recently summarized the cpus file very well to me: The actual
semantics of the CPUs file is to associate a CLOSid for a task that is
in the default resctrl group – while it is running on one of the listed
CPUs.
To answer your question the task's closid could be different from the
CPU's closid if the task's closid is 0 while it is running on a CPU that
is in the cpus file of a non-default resource group.
You can see a summary of the decision flow in section "Resource
allocation rules" in Documentation/x86/resctrl.rst
The "cpus" file was created in support of the real-time use cases. In
these use cases a group of CPUs can be designated as supporting the
real-time work and with their own resource group and assigned the needed
resources to do the real-time work. A real-time task can then be started
with affinity to those CPUs and dynamically any kernel threads (that
will be started on the same CPU) doing work on behalf of this task would
be able to use the resources set aside for the real-time work.
Reinette
Hi Reinette,
> On 7/7/2021 4:26 AM, [email protected] wrote:
> >>> Sorry, I have not explained A64FX's sector cache function well yet.
> >>> I think I need explain this function from different perspective.
> >>
> >> You have explained the A64FX's sector cache function well. I have
> >> also read both specs to understand it better. It appears to me that
> >> you are not considering the resctrl architecture as part of your
> >> solution but instead just forcing your architecture onto the resctrl
> >> filesystem. For example, in resctrl the resource groups are not just
> >> a directory structure but has significance in what is being
> >> represented within the directory (a class of service). The files
> >> within a resource group's directory build on that. From your side I
> >> have not seen any effort in aligning the sector cache function with the
> resctrl architecture but instead you are just changing resctrl interface to match
> the A64FX architecture.
> >>
> >> Could you please take a moment to understand what resctrl is and how
> >> it could be mapped to A64FX in a coherent way?
> >
> > Previously, my idea is based on how to make instructions use different
> > sectors in one task. After I studied resctrl, to utilize resctrl
> > architecture on A64FX, I think it$B!G(Bs better to assign one sector to one
> > task. Thanks for your idea that "sectors" could be considered the same
> > as the resctrl "classes of service".
> >
> > Based on your idea, I am considering the implementation details.
> > In this email, I will explain the outline of new proposal, and then
> > please allow me to confirm a few technologies about resctrl.
> >
> > The outline of my proposal is as follows.
> > - Add a sector function equivalent to Intel's CAT function into resctrl.
> > (divide shared L2 cache into multiple partitions for multiple cores
> > use)
> > - Allocate one sector to one resource group (one CLOSID). Since one
> > core can only be assigned to one resource group, on A64FX each core
> > only uses one sector at a time.
>
> ok, so a sector is a portion of cache and matches with what can be represented
> with a resource group.
>
> The second part of your comment is not clear to me. In the first part you
> mention: "one core can only be assigned to one resource group" - this seems to
> indicate some static assignment between cores and sectors and if this is the
Sorry, does "static assignment between cores and sectors" mean
each core always use a fixed sector id? For example, core 0 always
use sector 0 at any case. It is not.
> case this needs more thinking since the current implementation assumes that
> any core that can access the cache can access all resource groups associated
> with that cache. On the other hand, you mention "on A64FX each core only uses
> one sector at a time" - this now sounds dynamic and is how resctrl works since
> the CPU is assigned a single class of service to indicate all resources
> accessible to it.
It is correct. Each core can be assigned to any resource group, and
each core only uses one sector at a time. Additionally, which sector
each core uses depends on the resource group (class of service) ID.
> > - Disable A64FX's HPC tag address override function. We only set each
> > core's default sector value according to closid(default sector
> ID=CLOSID).
> > - No L1 cache control since L1 cache is not shared for cores. It is not
> > necessary to add L1 cache interface for schemata file.
> > - No need to update schemata interface. Resctrl's L2 cache interface
> > (L2: <cache_id0> = <cbm>; <cache_id1> = <cbm>; ...)
> > will be used as it is. However, on A64FX, <cbm> does not indicate
> > the position of cache partition, only indicate the number of
> > cache ways (size).
>
> From what I understand the upcoming MPAM support would make this easier
> to do.
>
> >
> > This is the smallest start of incorporating sector cache function into
> > resctrl. I will consider if we could add more sector cache features
> > into resctrl (e.g. selecting different sectors from one task) after
> > finishing this.
> >
> > (some questions are below)
> >
> >>>
> >>>> On 5/17/2021 1:31 AM, [email protected] wrote:
> >>
> >>> --------
> >>> A64FX NUMA-PE-Cache Architecture:
> >>> NUMA0:
> >>> PE0:
> >>> L1sector0,L1sector1,L1sector2,L1sector3
> >>> PE1:
> >>> L1sector0,L1sector1,L1sector2,L1sector3
> >>> ...
> >>> PE11:
> >>> L1sector0,L1sector1,L1sector2,L1sector3
> >>>
> >>> L2sector0,1/L2sector2,3
> >>> NUMA1:
> >>> PE0:
> >>> L1sector0,L1sector1,L1sector2,L1sector3
> >>> ...
> >>> PE11:
> >>> L1sector0,L1sector1,L1sector2,L1sector3
> >>>
> >>> L2sector0,1/L2sector2,3
> >>> NUMA2:
> >>> ...
> >>> NUMA3:
> >>> ...
> >>> --------
> >>> In A64FX processor, one L1 sector cache capacity setting register is
> >>> only for one PE and not shared among PEs. L2 sector cache maximum
> >>> capacity setting registers are shared among PEs in same NUMA, and it
> >>> is to be noted that changing these registers in one PE influences other PE.
> >>
> >> Understood. cache affinity is familiar to resctrl. When a CPU becomes
> >> online it is discovered which caches/resources it has affinity to.
> >> Resources then have CPU mask associated with them to indicate on
> >> which CPU a register could be changed to configure the
> >> resource/cache. See
> >> domain_add_cpu() and struct rdt_domain.
> >
> > Is the following understanding correct?
> > Struct rdt_domain is a group of online CPUs that share a same cache
> > instance. When a CPU is online(resctrl initialization), the
> > domain_add_cpu() function add the online cpu to corresponding
> > rdt_domain (in rdt_resource:domains list). For example, if there are
> > 4 L2 cache instances, then there will be 4 rdt_domain in the list and
> > each CPU is assigned to corresponding rdt_domain.
>
> Correct.
>
> >
> > The set values of cache/memory are stored in the *ctrl_val array
> > (indexed by CLOSID) of struct rdt_domain. For example, in CAT
> > function, the CBM value of CLOSID=x is stored in ctrl_val [x].
> > When we create a resource group and write set values of cache into the
> > schemata file, the update_domains() function updates the CBM value to
> > ctrl_val [CLOSID = resource group ID] in rdt_domain and updates the
> > CBM value to CBM register(MSR_IA32_Lx_CBM_BASE).
>
> For the most part, yes. The only part that I would like to clarify is that each
> CLOSID is represented by a different register, which register is updated
> depends on which CLOSID is changed. Could be written as
> MSR_IA32_L2_CBM_CLOSID/MSR_IA32_L3_CBM_CLOSID. The "BASE"
> register is CLOSID 0, the default, and the other registers are determined as
> offset from it.
>
> Also, the registers have the scope of the resource/cache. So, for example, if
> CPU 0 and CPU 1 share a L2 cache then it is only necessary to update the
> register on one of these CPUs.
Thanks for your explanation. I understood it.
In addition, A64FX's L2 cache setting registers have similar scopes
of resource/cache, and only necessary to update the register on one of
these CPUs.
> >>> The number of ways for L2 Sector ID (0,1 or 2,3) can be set through
> >>> any PEs in same NUMA. The sector ID 0,1 and 2,3 are not available at
> >>> the same time in same NUMA.
> >>>
> >>>
> >>> I think, in your idea, a resource group will be created for each sector ID.
> >>> (> "sectors" could be considered the same as the resctrl "classes of
> >>> service") Then, an example of resource group is created as follows.
> >>> $B!&(B L1: NUMAX-PEY-L1sector0 (X = 0,1,2,3.Y = 0,1,2 ... 11),
> >>> $B!&(B L2: NUMAX-L2sector0 (X = 0,1,2,3)
> >>>
> >>> In this example, sector with same ID(0) of all PEs is allocated to
> >>> resource group. The L1D caches are numbered from
> >>> NUMA0_PE0-L1sector0(0) to NUMA4_PE11-L1sector0(47) and the L2
> >> caches
> >>> numbered from
> >>> NUMA0-L2sector0(0) to NUM4-L2sector0(3).
> >>> (NUMA number X is from 0-4, PE number Y is from 0-11)
> >>> (1) The number of ways of NUMAX-PEY-L1sector0 can be set
> independently
> >>> for each PEs (0-47). When run a task on this resource group,
> >>> we cannot control on which PE the task is running on and how
> many
> >>> cache ways the task is using.
> >>
> >> resctrl does not control the affinity on which PE/CPU a task is run.
> >> resctrl is an interface with which to configure how resources are
> >> allocated on the system. resctrl could thus provide interface with
> >> which each sector of each cache instance is assigned a number of cache
> ways.
> >> resctrl also provides an interface to assign a task with a class of
> >> service (sector id?). Through this the task obtains access to all
> >> resources that is allocated to the particular class of service
> >> (sector id?). Depending on which CPU the task is running it may
> >> indeed experience different performance if the sector id it is
> >> running with does not have the same allocations on all cache instances.
> The affinity of the task needs to be managed separately using for example
> taskset.
> >> Please see Documentation/x86/resctrl.rst "Examples for RDT allocation
> usage"
> >
> > In resctrl_sched_in(), there are comments as follow:
> > /*
> > * If this task has a closid/rmid assigned, use it.
> > * Else use the closid/rmid assigned to this cpu.
> > */
> > I thought when we write PID to tasks file, this task (PID) will only
> > run on the CPUs which are specified in cpus file in the same resource
> > group. So, the task_struct's closid and cpu's closid is the same.
> > When task's closid is different from cpu's closid?
>
> resctrl does not manage the affinity of tasks.
>
> Tony recently summarized the cpus file very well to me: The actual semantics of
> the CPUs file is to associate a CLOSid for a task that is in the default resctrl
> group ? while it is running on one of the listed CPUs.
>
> To answer your question the task's closid could be different from the CPU's
> closid if the task's closid is 0 while it is running on a CPU that is in the cpus file
> of a non-default resource group.
>
> You can see a summary of the decision flow in section "Resource allocation
> rules" in Documentation/x86/resctrl.rst
>
> The "cpus" file was created in support of the real-time use cases. In these use
> cases a group of CPUs can be designated as supporting the real-time work and
> with their own resource group and assigned the needed resources to do the
> real-time work. A real-time task can then be started with affinity to those CPUs
> and dynamically any kernel threads (that will be started on the same CPU)
> doing work on behalf of this task would be able to use the resources set aside
> for the real-time work.
Thanks for your explanation. I understood it.
I will implement this sector function, and if I have other questions,
please allow me to mail you.
Best regards,
Tan Shaopeng
Hi Tan Shaopeng,
On 7/21/2021 1:10 AM, [email protected] wrote:
> Hi Reinette,
>
>> On 7/7/2021 4:26 AM, [email protected] wrote:
>>>>> Sorry, I have not explained A64FX's sector cache function well yet.
>>>>> I think I need explain this function from different perspective.
>>>>
>>>> You have explained the A64FX's sector cache function well. I have
>>>> also read both specs to understand it better. It appears to me that
>>>> you are not considering the resctrl architecture as part of your
>>>> solution but instead just forcing your architecture onto the resctrl
>>>> filesystem. For example, in resctrl the resource groups are not just
>>>> a directory structure but has significance in what is being
>>>> represented within the directory (a class of service). The files
>>>> within a resource group's directory build on that. From your side I
>>>> have not seen any effort in aligning the sector cache function with the
>> resctrl architecture but instead you are just changing resctrl interface to match
>> the A64FX architecture.
>>>>
>>>> Could you please take a moment to understand what resctrl is and how
>>>> it could be mapped to A64FX in a coherent way?
>>>
>>> Previously, my idea is based on how to make instructions use different
>>> sectors in one task. After I studied resctrl, to utilize resctrl
>>> architecture on A64FX, I think it$B!G(Bs better to assign one sector to one
>>> task. Thanks for your idea that "sectors" could be considered the same
>>> as the resctrl "classes of service".
>>>
>>> Based on your idea, I am considering the implementation details.
>>> In this email, I will explain the outline of new proposal, and then
>>> please allow me to confirm a few technologies about resctrl.
>>>
>>> The outline of my proposal is as follows.
>>> - Add a sector function equivalent to Intel's CAT function into resctrl.
>>> (divide shared L2 cache into multiple partitions for multiple cores
>>> use)
>>> - Allocate one sector to one resource group (one CLOSID). Since one
>>> core can only be assigned to one resource group, on A64FX each core
>>> only uses one sector at a time.
>>
>> ok, so a sector is a portion of cache and matches with what can be represented
>> with a resource group.
>>
>> The second part of your comment is not clear to me. In the first part you
>> mention: "one core can only be assigned to one resource group" - this seems to
>> indicate some static assignment between cores and sectors and if this is the
>
> Sorry, does "static assignment between cores and sectors" mean
> each core always use a fixed sector id? For example, core 0 always
> use sector 0 at any case. It is not.
>
>> case this needs more thinking since the current implementation assumes that
>> any core that can access the cache can access all resource groups associated
>> with that cache. On the other hand, you mention "on A64FX each core only uses
>> one sector at a time" - this now sounds dynamic and is how resctrl works since
>> the CPU is assigned a single class of service to indicate all resources
>> accessible to it.
>
> It is correct. Each core can be assigned to any resource group, and
> each core only uses one sector at a time. Additionally, which sector
> each core uses depends on the resource group (class of service) ID.
Thank you for clarifying. From what I understand this could be supported
by existing resctrl flows.
...
>>> In resctrl_sched_in(), there are comments as follow:
>>> /*
>>> * If this task has a closid/rmid assigned, use it.
>>> * Else use the closid/rmid assigned to this cpu.
>>> */
>>> I thought when we write PID to tasks file, this task (PID) will only
>>> run on the CPUs which are specified in cpus file in the same resource
>>> group. So, the task_struct's closid and cpu's closid is the same.
>>> When task's closid is different from cpu's closid?
>>
>> resctrl does not manage the affinity of tasks.
>>
>> Tony recently summarized the cpus file very well to me: The actual semantics of
>> the CPUs file is to associate a CLOSid for a task that is in the default resctrl
>> group ? while it is running on one of the listed CPUs.
>>
>> To answer your question the task's closid could be different from the CPU's
>> closid if the task's closid is 0 while it is running on a CPU that is in the cpus file
>> of a non-default resource group.
>>
>> You can see a summary of the decision flow in section "Resource allocation
>> rules" in Documentation/x86/resctrl.rst
>>
>> The "cpus" file was created in support of the real-time use cases. In these use
>> cases a group of CPUs can be designated as supporting the real-time work and
>> with their own resource group and assigned the needed resources to do the
>> real-time work. A real-time task can then be started with affinity to those CPUs
>> and dynamically any kernel threads (that will be started on the same CPU)
>> doing work on behalf of this task would be able to use the resources set aside
>> for the real-time work.
>
> Thanks for your explanation. I understood it.
>
> I will implement this sector function, and if I have other questions,
> please allow me to mail you.
I will help where I can. You may also be interested in the work James is
busy with. See his latest series at
https://lore.kernel.org/lkml/[email protected]/
Reinette