I am experiencing a very slow(32Kbytes per second) transfer rate on an
ASUS P5P800 mobo. This occurs on a special case where I am sending
individual 32Kbyte messages from a second server.
I suspect the hardware, but am not sure how to come up with a 'good'
regression test for this issue.
Configurations I have tried:
1. If I swap out the ethernet adapter(tried a intel 10/100 and intel
10/100/1000) the transfer rate jumps up into the MBytes / second.
2. If I do 'other' network activity on the box, like scp'ing' files
around, the transfer rate for my 32Kbyte packets goes up into the
Mbytes / second. So I am a little baffled with the behavior.
3. If I just 'scp' files around of various sizes the transfer rate goes
up into the Mbytes / second.
some of the relevant dmesg information:
eth0: Yukon Gigabit Ethernet 10/100/1000Base-T Adapter
PrefPort:A RlmtMode:Check Link State
...
eth0: network connection up using port A
speed: 1000
autonegotiation: yes
duplex mode: full
flowctrl: symmetric
role: slave
irq moderation: disabled
scatter-gather: disabled
tx-checksum: disabled
rx-checksum: disabled
lspci -vvv output for the ethernet adapter:
02:05.0 Ethernet controller: Marvell Technology Group Ltd. 88E8001
Gigabit Ethernet Controller (rev 13)
Subsystem: ASUSTeK Computer Inc. Marvell 88E8001 Gigabit
Ethernet Controller (Asus)
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop-
ParErr- Stepping- SERR+ FastB2B-
Status: Cap+ 66Mhz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort-
<TAbort- <MAbort- >SERR- <PERR-
Latency: 64 (5750ns min, 7750ns max), Cache Line Size 04
Interrupt: pin A routed to IRQ 7
Region 0: Memory at fbffc000 (32-bit, non-prefetchable)
[size=16K]
Region 1: I/O ports at e800 [size=256]
Expansion ROM at f0000000 [disabled] [size=128K]
Capabilities: [48] Power Management version 2
Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1
+,D2+,D3hot+,D3cold+)
Status: D0 PME-Enable- DSel=0 DScale=1 PME-
Capabilities: [50] Vital Product Data
The Marvel ethernet adapter is connected to a Linksys SD2005 10/100/1000
switch.
Any ideas why it would be doing this or a 'good' test for me to try?
Sean
On Thu, 03 Aug 2006 08:40:01 -0700
Sean Bruno <[email protected]> wrote:
> I am experiencing a very slow(32Kbytes per second) transfer rate on an
> ASUS P5P800 mobo. This occurs on a special case where I am sending
> individual 32Kbyte messages from a second server.
>
> I suspect the hardware, but am not sure how to come up with a 'good'
> regression test for this issue.
>
> Configurations I have tried:
>
> 1. If I swap out the ethernet adapter(tried a intel 10/100 and intel
> 10/100/1000) the transfer rate jumps up into the MBytes / second.
>
> 2. If I do 'other' network activity on the box, like scp'ing' files
> around, the transfer rate for my 32Kbyte packets goes up into the
> Mbytes / second. So I am a little baffled with the behavior.
>
> 3. If I just 'scp' files around of various sizes the transfer rate goes
> up into the Mbytes / second.
>
Which driver skge or sk98lin are you using? Sk98lin driver is being
obsoleted.
On Thu, 2006-08-03 at 09:40 -0700, Stephen Hemminger wrote:
> skge
I am using sk98lin under 2.6.17.7 ... should I give skge a shot?
Sean
On Thu, 03 Aug 2006 09:46:36 -0700
Sean Bruno <[email protected]> wrote:
> On Thu, 2006-08-03 at 09:40 -0700, Stephen Hemminger wrote:
> > skge
>
> I am using sk98lin under 2.6.17.7 ... should I give skge a shot?
>
> Sean
>
Yes, we will fix skge, sk98lin probably will stay busted...
You might be seeing checksum errors or flow control issues.
Look at ethtool stats (ethtool -S eth0)
and try with hardware checksumming off (ethtool -K rx off tx off)
--
On Thu, 2006-08-03 at 09:50 -0700, Stephen Hemminger wrote:
> e
ty
I have built skge and loaded it in leu of sk98lin and re-running my
failure case.
sean
On Thu, 2006-08-03 at 09:58 -0700, Sean Bruno wrote:
> On Thu, 2006-08-03 at 09:50 -0700, Stephen Hemminger wrote:
> > e
> ty
>
> I have built skge and loaded it in leu of sk98lin and re-running my
> failure case.
>
> sean
>
> -
Thanks for the guidance. Simply using the skge driver instead has
resolved the issue I was seeing.
Sean
Hi Sean,
Did you test with the current driver version 8.34, which is vailable on
our website?
Best regards,
Karim
Marvell(r) Semiconductor Germany GmbH
-------------------------------------
Karim Jamal
Technical Support Engineer
--------------------------------------
Phone: +49 (0) 7243502-330
Fax: +49 (0) 7243502-364
Mail: [email protected]
Web: http:\\http://www.syskonnect.de
-----Original Message-----
From: Sean Bruno [mailto:[email protected]]
Sent: Thursday, August 03, 2006 5:40 PM
To: [email protected]
Cc: [email protected]
Subject: sk98lin extremely slow transfer rate ASUS P5P800(2.6.17.7)
I am experiencing a very slow(32Kbytes per second) transfer rate on an
ASUS P5P800 mobo. This occurs on a special case where I am sending
individual 32Kbyte messages from a second server.
I suspect the hardware, but am not sure how to come up with a 'good'
regression test for this issue.
Configurations I have tried:
1. If I swap out the ethernet adapter(tried a intel 10/100 and intel
10/100/1000) the transfer rate jumps up into the MBytes / second.
2. If I do 'other' network activity on the box, like scp'ing' files
around, the transfer rate for my 32Kbyte packets goes up into the Mbytes
/ second. So I am a little baffled with the behavior.
3. If I just 'scp' files around of various sizes the transfer rate goes
up into the Mbytes / second.
some of the relevant dmesg information:
eth0: Yukon Gigabit Ethernet 10/100/1000Base-T Adapter
PrefPort:A RlmtMode:Check Link State
...
eth0: network connection up using port A
speed: 1000
autonegotiation: yes
duplex mode: full
flowctrl: symmetric
role: slave
irq moderation: disabled
scatter-gather: disabled
tx-checksum: disabled
rx-checksum: disabled
lspci -vvv output for the ethernet adapter:
02:05.0 Ethernet controller: Marvell Technology Group Ltd. 88E8001
Gigabit Ethernet Controller (rev 13)
Subsystem: ASUSTeK Computer Inc. Marvell 88E8001 Gigabit
Ethernet Controller (Asus)
Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop-
ParErr- Stepping- SERR+ FastB2B-
Status: Cap+ 66Mhz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort-
<TAbort- <MAbort- >SERR- <PERR-
Latency: 64 (5750ns min, 7750ns max), Cache Line Size 04
Interrupt: pin A routed to IRQ 7
Region 0: Memory at fbffc000 (32-bit, non-prefetchable)
[size=16K]
Region 1: I/O ports at e800 [size=256]
Expansion ROM at f0000000 [disabled] [size=128K]
Capabilities: [48] Power Management version 2
Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1
+,D2+,D3hot+,D3cold+)
Status: D0 PME-Enable- DSel=0 DScale=1 PME-
Capabilities: [50] Vital Product Data
The Marvel ethernet adapter is connected to a Linksys SD2005 10/100/1000
switch.
Any ideas why it would be doing this or a 'good' test for me to try?
Sean
Hi,
On Fri, Aug 04, 2006 at 08:50:33AM +0200, SysKonnect Support wrote:
> Hi Sean,
>
> Did you test with the current driver version 8.34, which is vailable on
> our website?
I've already observed the same behaviour (8.31 though). The problem was that
after sending a few thousands UDP packets, then next UDP packets would not
go out without some TCP traffic to "push" them outside. This is a real problem
on NFS (where I first noticed it). But for me, it happened only on Yukon
cards and not on Yukon2.
Hoping this helps,
Willy
> Best regards,
> Karim
>
> Marvell(r) Semiconductor Germany GmbH
> -------------------------------------
> Karim Jamal
> Technical Support Engineer
> --------------------------------------
> Phone: +49 (0) 7243502-330
> Fax: +49 (0) 7243502-364
> Mail: [email protected]
> Web: http:\\http://www.syskonnect.de
>
>
>
> -----Original Message-----
> From: Sean Bruno [mailto:[email protected]]
> Sent: Thursday, August 03, 2006 5:40 PM
> To: [email protected]
> Cc: [email protected]
> Subject: sk98lin extremely slow transfer rate ASUS P5P800(2.6.17.7)
>
>
> I am experiencing a very slow(32Kbytes per second) transfer rate on an
> ASUS P5P800 mobo. This occurs on a special case where I am sending
> individual 32Kbyte messages from a second server.
>
> I suspect the hardware, but am not sure how to come up with a 'good'
> regression test for this issue.
>
> Configurations I have tried:
>
> 1. If I swap out the ethernet adapter(tried a intel 10/100 and intel
> 10/100/1000) the transfer rate jumps up into the MBytes / second.
>
> 2. If I do 'other' network activity on the box, like scp'ing' files
> around, the transfer rate for my 32Kbyte packets goes up into the Mbytes
> / second. So I am a little baffled with the behavior.
>
> 3. If I just 'scp' files around of various sizes the transfer rate goes
> up into the Mbytes / second.
>
>
>
> some of the relevant dmesg information:
>
> eth0: Yukon Gigabit Ethernet 10/100/1000Base-T Adapter
> PrefPort:A RlmtMode:Check Link State
> ...
> eth0: network connection up using port A
> speed: 1000
> autonegotiation: yes
> duplex mode: full
> flowctrl: symmetric
> role: slave
> irq moderation: disabled
> scatter-gather: disabled
> tx-checksum: disabled
> rx-checksum: disabled
>
>
> lspci -vvv output for the ethernet adapter:
> 02:05.0 Ethernet controller: Marvell Technology Group Ltd. 88E8001
> Gigabit Ethernet Controller (rev 13)
> Subsystem: ASUSTeK Computer Inc. Marvell 88E8001 Gigabit
> Ethernet Controller (Asus)
> Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop-
> ParErr- Stepping- SERR+ FastB2B-
> Status: Cap+ 66Mhz+ UDF- FastB2B+ ParErr- DEVSEL=medium >TAbort-
> <TAbort- <MAbort- >SERR- <PERR-
> Latency: 64 (5750ns min, 7750ns max), Cache Line Size 04
> Interrupt: pin A routed to IRQ 7
> Region 0: Memory at fbffc000 (32-bit, non-prefetchable)
> [size=16K]
> Region 1: I/O ports at e800 [size=256]
> Expansion ROM at f0000000 [disabled] [size=128K]
> Capabilities: [48] Power Management version 2
> Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1
> +,D2+,D3hot+,D3cold+)
> Status: D0 PME-Enable- DSel=0 DScale=1 PME-
> Capabilities: [50] Vital Product Data
>
> The Marvel ethernet adapter is connected to a Linksys SD2005 10/100/1000
> switch.
>
> Any ideas why it would be doing this or a 'good' test for me to try?
>
> Sean
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
Hi Willy,
the current driver version 8.34 includes a new common module, so please
give it a try and let me know.
Best regards,
Karim
Marvell(r) Semiconductor Germany GmbH
-------------------------------------
Karim Jamal
Technical Support Engineer
--------------------------------------
Phone: +49 (0) 7243502-330
Fax: +49 (0) 7243502-364
Mail: [email protected]
Web: http:\\http://www.syskonnect.de
-----Original Message-----
From: Willy Tarreau [mailto:[email protected]]
Sent: Friday, August 04, 2006 9:00 AM
To: SysKonnect Support
Cc: Sean Bruno; [email protected]; [email protected]
Subject: Re: sk98lin extremely slow transfer rate ASUS P5P800(2.6.17.7)
Hi,
On Fri, Aug 04, 2006 at 08:50:33AM +0200, SysKonnect Support wrote:
> Hi Sean,
>
> Did you test with the current driver version 8.34, which is vailable
> on our website?
I've already observed the same behaviour (8.31 though). The problem was
that after sending a few thousands UDP packets, then next UDP packets
would not go out without some TCP traffic to "push" them outside. This
is a real problem on NFS (where I first noticed it). But for me, it
happened only on Yukon cards and not on Yukon2.
Hoping this helps,
Willy
> Best regards,
> Karim
>
> Marvell(r) Semiconductor Germany GmbH
> -------------------------------------
> Karim Jamal
> Technical Support Engineer
> --------------------------------------
> Phone: +49 (0) 7243502-330
> Fax: +49 (0) 7243502-364
> Mail: [email protected]
> Web: http:\\http://www.syskonnect.de
>
>
>
> -----Original Message-----
> From: Sean Bruno [mailto:[email protected]]
> Sent: Thursday, August 03, 2006 5:40 PM
> To: [email protected]
> Cc: [email protected]
> Subject: sk98lin extremely slow transfer rate ASUS P5P800(2.6.17.7)
>
>
> I am experiencing a very slow(32Kbytes per second) transfer rate on an
> ASUS P5P800 mobo. This occurs on a special case where I am sending
> individual 32Kbyte messages from a second server.
>
> I suspect the hardware, but am not sure how to come up with a 'good'
> regression test for this issue.
>
> Configurations I have tried:
>
> 1. If I swap out the ethernet adapter(tried a intel 10/100 and intel
> 10/100/1000) the transfer rate jumps up into the MBytes / second.
>
> 2. If I do 'other' network activity on the box, like scp'ing' files
> around, the transfer rate for my 32Kbyte packets goes up into the
> Mbytes / second. So I am a little baffled with the behavior.
>
> 3. If I just 'scp' files around of various sizes the transfer rate
> goes up into the Mbytes / second.
>
>
>
> some of the relevant dmesg information:
>
> eth0: Yukon Gigabit Ethernet 10/100/1000Base-T Adapter
> PrefPort:A RlmtMode:Check Link State
> ...
> eth0: network connection up using port A
> speed: 1000
> autonegotiation: yes
> duplex mode: full
> flowctrl: symmetric
> role: slave
> irq moderation: disabled
> scatter-gather: disabled
> tx-checksum: disabled
> rx-checksum: disabled
>
>
> lspci -vvv output for the ethernet adapter:
> 02:05.0 Ethernet controller: Marvell Technology Group Ltd. 88E8001
> Gigabit Ethernet Controller (rev 13)
> Subsystem: ASUSTeK Computer Inc. Marvell 88E8001 Gigabit
> Ethernet Controller (Asus)
> Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV+ VGASnoop-
> ParErr- Stepping- SERR+ FastB2B-
> Status: Cap+ 66Mhz+ UDF- FastB2B+ ParErr- DEVSEL=medium
> >TAbort-
> <TAbort- <MAbort- >SERR- <PERR-
> Latency: 64 (5750ns min, 7750ns max), Cache Line Size 04
> Interrupt: pin A routed to IRQ 7
> Region 0: Memory at fbffc000 (32-bit, non-prefetchable)
> [size=16K]
> Region 1: I/O ports at e800 [size=256]
> Expansion ROM at f0000000 [disabled] [size=128K]
> Capabilities: [48] Power Management version 2
> Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1
> +,D2+,D3hot+,D3cold+)
> Status: D0 PME-Enable- DSel=0 DScale=1 PME-
> Capabilities: [50] Vital Product Data
>
> The Marvel ethernet adapter is connected to a Linksys SD2005
> 10/100/1000 switch.
>
> Any ideas why it would be doing this or a 'good' test for me to try?
>
> Sean
>
> -
> To unsubscribe from this list: send the line "unsubscribe
> linux-kernel" in the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
On Fri, Aug 04, 2006 at 09:24:54AM +0200, SysKonnect Support wrote:
> Hi Willy,
>
> the current driver version 8.34 includes a new common module, so please
> give it a try and let me know.
OK, I will try ASAP (not before end of next week though, cause I don't have
the hardware with me).
> Best regards,
> Karim
Regards,
Willy