Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
MIME-Version: 1.0
References: <CA+1FSihpq_i-poiihdF0srE3fGXnncGtCMYqAGmNZu7fjNcY=w@mail.gmail.com>
 <CY4PR11MB15594EC5255084B8AE509068F02E0@CY4PR11MB1559.namprd11.prod.outlook.com>
 <1093525083.1169948.1598979998646@mail.yahoo.com> <559F4617-9633-4B40-BA55-E79305E20530@intel.com>
In-Reply-To: <559F4617-9633-4B40-BA55-E79305E20530@intel.com>
From:   Roman Shaposhnik <roman@zededa.com>
Date:   Wed, 2 Sep 2020 12:55:13 -0700
Message-ID: <CAMmSBy8vor1GBqNUSRjj=gNRBwDjzRWFcxuAU0EYkRDgwQ4K+g@mail.gmail.com>
Subject: Re: Various problems for the Xen for XenGT code and guide.
To:     "Lv, Zhiyuan" <zhiyuan.lv@intel.com>
Cc:     Jason Long <hack3rcon@yahoo.com>,
        Mario Marietto <marietto2008@gmail.com>,
        "igvt-g@lists.01.org" <igvt-g@lists.01.org>,
        "xen-devel@lists.xenproject.org" <xen-devel@lists.xenproject.org>,
        "xen-devel@lists.xen.org" <xen-devel@lists.xen.org>,
        "intel-gfx@lists.freedesktop.org" <intel-gfx@lists.freedesktop.org>,
        "linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
        "Li, Susie" <susie.li@intel.com>,
        "Tian, Kevin" <kevin.tian@intel.com>,
        "Li, Weinan Z" <weinan.z.li@intel.com>,
        "Downs, Mike" <mike.downs@intel.com>,
        "Xu, Terrence" <terrence.xu@intel.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Wed, Sep 2, 2020 at 5:48 AM Lv, Zhiyuan <zhiyuan.lv@intel.com> wrote:
>
> Hi,
>
> It is mainly due to the business priority change. XenGT project was origi=
nally created for data center usages with XEON E3 servers which have integr=
ated processor graphics. After SkyLake E3, there are no new servers capable=
 of running GVT-g, and Intel future graphics for data center will have diff=
erent approaches for GPU sharing. Another reason is the XenGT upstream diff=
iculty. Different from KVMGT which has been fully merged to upstream, Xen p=
art of GVT-g still has technical opens that are hard to close quickly.

This is extremely useful -- thanks for sharing. Any chance you can
elaborate on the later part "and Intel future graphics for data center
will have different approaches for GPU sharing"?

IOW, is there anything that Intel is cooking up that may help Xen in
that department?

Thanks,
Roman.