<html> <head> <meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1"> <style type="text/css" style="display:none;"> P {margin-top:0;margin-bottom:0;} </style> </head> <body dir="ltr"> <p style="font-family:Arial;font-size:10pt;color:#0078D7;margin:15pt;" align="Left"> [AMD Official Use Only - Internal Distribution Only]<br> </p> <br> <div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> Hi Tejun,</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> <br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> Thanks for taking the time to reply.</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> <br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> Perhaps we can even narrow things down to just gpu.weight/gpu.compute.weight as a start? In this aspect, is the key objection to the current implementation of gpu.compute.weight the work-conserving bit? This work-conserving requirement is probably what I have missed for the last two years (and hence going in circle.)</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> <br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> If this is the case, can you clarify/confirm the followings?</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> <br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> 1) Is resource scheduling goal of cgroup purely for the purpose of throughput? (at the expense of other scheduling goals such as latency.)</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> 2) If 1) is true, under what circumstances will the "Allocations" resource distribution model (as defined in the cgroup-v2) be acceptable?</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0); background-color: rgb(255, 255, 255);"> 3) If 1) is true, are things like cpuset from cgroup v1 no longer acceptable going forward?</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> <br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> To be clear, while some have framed this (time sharing vs spatial sharing) as a partisan issue, it is in fact a technical one. I have implemented the gpu cgroup support this way because we have a class of users that value low latency/low jitter/predictability/synchronicity. For example, they would like 4 tasks to share a GPU and they would like the tasks to start and finish at the same time.<br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> <br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> What is the rationale behind picking the Weight model over Allocations as the first acceptable implementation? Can't we have both work-conserving and non-work-conserving ways of distributing GPU resources? If we can, why not allow non-work-conserving implementation first, especially when we have users asking for such functionality?</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> <br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> Regards,</div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> Kenny<br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> <br> </div> <div style="font-family: Arial, Helvetica, sans-serif; font-size: 12pt; color: rgb(0, 0, 0);"> <br> </div> <hr tabindex="-1" style="display:inline-block; width:98%"> <div id="divRplyFwdMsg" dir="ltr"><font style="font-size: 11pt;" data-ogsc="" face="Calibri, sans-serif" color="#000000"><b>From:</b> Tejun Heo <htejun@gmail.com> on behalf of Tejun Heo <tj@kernel.org><br> <b>Sent:</b> Monday, April 13, 2020 3:11 PM<br> <b>To:</b> Kenny Ho <y2kenny@gmail.com><br> <b>Cc:</b> Ho, Kenny <Kenny.Ho@amd.com>; cgroups@vger.kernel.org <cgroups@vger.kernel.org>; dri-devel <dri-devel@lists.freedesktop.org>; amd-gfx list <amd-gfx@lists.freedesktop.org>; Deucher, Alexander <Alexander.Deucher@amd.com>; Koenig, Christian <Christian.Koenig@amd.com>; Kuehling, Felix <Felix.Kuehling@amd.com>; Greathouse, Joseph <Joseph.Greathouse@amd.com>; jsparks@cray.com <jsparks@cray.com><br> <b>Subject:</b> Re: [PATCH v2 00/11] new cgroup controller for gpu/drm subsystem</font> <div> </div> </div> <div class="BodyFragment"><font size="2"><span style="font-size:11pt"> <div class="PlainText">Hello, Kenny.<br> <br> On Tue, Mar 24, 2020 at 02:49:27PM -0400, Kenny Ho wrote:<br> > Can you elaborate more on what are the missing pieces?<br> <br> Sorry about the long delay, but I think we've been going in circles for quite<br> a while now. Let's try to make it really simple as the first step. How about<br> something like the following?<br> <br> * gpu.weight (should it be gpu.compute.weight? idk) - A single number<br> per-device weight similar to io.weight, which distributes computation<br> resources in work-conserving way.<br> <br> * gpu.memory.high - A single number per-device on-device memory limit.<br> <br> The above two, if works well, should already be plenty useful. And my guess is<br> that getting the above working well will be plenty challenging already even<br> though it's already excluding work-conserving memory distribution. So, let's<br> please do that as the first step and see what more would be needed from there.<br> <br> Thanks.<br> <br> -- <br> tejun<br> </div> </span></font></div> </div> </body> </html>