All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Marek Olšák" <maraeo-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>
To: John Brooks <john-xq/Ko7C6e2Bl57MIdRCFDg@public.gmane.org>
Cc: "Michel Dänzer" <michel-otUistvHUpPR7s880joybQ@public.gmane.org>,
	"amd-gfx mailing list"
	<amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org>
Subject: Re: [PATCH 0/3] drm/amdgpu: Tweaks for high pressure on CPU visible VRAM
Date: Fri, 19 May 2017 17:24:36 +0200	[thread overview]
Message-ID: <CAAxE2A6pcm8EiXDRO2kO64TWJRR4BVrzbpiVrLdzhh3uERzTUg@mail.gmail.com> (raw)
In-Reply-To: <1495163086-4747-1-git-send-email-john-xq/Ko7C6e2Bl57MIdRCFDg@public.gmane.org>

Where is your "attached" patch?

Marek

On Fri, May 19, 2017 at 5:04 AM, John Brooks <john@fastquake.com> wrote:
> I'm glad this is being worked on. However, somewhat to my surprise, this patch
> series didn't help Dying Light's BO eviction problem. For those who don't know,
> that game performs very badly in certain areas, and it is correlated with
> increased TTM eviction rates. Relevant screenshots of gallium HUD and sysprof:
>
> http://www.fastquake.com/images/screen-dlgalliumhud1-20170513-171241.png
> http://www.fastquake.com/images/screen-dlsysprof-20170515-225919.png
>
> I noticed last week that adding RADEON_DOMAIN_GTT to the domains in radeonsi
> (patch: http://www.fastquake.com/files/text/radeon-gtt.txt ) greatly improved
> performance in these areas, to the tune of about a 30fps increase. Obviously,
> putting GTT in every buffer's domain is not a proper solution. But it lead me
> to believe that perhaps the problem wasn't just the swapping of resident BOs,
> but the creation of new ones that only have VRAM in their domain, and they
> cause existing BOs to be evicted from visible VRAM unconditionally.
>
> The attached patch assigns GTT as the busy placement for newly created BOs that
> have the AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED flag, so that they will go to
> GTT if visible VRAM is full, instead of evicting established BOs. Since there
> is no way to know what the usage patterns of a new BO will be, we shouldn't
> evict established BOs (for which we have hypothetically had the opportunity to
> gather usage data) from visible VRAM for new, unknown BOs.
>
> With this patch I get hugely improved performance in Dying Light just like with
> the Mesa patch: I observed 30-40fps where I got 14 before, and 60fps where I
> got 40 before. TTM evictions and bytes moved have dropped to zero where they
> were exceedingly high before. Buffer evictions no longer dominate the prof
> trace. Screenshots:
>
> http://www.fastquake.com/images/screen-dl-gtt_busy_only-20170518-192602.png
> http://www.fastquake.com/images/screen-dlsysprof-gttpatch-20170518-223200.png
>
> --
> John Brooks (Frogging101)
>
> _______________________________________________
> amd-gfx mailing list
> amd-gfx@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/amd-gfx
_______________________________________________
amd-gfx mailing list
amd-gfx@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/amd-gfx

  parent reply	other threads:[~2017-05-19 15:24 UTC|newest]

Thread overview: 39+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-05-18  9:08 [PATCH 0/3] drm/amdgpu: Tweaks for high pressure on CPU visible VRAM Michel Dänzer
     [not found] ` <20170518090809.15916-1-michel-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-18  9:08   ` [PATCH 1/3] drm/amdgpu: Drop useless loops for placement restrictions Michel Dänzer
     [not found]     ` <20170518090809.15916-2-michel-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-18  9:17       ` Christian König
     [not found]         ` <6c3e7a55-d3b0-fc3f-39f4-e85934119e29-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org>
2017-05-18  9:47           ` Michel Dänzer
2017-05-18  9:08   ` [PATCH 2/3] drm/amdgpu: Don't evict other BOs from VRAM for page faults Michel Dänzer
2017-05-18  9:08   ` [PATCH 3/3] drm/amdgpu: Try evicting from CPU visible to invisible VRAM first Michel Dänzer
     [not found]     ` <20170518090809.15916-4-michel-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-18 15:43       ` John Brooks
     [not found]         ` <20170518154346.GA5730-6hIufAJW0g7Gr8qjsLp7YGXnswh1EIUO@public.gmane.org>
2017-05-19  1:20           ` Michel Dänzer
     [not found]             ` <b65d8565-47e6-39d7-12f7-f7b60cf9787e-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-19  1:43               ` John Brooks
     [not found]                 ` <20170519014314.GA29857-6hIufAJW0g7Gr8qjsLp7YGXnswh1EIUO@public.gmane.org>
2017-05-19  1:50                   ` Michel Dänzer
2017-05-19  3:04   ` [PATCH 0/3] drm/amdgpu: Tweaks for high pressure on CPU visible VRAM John Brooks
     [not found]     ` <1495163086-4747-1-git-send-email-john-xq/Ko7C6e2Bl57MIdRCFDg@public.gmane.org>
2017-05-19  3:04       ` [PATCH] drm/amdgpu: Place new CPU-accessbile BOs in GTT if visible VRAM is full John Brooks
     [not found]         ` <1495163086-4747-2-git-send-email-john-xq/Ko7C6e2Bl57MIdRCFDg@public.gmane.org>
2017-05-19  5:37           ` zhoucm1
2017-05-19  7:03           ` Michel Dänzer
     [not found]             ` <b2ecb0f8-cd1b-9a00-6ba6-082674fde776-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-19 15:52               ` Marek Olšák
     [not found]                 ` <CAAxE2A7DjWR8x1WYE=Fgqn7vZ7hyRdQHA3wKo_HcNdZ14KdmkA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-20  1:22                   ` Michel Dänzer
2017-05-19 19:23               ` John Brooks
     [not found]                 ` <20170519192352.GA21642-6hIufAJW0g7Gr8qjsLp7YGXnswh1EIUO@public.gmane.org>
2017-05-20  1:27                   ` Michel Dänzer
     [not found]                     ` <d0535946-a10d-3b77-0d94-c52098dce021-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-20  3:21                       ` John Brooks
     [not found]                         ` <20170520032101.GA16371-6hIufAJW0g7Gr8qjsLp7YGXnswh1EIUO@public.gmane.org>
2017-05-22  1:30                           ` Michel Dänzer
2017-05-19 15:24       ` Marek Olšák [this message]
     [not found]         ` <CAAxE2A6pcm8EiXDRO2kO64TWJRR4BVrzbpiVrLdzhh3uERzTUg-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-19 15:27           ` [PATCH 0/3] drm/amdgpu: Tweaks for high pressure on CPU visible VRAM John Brooks
     [not found]             ` <20170519152732.GA11484-6hIufAJW0g7Gr8qjsLp7YGXnswh1EIUO@public.gmane.org>
2017-05-19 15:47               ` Marek Olšák
     [not found]                 ` <CAAxE2A77Yu9c02Mu-wK3HD_gAKvPbN8q7aEtHcxrcOUhGG4Pfw-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-19 15:49                   ` Marek Olšák
2017-05-19 16:14   ` Marek Olšák
     [not found]     ` <CAAxE2A65deE+hue7oPMQ+=K5s8TEGZE+bJMNVOdp=LQuq_RnMQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-20  1:26       ` Michel Dänzer
     [not found]         ` <CAAxE2A6ZiNSswhArstqHpYnm30ieoDZj=j_-aFvZz1_ByXLOGA@mail.gmail.com>
     [not found]           ` <CAAxE2A4zdgFRh-GDqyDRrXmEWqjSrkfCo=Z41hfo-UNY0E5FhA@mail.gmail.com>
     [not found]             ` <CAAxE2A4zdgFRh-GDqyDRrXmEWqjSrkfCo=Z41hfo-UNY0E5FhA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-20  9:26               ` Marek Olšák
     [not found]                 ` <CAAxE2A67T3Mhw68thNewFtcdua3ufSP-FDX_RN153x-aLKUh1Q-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-20  9:35                   ` Michel Dänzer
2017-05-22 10:00                   ` Michel Dänzer
     [not found]                     ` <7649341f-6407-1a12-9674-fb65076125fc-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-22 10:09                       ` Marek Olšák
     [not found]                         ` <CAAxE2A446KFb+CiwdZusvdgRSfQS8jLgsCy4K7L0_tc00pvWKA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-22 18:20                           ` John Brooks
2017-05-23  0:45                           ` Michel Dänzer
     [not found]                             ` <c3dfa73a-e0c1-c6e9-0264-3c52e444c297-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-23 10:38                               ` Marek Olšák
     [not found]                                 ` <CAAxE2A7tKvvmZ+qf8f06FJODWNLgHcv-XPiJJgLana-PcXBR_g-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-24  7:56                                   ` Michel Dänzer
     [not found]                                     ` <f04e8239-e8cb-1c6c-a0ae-581facadb3cf-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-24 11:03                                       ` Marek Olšák
     [not found]                                         ` <CAAxE2A5uha-MPx4B=ZeW9kf2qQXOTNcK7sxMJDtEk54S=us47g-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-24 11:27                                           ` Christian König
     [not found]                                             ` <eafd4ca6-fbbf-2bb4-be97-16f54fe402dd-ANTagKRnAhcb1SvskN2V4Q@public.gmane.org>
2017-05-25  3:31                                               ` Michel Dänzer
     [not found]                                                 ` <0535f00f-9095-6209-72e9-e6bb46569a63-otUistvHUpPR7s880joybQ@public.gmane.org>
2017-05-25 11:24                                                   ` Marek Olšák
     [not found]                                                     ` <CAAxE2A6YFjK1VPuxW=HROqKX+-ejvVJ=YneULp+cDkj033YAbA-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2017-05-26  0:46                                                       ` Michel Dänzer

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=CAAxE2A6pcm8EiXDRO2kO64TWJRR4BVrzbpiVrLdzhh3uERzTUg@mail.gmail.com \
    --to=maraeo-re5jqeeqqe8avxtiumwx3w@public.gmane.org \
    --cc=amd-gfx-PD4FTy7X32lNgt0PjOBp9y5qC8QIuHrW@public.gmane.org \
    --cc=john-xq/Ko7C6e2Bl57MIdRCFDg@public.gmane.org \
    --cc=michel-otUistvHUpPR7s880joybQ@public.gmane.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.