From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mout-p-103.mailbox.org (mout-p-103.mailbox.org [80.241.56.161]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.subspace.kernel.org (Postfix) with ESMTPS id B5FC837BE64 for ; Mon, 10 Aug 2026 08:28:22 +0000 (UTC) Authentication-Results: smtp.subspace.kernel.org; arc=none smtp.client-ip=80.241.56.161 ARC-Seal:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1786350504; cv=none; b=KKLL1uJuVif2h4vX+QoANtY8x8ZykUJhjGWNn4rXr0apB3IGdhwXnvSPwMwkDjwVyqmbvjLvBQUJEAxEl/ysuuluWeGQCX10Hsj9xfzw5eOTgQskXTgV+CM4a6XuRkWbHoD67dl1vsIfAAQeEfOD+hZXq/Oy0aBrT5NMar4aHm0= ARC-Message-Signature:i=1; a=rsa-sha256; d=subspace.kernel.org; s=arc-20240116; t=1786350504; c=relaxed/simple; bh=ldqOSfs+yOOLvTvAeLkNgY9UCg41jz+vPOLK4eT1X5k=; h=Message-ID:Subject:From:To:Cc:Date:In-Reply-To:References: Content-Type:MIME-Version; b=gdlIp/7a31eaSdmaAeHdkjv3uDURQmRojRlzsItPf0bCJIRkqGxVy3te6BFbJ+O+cmqfB0XyLZn6g6B/Kc3GEO9HawnyXAAyIjJt7wcsy5mlw9VjYigWv1thkp4vdokOQatx6y0QEF78Ko7+VYIbegyXge+MtGbZvYb4Di9CLxU= ARC-Authentication-Results:i=1; smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=mailbox.org; spf=pass smtp.mailfrom=mailbox.org; dkim=pass (2048-bit key) header.d=mailbox.org header.i=@mailbox.org header.b=TO+sSvTH; arc=none smtp.client-ip=80.241.56.161 Authentication-Results: smtp.subspace.kernel.org; dmarc=pass (p=reject dis=none) header.from=mailbox.org Authentication-Results: smtp.subspace.kernel.org; spf=pass smtp.mailfrom=mailbox.org Authentication-Results: smtp.subspace.kernel.org; dkim=pass (2048-bit key) header.d=mailbox.org header.i=@mailbox.org header.b="TO+sSvTH" Received: from smtp2.mailbox.org (smtp2.mailbox.org [IPv6:2001:67c:2050:b231:465::2]) (using TLSv1.3 with cipher TLS_AES_256_GCM_SHA384 (256/256 bits) key-exchange x25519 server-signature RSA-PSS (4096 bits) server-digest SHA256) (No client certificate requested) by mout-p-103.mailbox.org (Postfix) with ESMTPS id 4hJSZ56WRTzKnR3; Mon, 10 Aug 2026 10:28:13 +0200 (CEST) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=mailbox.org; s=mail20150812; t=1786350499; h=from:from:reply-to:reply-to:subject:subject:date:date: message-id:message-id:to:to:cc:cc:mime-version:mime-version: content-type:content-type: content-transfer-encoding:content-transfer-encoding: in-reply-to:in-reply-to:references:references; bh=9yu6LEk2UcyjL9r965P8ZavzTzFAgQo0BfGlpvZT/OE=; b=TO+sSvTHYAlWYgEcUG/JQbXVlcu6tw7XIJZt+5rLQCLU39Iu2nwwzo0lAX2izkWGHYYgah 35eD3EGGny2IRS1Ysy0c0rJe8psXiYt2WRoRl3SiLkLVWI2g3YL6/a23swDOkuJjLSQtW7 ms3wADbqyooWyWXXalMEt9rrwn1BS7kfC3yjoY5bs2JhSHQhGkPwPPsrhNNBVxCj++R9Et 0Kzx7Si4k1XGQCRQJwQV/6Ldyik/Q65mOiNxOL9/gfZHaHZXAA+sRFfKDXafKQUsQTa2po +9SbYoatX4JI51IqMIUOZKB44Qse6aPGJeF2PjN6A0o68NZhzH7R/79L2G4I0Q== Authentication-Results: outgoing_mbo_mout; dkim=none; spf=pass (outgoing_mbo_mout: domain of phasta@mailbox.org designates 2001:67c:2050:b231:465::2 as permitted sender) smtp.mailfrom=phasta@mailbox.org Message-ID: Subject: Re: [REGRESSION] drm/sched: FAIR policy causes serious performance degradation at max GPU load on 9070XT From: Philipp Stanner Reply-To: phasta@kernel.org To: Luke.Wildhardt@proton.me, "matthew.brost@intel.com" , "dakr@kernel.org" , "phasta@kernel.org" , "ckoenig.leichtzumerken@gmail.com" , "dri-devel@lists.freedesktop.org" Cc: "regressions@lists.linux.dev" , "amd-gfx@lists.freedesktop.org" , "linux-kernel@vger.kernel.org" , Tvrtko Ursulin Date: Mon, 10 Aug 2026 10:28:04 +0200 In-Reply-To: References: Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Precedence: bulk X-Mailing-List: linux-kernel@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: MIME-Version: 1.0 X-MBO-RS-ID: 45f535531c94d2e02af X-MBO-RS-META: tgkpc1exkfrzt6o87it7isjs8m6bthoo X-Rspamd-Queue-Id: 4hJSZ56WRTzKnR3 +Cc Tvrtko On Sat, 2026-08-08 at 23:33 +0000, Luke.Wildhardt@proton.me wrote: > Since the DRM scheduler default policy was switched to FAIR in 7.2, susta= ined 100% GPU load in-game on my RX 9070 XT causes severe performance degra= dation. In the example title running through Proton, Project Silverfish, th= e foreground application degrades to roughly 10 fps or freezes outright whi= le audio continues, and the KDE Plasma Wayland session can lock up entirely= , requiring a reboot or killing the compositor to recover. The problem is n= ot limited to games. Running the game in the background + alt tabbing, then= playing a YouTube video can also cause the entire desktop to freeze.=20 >=20 > Then, if the game starts to stutter, either alt-tabbing, or opening up a = menu in game (which reduced the load on the GPU) immediately stops the stut= tering.=20 >=20 > The issue is reliably reproducible under sustained GPU saturation, althou= gh the time-to-failure varies. Increasing shadow load (which decreases in-g= ame frame rate) substantially shortens the time-to-failure. >=20 > In game, MangoHud does not reveal any frametime discrepancy.=20 >=20 > Bisection / test matrix >=20 > Reproducer: Project Silverfish (Proton) at settings that saturate the GPU= , with a second GPU client active (browser video). Other titles show the sa= me symptom under sustained load; I have not yet re-tested those specific ti= tles against the fix. >=20 > Kernel Result > torvalds/master stock FAIR (default) FAIL > same as above + reverts below gpu_sched.sched_policy=3D2 (FAIR) FAIL > same as above + reverts below gpu_sched.sched_policy=3D1 (FIFO) PASS > 7.1.5 release PASS > drm-next FAIR (default) FAIL >=20 > Reverted Commits: >=20 > d09339388b77 drm/sched: Remove drm_sched_init_args->num_rqs > 2833a0512b4c drm/sched: Remove drm_sched_init_args->num_rqs usage > 16e7698bc04d drm/sched: Embed run queue singleton into the scheduler > 77a6809f1dc3 drm/sched: Remove FIFO and RR and simplify to a single run q= ueue > 45c211ddf92a drm/sched: Switch default policy to fair > 2462a0ce23b0 drm/amdgpu: Remove drm_sched_init_args->num_rqs usage >=20 >=20 >=20 > Kernel (failing):=C2=A0 7.2.0-rc6 (stock), also current drm-next > Kernel (working):=C2=A0 7.2.0-rc6-default-revert-00006-g7e5de9b07e3d with= sched_policy=3D1 > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0= =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 7.1 release > Distro:=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0= Arch Linux > Session:=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 KDE = Plasma 6.7.4 / Wayland (KWin), Qt 6.11.1, KF 6.28.0 > CPU:=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2= =A0=C2=A0=C2=A0 AMD Ryzen 9 9950X3D > RAM:=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2= =A0=C2=A0=C2=A0 64 GiB > GPU:=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2= =A0=C2=A0=C2=A0 Sapphire Pulse Radeon RX 9070 XT > =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0= =C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 03:00.0 [1002:7550] rev c0, subsys Sap= phire 1478 > Motherboard:=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 ASUS (AM5 / 800-series c= hipset) > Mesa:=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0= =C2=A0=C2=A0 26.1.6-arch1.1 (RADV, driverVersion 26.1.6) > Vulkan:=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0= instance 1.4.357 / device 1.4.354 >=20 >=20 >=20 >=20 > On the same reverted kernel build, changing only gpu_sched.sched_policy f= rom 2 (FAIR) to 1 (FIFO) changes the result from reliably failing to passin= g extended stress testing. >=20 >=20 >=20 > Possible mitigation options >=20 > Given that 7.2 is late in the release cycle and 77a6809f1dc3 removes FIFO= /RR as selectable policies, there is currently no runtime workaround for af= fected systems. >=20 >=20 > Revert 45c211ddf92a so FIFO remains the default while FAIR remains availa= ble for further testing. That might be the wisest thing to do; but let's hear if Tvrtko has an idea for a hotfix first. >=20 > If necessary, also revert 77a6809f1dc3 and dependent changes to restore r= untime policy selection. >=20 > Alternatively, fix the underlying FAIR regression if a suitable fix can b= e identified in time for 7.2. >=20 >=20 > I mention the revert options because the justification for removing FIFO/= RR included the absence of known regressions relative to those policies. Th= is report appears to provide at least one counterexample. >=20 >=20 > I am willing to test patches responsively and collect data if necessary.= =C2=A0=20 Thanks for your effort and help. I had dared to hope that we could get this one through without a regression, but reality always catches up.. P.