drm/i915/bdw: Use timeout mode for RC6 on bdw
Subsystems (2)
PATCH 1/1 drm/i915/bdw: Use timeout mode for RC6 on bdw
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
--- drivers/gpu/drm/i915/intel_pm.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-)
drivers/gpu/drm/i915/intel_pm.c +12 -4
@@ -3335,15 +3335,23 @@ static void gen8_enable_rps(struct drm_device *dev)for_each_ring (ring, dev_priv, unused)I915_WRITE (RING_MAX_IDLE (ring->mmio_base),10 );I915_WRITE (GEN6_RC_SLEEP ,0 ); -I915_WRITE (GEN6_RC6_THRESHOLD ,50000 );/* 50/125ms per EI */ +if (IS_BROADWELL (dev)) +I915_WRITE (GEN6_RC6_THRESHOLD ,625 );/* 800us/1.28 for TO */ +else +I915_WRITE (GEN6_RC6_THRESHOLD ,50000 );/* 50/125ms per EI */ /* 3: Enable RC6 */ if (intel_enable_rc6 (dev) &INTEL_RC6_ENABLE ) rc6_mask =GEN6_RC_CTL_RC6_ENABLE ;intel_print_rc6_info (dev, rc6_mask); -I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | -GEN6_RC_CTL_EI_MODE (1 ) | - rc6_mask); +if (IS_BROADWELL (dev)) +I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | +GEN7_RC_CTL_TO_MODE | + rc6_mask); +else +I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | +GEN6_RC_CTL_EI_MODE (1 ) | + rc6_mask);/* 4 Program defaults and thresholds for RPS*/ I915_WRITE(GEN6_RPNSWREQ,
Signature · 1 line
1.7.9.5
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>
How recent a nightly branch have you used to obtain these results? Chris just fixed some serious bugs in the gpu booster logic which would have affected all intermediate workloads.
-Daniel
Quoted 43 lines
--- drivers/gpu/drm/i915/intel_pm.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-)drivers/gpu/drm/i915/intel_pm.c +12
-4@@ -3335,15 +3335,23 @@ static void gen8_enable_rps(struct drm_device *dev)for_each_ring (ring, dev_priv, unused)I915_WRITE (RING_MAX_IDLE (ring->mmio_base),10 );I915_WRITE (GEN6_RC_SLEEP ,0 ); -I915_WRITE (GEN6_RC6_THRESHOLD ,50000 );/* 50/125ms per EI */ +if (IS_BROADWELL (dev)) +I915_WRITE (GEN6_RC6_THRESHOLD ,625 );/* 800us/1.28 for TO */ +else +I915_WRITE (GEN6_RC6_THRESHOLD ,50000 );/* 50/125ms per EI */ /* 3: Enable RC6 */ if (intel_enable_rc6 (dev) &INTEL_RC6_ENABLE ) rc6_mask =GEN6_RC_CTL_RC6_ENABLE ;intel_print_rc6_info (dev, rc6_mask); -I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | -GEN6_RC_CTL_EI_MODE (1 ) | - rc6_mask); +if (IS_BROADWELL (dev)) +I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | +GEN7_RC_CTL_TO_MODE | + rc6_mask); +else +I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | +GEN6_RC_CTL_EI_MODE (1 ) | + rc6_mask);/* 4 Program defaults and thresholds for RPS*/ I915_WRITE(GEN6_RPNSWREQ,Signature · 6 lines
1.7.9.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch
Quoted 6 lines · Daniel Vetter
Quoted 5 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Quoted 8 lines
Quoted 3 lines
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>How recent a nightly branch have you used to obtain these results? Chris just fixed some serious bugs in the gpu booster logic which would have affected all intermediate workloads.
-Daniel
He must not be using nightly if he has any BDW RC6 residency at all.
Quoted 53 lines
Quoted 44 lines
--- drivers/gpu/drm/i915/intel_pm.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-)drivers/gpu/drm/i915/intel_pm.c +12
-4@@ -3335,15 +3335,23 @@ static void gen8_enable_rps(struct drm_device *dev)for_each_ring (ring, dev_priv, unused)I915_WRITE (RING_MAX_IDLE (ring->mmio_base),10 );I915_WRITE (GEN6_RC_SLEEP ,0 ); -I915_WRITE (GEN6_RC6_THRESHOLD ,50000 );/* 50/125ms per EI */ +if (IS_BROADWELL (dev)) +I915_WRITE (GEN6_RC6_THRESHOLD ,625 );/* 800us/1.28 for TO */ +else +I915_WRITE (GEN6_RC6_THRESHOLD ,50000 );/* 50/125ms per EI */ /* 3: Enable RC6 */ if (intel_enable_rc6 (dev) &INTEL_RC6_ENABLE ) rc6_mask =GEN6_RC_CTL_RC6_ENABLE ;intel_print_rc6_info (dev, rc6_mask); -I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | -GEN6_RC_CTL_EI_MODE (1 ) | - rc6_mask); +if (IS_BROADWELL (dev)) +I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | +GEN7_RC_CTL_TO_MODE | + rc6_mask); +else +I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | +GEN6_RC_CTL_EI_MODE (1 ) | + rc6_mask);/* 4 Program defaults and thresholds for RPS*/ I915_WRITE(GEN6_RPNSWREQ,Signature · 6 lines
1.7.9.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfxSignature · 7 lines
Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 1 line
Ben Widawsky, Intel Open Source Technology Center
Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only.
The difference is particularly noticeable with video playback.Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>
Quoted 7 lines
Quoted 4 lines
How recent a nightly branch have you used to obtain these results? Chris just fixed some serious bugs in the gpu booster logic which would have affected all intermediate workloads.
-DanielHe must not be using nightly if he has any BDW RC6 residency at all.
[TOR:] Ben is correct. I was testing mostly with a kernel for Android. I also tested with Ben's broadwell branch and saw similar improvement.
Quoted 13 lines · Tom
Quoted 13 lines
Quoted 12 lines
Quoted 7 lines
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only.
The difference is particularly noticeable with video playback.Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>How recent a nightly branch have you used to obtain these results? Chris just fixed some serious bugs in the gpu booster logic which would have affected all intermediate workloads.
-Daniel
Quoted 4 lines
Quoted 2 lines
He must not be using nightly if he has any BDW RC6 residency at all.
[TOR:] Ben is correct. I was testing mostly with a kernel for Android. I also tested with Ben's broadwell branch and saw similar improvement.
Ok I think it'd be good to have this when we actually merge/enable the final pieces of the bdw rc6 code. Can you please work together with Ben to make sure this patch isn't lost?
Or who is shepherding bdw rc6 nowadays?
-Daniel
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch
Quoted 19 lines · Daniel Vetter
Quoted 13 lines · Tom
Quoted 13 lines
Quoted 12 lines
Quoted 7 lines
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only.
The difference is particularly noticeable with video playback.Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>How recent a nightly branch have you used to obtain these results? Chris just fixed some serious bugs in the gpu booster logic which would have affected all intermediate workloads.
-DanielQuoted 4 lines
Quoted 2 lines
He must not be using nightly if he has any BDW RC6 residency at all.
[TOR:] Ben is correct. I was testing mostly with a kernel for Android. I also tested with Ben's broadwell branch and saw similar improvement.
Quoted 6 lines
Ok I think it'd be good to have this when we actually merge/enable the final pieces of the bdw rc6 code. Can you please work together with Ben to make sure this patch isn't lost?
Or who is shepherding bdw rc6 nowadays?
-Daniel
Locally, people are still using my RC6 "workaround." We're waiting on Mika at the moment. I am going to merge this patch to my bdw-rc6/broadwell branches.
I think it would be great to get a little more detail under what circumstances this improves residency. We can discuss that internally.
Signature · 1 line
Ben Widawsky, Intel Open Source Technology Center
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>
I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Quoted 43 lines
--- drivers/gpu/drm/i915/intel_pm.c | 16 ++++++++++++---- 1 file changed, 12 insertions(+), 4 deletions(-)drivers/gpu/drm/i915/intel_pm.c +12
-4@@ -3335,15 +3335,23 @@ static void gen8_enable_rps(struct drm_device *dev)for_each_ring (ring, dev_priv, unused)I915_WRITE (RING_MAX_IDLE (ring->mmio_base),10 );I915_WRITE (GEN6_RC_SLEEP ,0 ); -I915_WRITE (GEN6_RC6_THRESHOLD ,50000 );/* 50/125ms per EI */ +if (IS_BROADWELL (dev)) +I915_WRITE (GEN6_RC6_THRESHOLD ,625 );/* 800us/1.28 for TO */ +else +I915_WRITE (GEN6_RC6_THRESHOLD ,50000 );/* 50/125ms per EI */ /* 3: Enable RC6 */ if (intel_enable_rc6 (dev) &INTEL_RC6_ENABLE ) rc6_mask =GEN6_RC_CTL_RC6_ENABLE ;intel_print_rc6_info (dev, rc6_mask); -I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | -GEN6_RC_CTL_EI_MODE (1 ) | - rc6_mask); +if (IS_BROADWELL (dev)) +I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | +GEN7_RC_CTL_TO_MODE | + rc6_mask); +else +I915_WRITE (GEN6_RC_CONTROL ,GEN6_RC_CTL_HW_ENABLE | +GEN6_RC_CTL_EI_MODE (1 ) | + rc6_mask);/* 4 Program defaults and thresholds for RPS*/ I915_WRITE(GEN6_RPNSWREQ,Signature · 6 lines
1.7.9.5 _______________________________________________ Intel-gfx mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/intel-gfx
Signature · 1 line
Ben Widawsky, Intel Open Source Technology Center
Quoted 6 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 5 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Quoted 6 lines
Quoted 3 lines
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
Quoted 13 lines · Kristen Carlson Accardi
Quoted 10 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>
Quoted 6 lines
Quoted 2 lines
I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
I'm not actually sure. You can try it and let me know. I haven't had any time to verify the rebase. We can check my hack.
Signature · 1 line
Ben Widawsky, Intel Open Source Technology Center
Quoted 17 lines · Ben Widawsky
Quoted 16 lines · Kristen Carlson Accardi
Quoted 12 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Quoted 6 lines
Quoted 3 lines
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
I'm not actually sure. You can try it and let me know. I haven't had any time to verify the rebase. We can check my hack.
Note that in -nightly you also have to update sanitize_rc6_option() along with intel_enable_gt_powersave() and intel_disable_gt_powersave() since atm these keep RC6 disabled on BDW.
--Imre
Quoted 22 lines · On Thu, 01 May 2014 00:03:15 +0300 Imre Deak <[email protected]> wrote:
Quoted 21 lines · Ben Widawsky
Quoted 19 lines · Kristen Carlson Accardi
Quoted 6 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 5 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Quoted 6 lines
Quoted 3 lines
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
Quoted 9 lines
Quoted 2 lines
I'm not actually sure. You can try it and let me know. I haven't had any time to verify the rebase. We can check my hack.
Note that in -nightly you also have to update
sanitize_rc6_option()along withintel_enable_gt_powersave()andintel_disable_gt_powersave()since atm these keep RC6 disabled on BDW.--Imre
Yes, I reverted fb5ed3b201fe5670c9ffeec3b5f6ff044d543c9e and was able to see some rc6 residency. With the idle workload, residency appears to be similar to before, so no regression.
Quoted 32 lines · Kristen Carlson Accardi
Quoted 29 lines · On Thu, 01 May 2014 00:03:15 +0300 Imre Deak <[email protected]> wrote:
Quoted 17 lines · Ben Widawsky
Quoted 16 lines · Kristen Carlson Accardi
Quoted 12 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Quoted 6 lines
Quoted 3 lines
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
I'm not actually sure. You can try it and let me know. I haven't had any time to verify the rebase. We can check my hack.
Note that in -nightly you also have to update
sanitize_rc6_option()along withintel_enable_gt_powersave()andintel_disable_gt_powersave()since atm these keep RC6 disabled on BDW.
Quoted 7 lines
Quoted 2 lines
--Imre
Yes, I reverted
fb5ed3b201fe5670c9ffeec3b5f6ff044d543c9eand was able to see some rc6 residency. With the idle workload, residency appears to be similar to before, so no regression.
Thanks. I'll squash this in where appropriate.
Signature · 1 line
Ben Widawsky, Intel Open Source Technology Center
Quoted 37 lines · Kristen Carlson Accardi
Quoted 36 lines
Quoted 31 lines · On Thu, 01 May 2014 00:03:15 +0300 Imre Deak <[email protected]> wrote:
Quoted 17 lines · Ben Widawsky
Quoted 16 lines · Kristen Carlson Accardi
Quoted 12 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only.
The difference is particularly noticeable with video playback.Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Quoted 6 lines
Quoted 3 lines
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
I'm not actually sure. You can try it and let me know. I haven't had any time to verify the rebase. We can check my hack.
Note that in -nightly you also have to update
sanitize_rc6_option()along withintel_enable_gt_powersave()andintel_disable_gt_powersave()since atm these keep RC6 disabled on BDW.--Imre
Quoted 8 lines
Quoted 3 lines
Yes, I reverted
fb5ed3b201fe5670c9ffeec3b5f6ff044d543c9eand was able to see some rc6 residency. With the idle workload, residency appears to be similar to before, so no regression.Thanks. I'll squash this in where appropriate.
--
Ben Widawsky, Intel Open Source Technology Center
[TOR:] Can we get this patch merged now that RC6 is working on drm-intel-nightly? Thanks,
Tom
Quoted 41 lines · Tom
Quoted 41 lines · Kristen Carlson Accardi
Quoted 39 lines
Quoted 22 lines · On Thu, 01 May 2014 00:03:15 +0300 Imre Deak <[email protected]> wrote:
Quoted 21 lines · Ben Widawsky
Quoted 19 lines · Kristen Carlson Accardi
Quoted 6 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 5 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only.
The difference is particularly noticeable with video playback.Quoted 6 lines
Quoted 3 lines
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
Quoted 9 lines
Quoted 2 lines
I'm not actually sure. You can try it and let me know. I haven't had any time to verify the rebase. We can check my hack.
Note that in -nightly you also have to update
sanitize_rc6_option()along withintel_enable_gt_powersave()andintel_disable_gt_powersave()since atm these keep RC6 disabled on BDW.--Imre
Yes, I reverted
fb5ed3b201fe5670c9ffeec3b5f6ff044d543c9eand was able to see some rc6 residency. With the idle workload, residency appears to be similar to before, so no regression.
Quoted 6 lines
Quoted 4 lines
Thanks. I'll squash this in where appropriate.
--
Ben Widawsky, Intel Open Source Technology Center[TOR:] Can we get this patch merged now that RC6 is working on drm-intel-nightly?
Needs some review from bdw people. Also some relative residency improvement date should be added to the commit message (yes, we're allowed to do that now officially).
-Daniel
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch
Quoted 55 lines
From: Daniel Vetter [mailto:[email protected]] On Behalf Of Daniel Vetter
Sent: Monday, June 02, 2014 1:26 AM
To: O'Rourke, Tom
Cc: [email protected]; Ben Widawsky; Kristen Carlson Accardi
Subject: Re: [Intel-gfx] [PATCH] drm/i915/bdw: Use timeout mode for RC6 on bdwQuoted 6 lines · Tom
Quoted 6 lines · Kristen Carlson Accardi
Quoted 5 lines
Quoted 2 lines · On Thu, 01 May 2014 00:03:15 +0300 Imre Deak <[email protected]> wrote:
Quoted 1 line · Ben Widawsky
On Wed, Apr 30, 2014 at 01:34:36PM -0700, Kristen Carlson Accardi
Quoted 40 lines · wrote:
Quoted 39 lines
Quoted 27 lines
Quoted 27 lines
Quoted 16 lines
Quoted 16 lines
Quoted 12 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Quoted 6 lines
Quoted 3 lines
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
I'm not actually sure. You can try it and let me know. I haven't had any time to verify the rebase. We can check my hack.
Note that in -nightly you also have to update
sanitize_rc6_option()along withintel_enable_gt_powersave()andintel_disable_gt_powersave()since atm these keep RC6 disabled on BDW.Quoted 7 lines
Quoted 2 lines
--Imre
Yes, I reverted
fb5ed3b201fe5670c9ffeec3b5f6ff044d543c9eand was able to see some rc6 residency. With the idle workload, residency appears to be similar to before, so no regression.Thanks. I'll squash this in where appropriate.
--
Ben Widawsky, Intel Open Source Technology Center
Quoted 8 lines
Quoted 1 line
[TOR:] Can we get this patch merged now that RC6 is working on drm-intel-
nightly?
Needs some review from bdw people. Also some relative residency improvement date should be added to the commit message (yes, we're allowed to do that now officially).
-Daniel
--
[TOR:] Hello bdw people, please review this patch.
Is relative performance data now required in the commit message? A week ago this would have been prohibited.
Thanks,
Tom
Quoted 64 lines · Tom
Quoted 55 lines
From: Daniel Vetter [mailto:[email protected]] On Behalf Of Daniel Vetter
Sent: Monday, June 02, 2014 1:26 AM
To: O'Rourke, Tom
Cc: [email protected]; Ben Widawsky; Kristen Carlson Accardi
Subject: Re: [Intel-gfx] [PATCH] drm/i915/bdw: Use timeout mode for RC6 on bdwQuoted 6 lines · Tom
Quoted 6 lines · Kristen Carlson Accardi
Quoted 5 lines
Quoted 2 lines · On Thu, 01 May 2014 00:03:15 +0300 Imre Deak <[email protected]> wrote:
Quoted 1 line · Ben Widawsky
On Wed, Apr 30, 2014 at 01:34:36PM -0700, Kristen Carlson Accardi
Quoted 40 lines · wrote:
Quoted 39 lines
Quoted 27 lines
Quoted 27 lines
Quoted 16 lines
Quoted 16 lines
Quoted 12 lines · On Tue, 29 Apr 2014 22:31:49 -0700 Ben Widawsky <[email protected]> wrote:
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>I've merged this one to my bdw-rc6 branch, and therefore my broadwell branch. Hopefully Kristen will see some improvement.
Quoted 6 lines
Quoted 3 lines
Unfortunately, I built your bdw-rc6 branch along with the revert I need to get my panel to work, and I get zero rc6 residency. Do I have to explicitly enable it?
I'm not actually sure. You can try it and let me know. I haven't had any time to verify the rebase. We can check my hack.
Note that in -nightly you also have to update
sanitize_rc6_option()along withintel_enable_gt_powersave()andintel_disable_gt_powersave()since atm these keep RC6 disabled on BDW.Quoted 7 lines
Quoted 2 lines
--Imre
Yes, I reverted
fb5ed3b201fe5670c9ffeec3b5f6ff044d543c9eand was able to see some rc6 residency. With the idle workload, residency appears to be similar to before, so no regression.Thanks. I'll squash this in where appropriate.
--
Ben Widawsky, Intel Open Source Technology CenterQuoted 8 lines
Quoted 1 line
[TOR:] Can we get this patch merged now that RC6 is working on drm-intel-
nightly?
Needs some review from bdw people. Also some relative residency improvement date should be added to the commit message (yes, we're allowed to do that now officially).
-Daniel
--
Quoted 4 lines
[TOR:] Hello bdw people, please review this patch.
Is relative performance data now required in the commit message? A week ago this would have been prohibited.
You might need to double-check with your own manager, but we're now again allowed to officially add it to the commit message. It was kinda always required, just had to be washed down more.
-Daniel
Signature · 3 lines
Daniel Vetter Software Engineer, Intel Corporation +41 (0) 79 365 57 48 - http://blog.ffwll.ch
Quoted 8 lines
From: Daniel Vetter [mailto:[email protected]] On Behalf Of Daniel Vetter
Sent: Tuesday, June 03, 2014 12:38 AM
To: O'Rourke, Tom
Cc: Daniel Vetter; [email protected]; Ben Widawsky; Kristen Carlson Accardi
Subject: Re: [Intel-gfx] [PATCH] drm/i915/bdw: Use timeout mode for RC6 on bdw
[TOR:]...snip
Quoted 13 lines
Quoted 13 lines
Quoted 10 lines
Quoted 3 lines
[TOR:] Can we get this patch merged now that RC6 is working on drm-intel-
nightly?
Needs some review from bdw people. Also some relative residency improvement date should be added to the commit message (yes, we're allowed to do that now officially).
-Daniel
--[TOR:] Hello bdw people, please review this patch.
Quoted 7 lines
Quoted 2 lines
Is relative performance data now required in the commit message? A week ago this would have been prohibited.
You might need to double-check with your own manager, but we're now again allowed to officially add it to the commit message. It was kinda always required, just had to be washed down more.
-Daniel
[TOR:] Sorry, I am not allowed to add the performance improvement claims to the commit message. Tom
Quoted 8 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only. The difference is particularly noticeable with video playback.
Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>
Now that CHV is out, does it apply there too?
[snip]
Signature · 1 line
Ben Widawsky, Intel Open Source Technology Center
Quoted 12 lines
From: Ben Widawsky [mailto:[email protected]]
Sent: Friday, June 20, 2014 9:43 AMQuoted 7 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only.
The difference is particularly noticeable with video playback.Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>
Quoted 8 lines
Now that CHV is out, does it apply there too?
Reviewed-by Ben Widawsky <[email protected]>[snip]
--
Ben Widawsky, Intel Open Source Technology Center
[TOR:] For CHV, we expect timeout mode will provide benefit on some pre-production steppings and no benefit on the production stepping.
Quoted 15 lines
Quoted 15 lines
From: Ben Widawsky [mailto:[email protected]]
Sent: Friday, June 20, 2014 9:43 AMQuoted 7 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only.
The difference is particularly noticeable with video playback.Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>Now that CHV is out, does it apply there too?
Reviewed-by Ben Widawsky <[email protected]>
Quoted 7 lines
Quoted 5 lines
[snip]
--
Ben Widawsky, Intel Open Source Technology Center[TOR:] For CHV, we expect timeout mode will provide benefit on some pre- production steppings and no benefit on the production stepping.
[TOR:] Can we get this patch merged now that Ben has reviewed?
Quoted 18 lines · Tom
Quoted 18 lines
Quoted 18 lines
From: Ben Widawsky [mailto:[email protected]]
Sent: Friday, June 20, 2014 9:43 AMQuoted 7 lines · Tom O'Rourke
Higher RC6 residency is observed using timeout mode instead of EI mode. This applies to Broadwell only.
The difference is particularly noticeable with video playback.Issue: VIZ-3778
Change-Id I62bb12e21caf19651034826b45cde7f73a80938dSigned-off-by Tom O'Rourke <Tom.O'[email protected]>Now that CHV is out, does it apply there too?
Reviewed-by Ben Widawsky <[email protected]>[snip]
Quoted 6 lines
Quoted 4 lines
Quoted 2 lines
--
Ben Widawsky, Intel Open Source Technology Center[TOR:] For CHV, we expect timeout mode will provide benefit on some pre- production steppings and no benefit on the production stepping.
[TOR:] Can we get this patch merged now that Ben has reviewed?
Pushed to dinq, thanks for the patch and review.
BR,
Jani.
Signature · 1 line
Jani Nikula, Intel Open Source Technology Center