Skip to contentSkip to conversation

drm/i915: Add a new modparam for customized ring multiplier

RFC In review Jackie Li 5 participants 18 messages Dec 18, 2017 – Jan 30, 2018

Subsystems (2)
lore
Share & tools
lore

Jackie Li

author
#0 lore raw

RFC 1/1 drm/i915: Add a new modparam for customized ring multiplier

From: Zhipeng Gong <[email protected]>

SKL platforms requires a higher ring multiplier when there's massive GPU load. Current driver doesn't provide a way to override the ring multiplier.

This patch adds a new module parameter to allow the overriding of ring multiplier for Gen9 platforms.

Cc Ben Widawsky <[email protected]>
Cc Chris Wilson <[email protected]>
Cc Oscar Mateo <[email protected]>
Cc Sagar Arun Kamble <[email protected]>
Cc Dmitry Rogozhkin <[email protected]>
Cc Zhipeng Gong <[email protected]>
---
 drivers/gpu/drm/i915/i915_params.c |  4 +++
 drivers/gpu/drm/i915/i915_params.h |  1 +
 drivers/gpu/drm/i915/intel_pm.c    | 57 ++++++++++++++++++++++++++++++++------
 3 files changed, 53 insertions(+), 9 deletions(-)
drivers/gpu/drm/i915/i915_params.c +4 -0
index 8dfea03..f3dd179 100644
@@ -101,6 +101,10 @@ i915_param_named_unsafe(disable_power_well, int, 0400,
 	"Disable display power wells when possible "
 	"(-1=auto [default], 0=power wells always on, 1=power wells disabled when possible)");
 
+i915_param_named_unsafe(ring_multiplier, uint, 0400,
+	"Override Ring/GT frequency multiplier."
+	"(2=2x ring multiplier [defaut], 3=3x ring multiplier)");
+
 i915_param_named_unsafe(enable_ips, int, 0600, "Enable IPS (default: true)");
 
 i915_param_named(fastboot, bool, 0600,
drivers/gpu/drm/i915/i915_params.h +1 -0
index 792ce26..6073d1c 100644
@@ -43,6 +43,7 @@
 	param(int, enable_ppgtt, -1) \
 	param(int, enable_psr, -1) \
 	param(int, disable_power_well, -1) \
+	param(unsigned int, ring_multiplier, 2) \
 	param(int, enable_ips, 1) \
 	param(int, invert_brightness, 0) \
 	param(int, enable_guc, 0) \
drivers/gpu/drm/i915/intel_pm.c +48 -9

84 lines not on this page · raw message

Signature · 6 lines
2.7.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#1 lore raw

✓ Fi.CI.BAT: success for drm/i915: Add a new modparam for customized ring multiplier

== Series Details ==

== Series Details ==

Series: drm/i915: Add a new modparam for customized ring multiplier
URL : https://patchwork.freedesktop.org/series/35535/
State : success

== Summary ==

Series 35535v1 drm/i915: Add a new modparam for customized ring multiplier
https://patchwork.freedesktop.org/api/1.0/series/35535/revisions/1/mbox/

Test debugfs_test:

        Subgroup read_all_entries:
                pass       -> INCOMPLETE (fi-snb-2520m) fdo#103713
Test kms_pipe_crc_basic:
        Subgroup suspend-read-crc-pipe-a:
                pass       -> DMESG-WARN (fi-kbl-r) fdo#104172

fdo#103713 https://bugs.freedesktop.org/show_bug.cgi?id=103713
fdo#104172 https://bugs.freedesktop.org/show_bug.cgi?id=104172

fi-bdw-5557u     total:288  pass:267  dwarn:0   dfail:0   fail:0   skip:21  time:433s
fi-bdw-gvtdvm    total:288  pass:264  dwarn:0   dfail:0   fail:0   skip:24  time:439s
fi-blb-e6850     total:288  pass:223  dwarn:1   dfail:0   fail:0   skip:64  time:377s
fi-bsw-n3050     total:288  pass:242  dwarn:0   dfail:0   fail:0   skip:46  time:498s
fi-bwr-2160      total:288  pass:183  dwarn:0   dfail:0   fail:0   skip:105 time:276s
fi-bxt-dsi       total:288  pass:258  dwarn:0   dfail:0   fail:0   skip:30  time:493s
fi-bxt-j4205     total:288  pass:259  dwarn:0   dfail:0   fail:0   skip:29  time:493s
fi-byt-j1900     total:288  pass:253  dwarn:0   dfail:0   fail:0   skip:35  time:484s
fi-byt-n2820     total:288  pass:249  dwarn:0   dfail:0   fail:0   skip:39  time:471s
fi-elk-e7500     total:224  pass:163  dwarn:15  dfail:0   fail:0   skip:45 
fi-gdg-551       total:288  pass:178  dwarn:1   dfail:0   fail:1   skip:108 time:261s
fi-glk-1         total:288  pass:260  dwarn:0   dfail:0   fail:0   skip:28  time:529s
fi-hsw-4770      total:288  pass:261  dwarn:0   dfail:0   fail:0   skip:27  time:406s
fi-hsw-4770r     total:288  pass:261  dwarn:0   dfail:0   fail:0   skip:27  time:414s
fi-ilk-650       total:288  pass:228  dwarn:0   dfail:0   fail:0   skip:60  time:387s
fi-ivb-3520m     total:288  pass:259  dwarn:0   dfail:0   fail:0   skip:29  time:475s
fi-ivb-3770      total:288  pass:255  dwarn:0   dfail:0   fail:0   skip:33  time:425s
fi-kbl-7500u     total:288  pass:263  dwarn:1   dfail:0   fail:0   skip:24  time:481s
fi-kbl-7560u     total:288  pass:268  dwarn:1   dfail:0   fail:0   skip:19  time:527s
fi-kbl-7567u     total:288  pass:268  dwarn:0   dfail:0   fail:0   skip:20  time:462s
fi-kbl-r         total:288  pass:260  dwarn:1   dfail:0   fail:0   skip:27  time:526s
fi-pnv-d510      total:288  pass:222  dwarn:1   dfail:0   fail:0   skip:65  time:575s
fi-skl-6260u     total:288  pass:268  dwarn:0   dfail:0   fail:0   skip:20  time:445s
fi-skl-6600u     total:288  pass:260  dwarn:1   dfail:0   fail:0   skip:27  time:532s
fi-skl-6700hq    total:288  pass:262  dwarn:0   dfail:0   fail:0   skip:26  time:552s
fi-skl-6700k2    total:288  pass:264  dwarn:0   dfail:0   fail:0   skip:24  time:505s
fi-skl-6770hq    total:288  pass:268  dwarn:0   dfail:0   fail:0   skip:20  time:501s
fi-skl-gvtdvm    total:288  pass:265  dwarn:0   dfail:0   fail:0   skip:23  time:446s
fi-snb-2520m     total:3    pass:2    dwarn:0   dfail:0   fail:0   skip:0  
fi-snb-2600      total:288  pass:248  dwarn:0   dfail:0   fail:0   skip:40  time:405s
Blacklisted hosts:
fi-cfl-s2        total:288  pass:262  dwarn:0   dfail:0   fail:0   skip:26  time:585s
fi-cnl-y         total:285  pass:259  dwarn:0   dfail:0   fail:0   skip:25 
fi-glk-dsi       total:54   pass:46   dwarn:0   dfail:0   fail:0   skip:7  

dcb649c7cee0af1ed444d25c2230f18fa4bad49e drm-tip: 2017y-12m-18d-16h-15m-16s UTC integration manifest fe7a55adb46a drm/i915: Add a new modparam for customized ring multiplier

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_7530/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#2 lore raw
In reply to Jackie Li

Quoting Jackie Li (2017-12-18 21:22:08)

Quoted 8 lines

From: Zhipeng Gong <[email protected]>

SKL platforms requires a higher ring multiplier when there's massive GPU load. Current driver doesn't provide a way to override the ring multiplier.

This patch adds a new module parameter to allow the overriding of ring multiplier for Gen9 platforms.

So the default ring-scaling is not good enough, the first thing we do is to try and ensure the defaults work for nearly all use cases. My impression is that you want a nonlinear scalefactor, low power workloads don't try and ramp up the ring frequencies as aggressively, high power workloads try hard for higher frequencies, and then get throttled back harder as well. How well can we autotune it? What events tells us if the ratio is too high or too low?

Adding an untested unsafe modparam is a big no. If you want us to support such a parameter, we at least need it not to taint the kernel and come with some tests that prove it functions as intended. Better still might be a I915_SETPARAM (or hooking up the write func of the modparam) to support changing the value on the fly.
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#3 lore raw
Quoted 9 lines · Chris Wilson

Quoting Jackie Li (2017-12-18 21:22:08)

Quoted 8 lines

From: Zhipeng Gong <[email protected]>

SKL platforms requires a higher ring multiplier when there's massive GPU load. Current driver doesn't provide a way to override the ring multiplier.

This patch adds a new module parameter to allow the overriding of ring multiplier for Gen9 platforms.

Quoted 7 lines

So the default ring-scaling is not good enough, the first thing we do is to try and ensure the defaults work for nearly all use cases. My impression is that you want a nonlinear scalefactor, low power workloads don't try and ramp up the ring frequencies as aggressively, high power workloads try hard for higher frequencies, and then get throttled back harder as well. How well can we autotune it? What events tells us if the ratio is too high or too low?

Thanks Chris! the background was that we found that the performance would drop on some SKL platforms with the default ring scaling factor. and use of a 3x ring scaler can solve this problem.

One of the trigger of this issue is when there's massive GPU workloads while having little/negligible CPU load - currently, we have no way to monitor GPU/CPU workload in our driver. This is the reason why we use a modparam. Do you have any suggestion about this?

Zhipeng and Dmitry may add more insights on this issue and related tuning.

Quoted 5 lines

Adding an untested unsafe modparam is a big no. If you want us to support such a parameter, we at least need it not to taint the kernel and come with some tests that prove it functions as intended. Better still might be a I915_SETPARAM (or hooking up the write func of the modparam) to support changing the value on the fly.

the intention of this patch was to add a way to override the default 2x ring scaling factor so that we would be able choose a different ring scaler for SKL platforms which were known that would have massive GPU workload. And related changes (using a 3x scaler) had been tested.

It would be great if we can do this on the fly.  However, we're going to need a way to monitor the trigger events in order to change the value dynamically.

By saying use I915_SETPARAM (or different write func), are you suggesting to request the new ring freq directly while setting the new ring scaling factor?

Regards,
-Jackie

Quoted 1 line

-Chris

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#4 lore raw
In reply to Chris Wilson

Quoting Chris Wilson (2017-12-18 21:47:25)

Quoted 10 lines

Quoting Jackie Li (2017-12-18 21:22:08)

Quoted 8 lines

From: Zhipeng Gong <[email protected]>

SKL platforms requires a higher ring multiplier when there's massive GPU load. Current driver doesn't provide a way to override the ring multiplier.

This patch adds a new module parameter to allow the overriding of ring multiplier for Gen9 platforms.

Quoted 7 lines

So the default ring-scaling is not good enough, the first thing we do is to try and ensure the defaults work for nearly all use cases. My impression is that you want a nonlinear scalefactor, low power workloads don't try and ramp up the ring frequencies as aggressively, high power workloads try hard for higher frequencies, and then get throttled back harder as well. How well can we autotune it? What events tells us if the ratio is too high or too low?

One thing that came to mind is that we don't know the min/max ring frequencies and just program them blindly. Is it the case that at max gpu freq, there is still headroom on the ring freq, or do you require a steeper ramp so that you hit the max ringfreq earlier for your workload (which then presumably can run at less than max gpufreq, so pushing power elsewhere).
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#5 lore raw

Clarification on the issue. Consider that you have a massive load on GT and just tiny one on IA. If GT will program the RING frequency to be lower than IA frequency, then you will fall into the situation when RING frequency constantly transits from GT to IA level and back. Each transition of a RING frequency is a full system stall. If you will have "good" transition rate with few transitions per few milliseconds you will lose ~10% of performance. That's the case for media workloads when you easily can step into this since 1) media utilizes few GPU engines and with few parallel workloads you can make sure that at least 1 engine is _always_ doing something, 2) media BB are relatively small, so you have regular wakeups of the IA to manage requests. This will affect Gen9 platforms due to HW design change (we've spot this in SKL). This will not happen in Gen8 (old HW design). This will be fixed in Gen10+ (CNL+).

On SKL we ran into this with the GPU frequency pinned to 700MHz, CPU to 2GHz. Multipliers were x2 for GT, x1 for IA.

So, effectively, what we need to do is to make sure that RING frequency request from GT is _not_ below the request from IA. If IA requests 2GHz, we can't request 1.4GHz, we need request at least 2GHz. Multiplier patch was intended to do exactly that, but manually. Can we somehow automate that managing IA frequency requests to the RING?

Dmitry.

-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, December 26, 2017 6:36 AM
To: Li, Yaodong <[email protected]>; [email protected]
Cc: Gong, Zhipeng <[email protected]>; Widawsky, Benjamin <[email protected]>; Mateo Lozano, Oscar <[email protected]>; Kamble, Sagar A <[email protected]>; Rogozhkin, Dmitry V <[email protected]>; Li, Yaodong <[email protected]>
Subject: Re: [RFC] drm/i915: Add a new modparam for customized ring multiplier

Quoting Chris Wilson (2017-12-18 21:47:25)

Quoted 10 lines

Quoting Jackie Li (2017-12-18 21:22:08)

Quoted 8 lines

From: Zhipeng Gong <[email protected]>

SKL platforms requires a higher ring multiplier when there's massive GPU load. Current driver doesn't provide a way to override the ring multiplier.

This patch adds a new module parameter to allow the overriding of ring multiplier for Gen9 platforms.

Quoted 7 lines

So the default ring-scaling is not good enough, the first thing we do is to try and ensure the defaults work for nearly all use cases. My impression is that you want a nonlinear scalefactor, low power workloads don't try and ramp up the ring frequencies as aggressively, high power workloads try hard for higher frequencies, and then get throttled back harder as well. How well can we autotune it? What events tells us if the ratio is too high or too low?

One thing that came to mind is that we don't know the min/max ring frequencies and just program them blindly. Is it the case that at max gpu freq, there is still headroom on the ring freq, or do you require a steeper ramp so that you hit the max ringfreq earlier for your workload (which then presumably can run at less than max gpufreq, so pushing power elsewhere). -Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#6 lore raw

Quoting Rogozhkin, Dmitry V (2017-12-26 16:39:23)

Quoted 1 line

Clarification on the issue. Consider that you have a massive load on GT and just tiny one on IA. If GT will program the RING frequency to be lower than IA frequency, then you will fall into the situation when RING frequency constantly transits from GT to IA level and back. Each transition of a RING frequency is a full system stall. If you will have "good" transition rate with few transitions per few milliseconds you will lose ~10% of performance. That's the case for media workloads when you easily can step into this since 1) media utilizes few GPU engines and with few parallel workloads you can make sure that at least 1 engine is _always_ doing something, 2) media BB are relatively small, so you have regular wakeups of the IA to manage requests. This will affect Gen9 platforms due to HW design change (we've spot this in SKL). This will not happen in Gen8 (old HW design). This will be fixed in Gen10+ (CNL+).

To clarify, the HW will flip between the two GT/IA requests rather than stick to the highest? Iirc, the expectation was that we were setting a requested minimum frequency for the ring/ia based off the gpu freq.

Quoted 1 line

On SKL we ran into this with the GPU frequency pinned to 700MHz, CPU to 2GHz. Multipliers were x2 for GT, x1 for IA.

Basically, with the GPU clocked to mid frequency, memory throughput is insufficient to keep the fixed functions occupied, and you need to increase the ring frequency. Is there ever a case where we don't need max ring frequency? (Perhaps we still need to set low frequency for GT idle?) I guess media is more susceptible to this as that workload should be sustainable at reduced clocks, GL et al are much more likely to keep the clocks ramped all the way up.

Do you know anything about the opposite position. I heard a suggestion that simply increasing the ringfreq universally caused thermal throttling in some other workloads. Do you have any knowledge of those?

Quoted 1 line

So, effectively, what we need to do is to make sure that RING frequency request from GT is _not_ below the request from IA. If IA requests 2GHz, we can't request 1.4GHz, we need request at least 2GHz. Multiplier patch was intended to do exactly that, but manually. Can we somehow automate that managing IA frequency requests to the RING?

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions? That seems possible, certainly. I'm not sure if the ring frequency is actually poked from anywhere else in the kernel, would be interesting to find out.
-Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#7 lore raw
Quoted 1 line
Quoted 1 line

To clarify, the HW will flip between the two GT/IA requests rather than stick to the highest?

Yes, it will flip on Gen9. On Gen8 there was some mechanism (HW) which flattened that. But it was removed/substituted in Gen9. In Gen10 it was tuned to close the mentioned issue.

Quoted 1 line
Quoted 1 line

Do you know anything about the opposite position. I heard a suggestion that simply increasing the ringfreq universally caused thermal throttling in some other workloads. Do you have any knowledge of those?

Initially we tried to just increase GT multiplier to x3 and stepped into the throttling. Thus, we introduced parameter to be able to mitigate all that depending on the SKU and user needs. I definitely asked what will be if GT request will be bigger than IA request. But that was a year ago and I don't remember the answer. Let me ask again. I will mail back in few days.

Quoted 1 line
Quoted 1 line

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions?

Yep. This seems a correct step to give some automatic support instead of parameter/hardcoded multiplier.

Dmitry.

-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, December 26, 2017 8:59 AM
To: Rogozhkin, Dmitry V <[email protected]>; Li, Yaodong <[email protected]>; [email protected]
Cc: Gong, Zhipeng <[email protected]>; Widawsky, Benjamin <[email protected]>; Mateo Lozano, Oscar <[email protected]>; Kamble, Sagar A <[email protected]>; Li, Yaodong <[email protected]>
Subject: RE: [RFC] drm/i915: Add a new modparam for customized ring multiplier

Quoting Rogozhkin, Dmitry V (2017-12-26 16:39:23)

Quoted 1 line

Clarification on the issue. Consider that you have a massive load on GT and just tiny one on IA. If GT will program the RING frequency to be lower than IA frequency, then you will fall into the situation when RING frequency constantly transits from GT to IA level and back. Each transition of a RING frequency is a full system stall. If you will have "good" transition rate with few transitions per few milliseconds you will lose ~10% of performance. That's the case for media workloads when you easily can step into this since 1) media utilizes few GPU engines and with few parallel workloads you can make sure that at least 1 engine is _always_ doing something, 2) media BB are relatively small, so you have regular wakeups of the IA to manage requests. This will affect Gen9 platforms due to HW design change (we've spot this in SKL). This will not happen in Gen8 (old HW design). This will be fixed in Gen10+ (CNL+).

To clarify, the HW will flip between the two GT/IA requests rather than stick to the highest? Iirc, the expectation was that we were setting a requested minimum frequency for the ring/ia based off the gpu freq.

Quoted 1 line

On SKL we ran into this with the GPU frequency pinned to 700MHz, CPU to 2GHz. Multipliers were x2 for GT, x1 for IA.

Basically, with the GPU clocked to mid frequency, memory throughput is insufficient to keep the fixed functions occupied, and you need to increase the ring frequency. Is there ever a case where we don't need max ring frequency? (Perhaps we still need to set low frequency for GT idle?) I guess media is more susceptible to this as that workload should be sustainable at reduced clocks, GL et al are much more likely to keep the clocks ramped all the way up.

Do you know anything about the opposite position. I heard a suggestion that simply increasing the ringfreq universally caused thermal throttling in some other workloads. Do you have any knowledge of those?

Quoted 1 line

So, effectively, what we need to do is to make sure that RING frequency request from GT is _not_ below the request from IA. If IA requests 2GHz, we can't request 1.4GHz, we need request at least 2GHz. Multiplier patch was intended to do exactly that, but manually. Can we somehow automate that managing IA frequency requests to the RING?

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions? That seems possible, certainly. I'm not sure if the ring frequency is actually poked from anywhere else in the kernel, would be interesting to find out. -Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#8 lore raw
Quoted 1 line
Quoted 1 line

I definitely asked what will be if GT request will be bigger than IA request. But that was a year ago and I don't remember the answer. Let me ask again. I will mail back in few days.

Hi Chris, here is the response.

Question was: "Whether we can meet with the RING transition penalty (at least theoretically) if we will have GT request higher than IA request with the dominant IA load and tiny GT load, i.e. reverted situation of what we have actually faced? For example, if we will try to pin IA frequency to 800MHz (x1 multiplier) and GT frequency to 700MHz (x2 multiplier): in that case we will have requests for ring 800 vs. 1400."

Answer is: "In this case, if the GT will toggle between RC0 and RC6, it will force ring frequency to toggle between 800 and 1400, which in the toggling time will stall IA execution. This will lead to performance loss." However, this is a case if we have really few toggle events within few milliseconds. It is quite probable that GT driver will not allow such behavior to happen if it simply doesn't often toggle between RC0 and RC6. Considering that GT driver probably handles much less interrupts than IA, this can be the case. So, I think Chris, that's now question to you: how often you toggle between RC0 and RC6 to see the reverted issue to happen? If you don't toggle much, then RING will simply remain on 1400 almost all the time and you will see no issue.

Again, I remind that's the talk about Gen9 only.

Dmitry.

-----Original Message-----
From: Intel-gfx [mailto:[email protected]] On Behalf Of Rogozhkin, Dmitry V
Sent: Tuesday, December 26, 2017 9:39 AM
To: Chris Wilson <[email protected]>; Li, Yaodong <[email protected]>; [email protected]
Cc: Widawsky, Benjamin <[email protected]>
Subject: Re: [Intel-gfx] [RFC] drm/i915: Add a new modparam for customized ring multiplier

Quoted 1 line
Quoted 1 line

To clarify, the HW will flip between the two GT/IA requests rather than stick to the highest?

Yes, it will flip on Gen9. On Gen8 there was some mechanism (HW) which flattened that. But it was removed/substituted in Gen9. In Gen10 it was tuned to close the mentioned issue.

Quoted 1 line
Quoted 1 line

Do you know anything about the opposite position. I heard a suggestion that simply increasing the ringfreq universally caused thermal throttling in some other workloads. Do you have any knowledge of those?

Initially we tried to just increase GT multiplier to x3 and stepped into the throttling. Thus, we introduced parameter to be able to mitigate all that depending on the SKU and user needs. I definitely asked what will be if GT request will be bigger than IA request. But that was a year ago and I don't remember the answer. Let me ask again. I will mail back in few days.

Quoted 1 line
Quoted 1 line

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions?

Yep. This seems a correct step to give some automatic support instead of parameter/hardcoded multiplier.

Dmitry.

-----Original Message-----
From: Chris Wilson [mailto:[email protected]]
Sent: Tuesday, December 26, 2017 8:59 AM
To: Rogozhkin, Dmitry V <[email protected]>; Li, Yaodong <[email protected]>; [email protected]
Cc: Gong, Zhipeng <[email protected]>; Widawsky, Benjamin <[email protected]>; Mateo Lozano, Oscar <[email protected]>; Kamble, Sagar A <[email protected]>; Li, Yaodong <[email protected]>
Subject: RE: [RFC] drm/i915: Add a new modparam for customized ring multiplier

Quoting Rogozhkin, Dmitry V (2017-12-26 16:39:23)

Quoted 1 line

Clarification on the issue. Consider that you have a massive load on GT and just tiny one on IA. If GT will program the RING frequency to be lower than IA frequency, then you will fall into the situation when RING frequency constantly transits from GT to IA level and back. Each transition of a RING frequency is a full system stall. If you will have "good" transition rate with few transitions per few milliseconds you will lose ~10% of performance. That's the case for media workloads when you easily can step into this since 1) media utilizes few GPU engines and with few parallel workloads you can make sure that at least 1 engine is _always_ doing something, 2) media BB are relatively small, so you have regular wakeups of the IA to manage requests. This will affect Gen9 platforms due to HW design change (we've spot this in SKL). This will not happen in Gen8 (old HW design). This will be fixed in Gen10+ (CNL+).

To clarify, the HW will flip between the two GT/IA requests rather than stick to the highest? Iirc, the expectation was that we were setting a requested minimum frequency for the ring/ia based off the gpu freq.

Quoted 1 line

On SKL we ran into this with the GPU frequency pinned to 700MHz, CPU to 2GHz. Multipliers were x2 for GT, x1 for IA.

Basically, with the GPU clocked to mid frequency, memory throughput is insufficient to keep the fixed functions occupied, and you need to increase the ring frequency. Is there ever a case where we don't need max ring frequency? (Perhaps we still need to set low frequency for GT idle?) I guess media is more susceptible to this as that workload should be sustainable at reduced clocks, GL et al are much more likely to keep the clocks ramped all the way up.

Do you know anything about the opposite position. I heard a suggestion that simply increasing the ringfreq universally caused thermal throttling in some other workloads. Do you have any knowledge of those?

Quoted 1 line

So, effectively, what we need to do is to make sure that RING frequency request from GT is _not_ below the request from IA. If IA requests 2GHz, we can't request 1.4GHz, we need request at least 2GHz. Multiplier patch was intended to do exactly that, but manually. Can we somehow automate that managing IA frequency requests to the RING?

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions? That seems possible, certainly. I'm not sure if the ring frequency is actually poked from anywhere else in the kernel, would be interesting to find out. -Chris
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#9 lore raw
In reply to Rogozhkin, Dmitry V
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions?

Yep. This seems a correct step to give some automatic support instead of parameter/hardcoded multiplier.

Does this mean we should use cpufreq/intel_pstate based approach instead of the current modparam solution for Gen9?

Some concerns and questions about intel_pstate approach:
a) Currently, we cannot get the accurate pstate/target freq value from cpufreq in intel_pstate active mode since these values won't be exported to cpufreq layer, so if we won't change intel_pstate code then we only can get the max cpu freq of a new policy.
b) intel_pstate policy is attached to each logic cpu, which means we will receive policy/freq transition notification for each logic cpu freq change. One question is how we are going to decide the freq of the ring? just use the max cpu freq reported?
c) With the intel_pstate approach we may still run into thermal throttling, in this case, can a certain cooling device be triggered to lower the cpu freq?

Thanks and Regards,
-Jackie

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#10 lore raw

Since ring frequency programming needs consideration of both IA and GT frequency requests I think keeping the logic to program the ring frequency table in driver that monitors both IA/GT busyness and power budgets like intel_ips will be more appropriate. intel_ips is relying on global load derived from all CPUs.
I understand that power awareness and busyness based policy might be trickier but having that as tunable will give better flexibility.

Quoted 16 lines · Yaodong Li
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions?

Yep. This seems a correct step to give some automatic support instead of parameter/hardcoded multiplier.

Does this mean we should use cpufreq/intel_pstate based approach instead of the current modparam solution for Gen9?

Some concerns and questions about intel_pstate approach:
a) Currently, we cannot get the accurate pstate/target freq value from cpufreq in intel_pstate active mode since these values won't be exported to cpufreq layer, so if we won't change intel_pstate code then we only can get the max cpu freq of a new policy.
b) intel_pstate policy is attached to each logic cpu, which means we will receive policy/freq transition notification for each logic cpu freq change. One question is how we are going to decide the freq of the ring? just use the max cpu freq reported?
c) With the intel_pstate approach we may still run into thermal throttling, in this case, can a certain cooling device be triggered to lower the cpu freq?

Quoted 3 lines

Thanks and Regards,
-Jackie

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#11 lore raw
Quoted 5 lines · Sagar Arun Kamble

Since ring frequency programming needs consideration of both IA and GT frequency requests I think keeping the logic to program the ring frequency table in driver that monitors both IA/GT busyness and power budgets like intel_ips will be more appropriate. intel_ips is relying on global load derived from all CPUs.
I understand that power awareness and busyness based policy might be trickier but having that as tunable will give better flexibility.

By just looking into the current code, the way intel_ips checks gpu busyness cannot reflect the actual workload of GT (e.g. gpu busy is true even if there's only one pending request), in this case, we shall not increase the ring freq if we want to use a "workload monitoring" based solution. so we need a more accurate way to monitor the current GT workload (e.g.  when the pending request count reaches a center tunable threshold??).

Quoted 17 lines · Yaodong Li
Quoted 16 lines
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions?

Yep. This seems a correct step to give some automatic support instead of parameter/hardcoded multiplier.

Does this mean we should use cpufreq/intel_pstate based approach instead of the current modparam solution for Gen9?

Some concerns and questions about intel_pstate approach:
a) Currently, we cannot get the accurate pstate/target freq value from cpufreq in intel_pstate active mode since these values won't be exported to cpufreq layer, so if we won't change intel_pstate code then we only can get the max cpu freq of a new policy.
b) intel_pstate policy is attached to each logic cpu, which means we will receive policy/freq transition notification for each logic cpu freq change. One question is how we are going to decide the freq of the ring? just use the max cpu freq reported?
c) With the intel_pstate approach we may still run into thermal throttling, in this case, can a certain cooling device be triggered to lower the cpu freq?

Quoted 4 lines
Quoted 3 lines

Thanks and Regards,
-Jackie

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#12 lore raw
Quoted 11 lines · Yaodong Li
Quoted 5 lines · Sagar Arun Kamble

Since ring frequency programming needs consideration of both IA and GT frequency requests I think keeping the logic to program the ring frequency table in driver that monitors both IA/GT busyness and power budgets like intel_ips will be more appropriate. intel_ips is relying on global load derived from all CPUs.
I understand that power awareness and busyness based policy might be trickier but having that as tunable will give better flexibility.

By just looking into the current code, the way intel_ips checks gpu busyness cannot reflect the actual workload of GT (e.g. gpu busy is true even if there's only one pending request), in this case, we shall not increase the ring freq if we want to use a "workload monitoring" based solution. so we need a more accurate way to monitor the current GT workload (e.g.  when the pending request count reaches a center tunable threshold??).

Yes. May be we can share the PMU data about engine busyness with intel_ips.

Quoted 18 lines
Quoted 18 lines · Yaodong Li
Quoted 16 lines
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions?

Yep. This seems a correct step to give some automatic support instead of parameter/hardcoded multiplier.

Does this mean we should use cpufreq/intel_pstate based approach instead of the current modparam solution for Gen9?

Some concerns and questions about intel_pstate approach:
a) Currently, we cannot get the accurate pstate/target freq value from cpufreq in intel_pstate active mode since these values won't be exported to cpufreq layer, so if we won't change intel_pstate code then we only can get the max cpu freq of a new policy.
b) intel_pstate policy is attached to each logic cpu, which means we will receive policy/freq transition notification for each logic cpu freq change. One question is how we are going to decide the freq of the ring? just use the max cpu freq reported?
c) With the intel_pstate approach we may still run into thermal throttling, in this case, can a certain cooling device be triggered to lower the cpu freq?

Quoted 5 lines
Quoted 4 lines
Quoted 3 lines

Thanks and Regards,
-Jackie

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#13 lore raw
Quoted 10 lines · Sagar Arun Kamble
Quoted 7 lines · Yaodong Li
Quoted 5 lines · Sagar Arun Kamble

Since ring frequency programming needs consideration of both IA and GT frequency requests I think keeping the logic to program the ring frequency table in driver that monitors both IA/GT busyness and power budgets like intel_ips will be more appropriate. intel_ips is relying on global load derived from all CPUs.
I understand that power awareness and busyness based policy might be trickier but having that as tunable will give better flexibility.

Quoted 5 lines
Quoted 4 lines

By just looking into the current code, the way intel_ips checks gpu busyness cannot reflect the actual workload of GT (e.g. gpu busy is true even if there's only one pending request), in this case, we shall not increase the ring freq if we want to use a "workload monitoring" based solution. so we need a more accurate way to monitor the current GT workload (e.g.  when the pending request count reaches a center tunable threshold??).

Yes. May be we can share the PMU data about engine busyness with intel_ips.

Thank you Sagar! Can you tell more about how we can get the gpu busyness from PMU data?

I think the solution would be to set the ring freq table to use a ring freq >= current ia freq (for all possible gpu freq) once we found gpu workload is high (need to tune the threshold), and we will decrease the ring freq (use a 2x multiplier?) once we found the GT workload is low. The benefit to use the intel_ips is it can tell both the cpu & gpu busyness. However, we do need an accurate way to check at least the busyness of gpu for this issue.

Quoted 17 lines
Quoted 17 lines
Quoted 17 lines · Yaodong Li
Quoted 16 lines
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions?

Yep. This seems a correct step to give some automatic support instead of parameter/hardcoded multiplier.

Does this mean we should use cpufreq/intel_pstate based approach instead of the current modparam solution for Gen9?

Some concerns and questions about intel_pstate approach:
a) Currently, we cannot get the accurate pstate/target freq value from cpufreq in intel_pstate active mode since these values won't be exported to cpufreq layer, so if we won't change intel_pstate code then we only can get the max cpu freq of a new policy.
b) intel_pstate policy is attached to each logic cpu, which means we will receive policy/freq transition notification for each logic cpu freq change. One question is how we are going to decide the freq of the ring? just use the max cpu freq reported?
c) With the intel_pstate approach we may still run into thermal throttling, in this case, can a certain cooling device be triggered to lower the cpu freq?

Quoted 6 lines
Quoted 5 lines
Quoted 4 lines
Quoted 3 lines

Thanks and Regards,
-Jackie

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#14 lore raw
Quoted 11 lines · Yaodong Li
Quoted 10 lines · Sagar Arun Kamble
Quoted 7 lines · Yaodong Li
Quoted 5 lines · Sagar Arun Kamble

Since ring frequency programming needs consideration of both IA and GT frequency requests I think keeping the logic to program the ring frequency table in driver that monitors both IA/GT busyness and power budgets like intel_ips will be more appropriate. intel_ips is relying on global load derived from all CPUs.
I understand that power awareness and busyness based policy might be trickier but having that as tunable will give better flexibility.

Quoted 7 lines
Quoted 5 lines
Quoted 4 lines

By just looking into the current code, the way intel_ips checks gpu busyness cannot reflect the actual workload of GT (e.g. gpu busy is true even if there's only one pending request), in this case, we shall not increase the ring freq if we want to use a "workload monitoring" based solution. so we need a more accurate way to monitor the current GT workload (e.g.  when the pending request count reaches a center tunable threshold??).

Yes. May be we can share the PMU data about engine busyness with intel_ips.

Thank you Sagar! Can you tell more about how we can get the gpu busyness from PMU data?

It seems to be not possible as of now since pmu is reporting accumulated busy time per engine without info about time they became active/idle, unless I am missing something. Also having pmu ON in release environment needs to be checked. Chris, could you please clarify on pmu usage here.

Also thinking about below two options to get the GPU busyness:
1. Sharing i915 rps power zone (LOW_POWER, BETWEEN, HIGH_POWER) to intel_ips (kind of indicators of GPU load)
2. Sampling GT C0 counter to derive full GPU busyness. This though may not have been tested much so far across platforms.

Quoted 4 lines

I think the solution would be to set the ring freq table to use a ring freq >= current ia freq (for all possible gpu freq) once we found gpu workload is high (need to tune the threshold), and we will decrease the ring freq (use a 2x multiplier?) once we found the GT workload is low. The benefit to use the intel_ips is it can tell both the cpu & gpu busyness. However, we do need an accurate way to check at least the busyness of gpu for this issue.

Agree.

Thanks
Sagar

Quoted 18 lines
Quoted 18 lines
Quoted 18 lines
Quoted 18 lines · Yaodong Li
Quoted 16 lines
Quoted 4 lines
Quoted 2 lines
Quoted 2 lines

You are thinking of plugging into intel_pstate to make it smarter for ia freq transitions?

Yep. This seems a correct step to give some automatic support instead of parameter/hardcoded multiplier.

Does this mean we should use cpufreq/intel_pstate based approach instead of the current modparam solution for Gen9?

Some concerns and questions about intel_pstate approach:
a) Currently, we cannot get the accurate pstate/target freq value from cpufreq in intel_pstate active mode since these values won't be exported to cpufreq layer, so if we won't change intel_pstate code then we only can get the max cpu freq of a new policy.
b) intel_pstate policy is attached to each logic cpu, which means we will receive policy/freq transition notification for each logic cpu freq change. One question is how we are going to decide the freq of the ring? just use the max cpu freq reported?
c) With the intel_pstate approach we may still run into thermal throttling, in this case, can a certain cooling device be triggered to lower the cpu freq?

Quoted 7 lines
Quoted 6 lines
Quoted 5 lines
Quoted 4 lines
Quoted 3 lines

Thanks and Regards,
-Jackie

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

author
#15 lore raw

RFC 1/1 drm/i915: cpufreq based dynamic ring freqency table on Gen9

In reply to Yaodong Li

A known issue that ring frequency transition would lead to a full system stall had impacted the media performance when the GT frequency is lower than IA freqency and there's heavy GT workload while having low IA workload. In this case, the ring frequency will be toggled between GT and IA frequency. The solution is to keep the ring frequency higher than or equal to current IA frequency so that we can avoid the toggling between different frequencies issue.

This patch monitors the cpufreq frequency notifications (freqency transition notification when governors are used, or new frequency policy when bypassing the governors such as the active mode of intel_pstate) and updates the ring frequency table by calculating the ring frequency for each possible gpu frequency with a new formula ring_freq = max(current IA freq, 2 * gpu_freq). with this we can guarantee that the ring frequency would depend on IA freq when the 2x gpu freq is lower then the active IA freq, and it would depend on 2x gpu freq once the 2x gpu freq is higher than current IA freq.

Potential issues with this solution:
0) currently, this patch is now only monitoring the frequency change of CPU 0. it would fail if another logic CPU was using a higher frequency.
1) the latency of the ring freq table update was not considered. Need more understanding to the cost of each ring freq table update.
2) Not sure the impact for low GT workload while the ring frequency is still set to higher or equal to IA frequency.

Cc Chris Wilson <[email protected]>
Cc Sagar Arun Kamble <[email protected]>
Cc Dmitry Rogozhkin <[email protected]>
---
 drivers/gpu/drm/i915/i915_drv.h |   4 ++
 drivers/gpu/drm/i915/intel_pm.c | 106 ++++++++++++++++++++++++++++++++++++++--
 2 files changed, 107 insertions(+), 3 deletions(-)
drivers/gpu/drm/i915/i915_drv.h +4 -0
index a689396..39311cd 100644
@@ -2044,6 +2044,10 @@ struct drm_i915_private {
 	/* gen6+ GT PM state */
 	struct intel_gen6_power_mgmt gt_pm;
 
+	/* cpufreq notifiers*/
+	struct notifier_block cpufreq_transition;
+	struct notifier_block cpufreq_policy;
+
 	/* ilk-only ips/rps state. Everything in here is protected by the global
 	 * mchdev_lock in intel_pm.c */
 	struct intel_ilk_power_mgmt ips;
drivers/gpu/drm/i915/intel_pm.c +103 -3

130 lines not on this page · raw message

Signature · 6 lines
2.7.4

_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx
#16 lore raw

✗ Fi.CI.IGT: failure for drm/i915: Add a new modparam for customized ring multiplier

In reply to Jackie Li
== Series Details ==

== Series Details ==

Series: drm/i915: Add a new modparam for customized ring multiplier
URL : https://patchwork.freedesktop.org/series/35535/
State : failure

== Summary ==

Test kms_fbcon_fbt:

        Subgroup fbc-suspend:
                pass       -> SKIP       (shard-hsw)
Test kms_frontbuffer_tracking:
        Subgroup fbc-1p-offscren-pri-shrfb-draw-render:
                fail       -> PASS       (shard-snb) fdo#101623
Test drv_selftest:
        Subgroup live_hangcheck:
                incomplete -> PASS       (shard-snb) fdo#104313
Test kms_cursor_crc:
        Subgroup cursor-128x128-suspend:
                pass       -> SKIP       (shard-snb) fdo#103375
Test kms_plane:
        Subgroup plane-position-covered-pipe-c-planes:
                pass       -> INCOMPLETE (shard-hsw)
Test drv_suspend:
        Subgroup debugfs-reader:
                skip       -> PASS       (shard-hsw)

fdo#101623 https://bugs.freedesktop.org/show_bug.cgi?id=101623
fdo#104313 https://bugs.freedesktop.org/show_bug.cgi?id=104313
fdo#103375 https://bugs.freedesktop.org/show_bug.cgi?id=103375

shard-hsw        total:2637 pass:1490 dwarn:1   dfail:0   fail:10  skip:1135 time:9281s
shard-snb        total:2712 pass:1307 dwarn:1   dfail:0   fail:12  skip:1392 time:7951s
Blacklisted hosts:
shard-apl        total:2712 pass:1686 dwarn:2   dfail:1   fail:22  skip:1001 time:13786s
shard-kbl        total:2699 pass:1793 dwarn:1   dfail:1   fail:25  skip:878 time:10894s

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_7530/shards.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx

#17 lore raw

✗ Fi.CI.BAT: failure for drm/i915: Add a new modparam for customized ring multiplier (rev2)

In reply to Jackie Li
== Series Details ==

== Series Details ==

Series: drm/i915: Add a new modparam for customized ring multiplier (rev2)
URL : https://patchwork.freedesktop.org/series/35535/
State : failure

== Summary ==

Series 35535v2 drm/i915: Add a new modparam for customized ring multiplier
https://patchwork.freedesktop.org/api/1.0/series/35535/revisions/2/mbox/

Test debugfs_test:

        Subgroup read_all_entries:
                dmesg-warn -> PASS       (fi-elk-e7500) fdo#103989
Test gem_ctx_switch:
        Subgroup basic-default-heavy:
                pass       -> DMESG-WARN (fi-skl-6260u)
                pass       -> DMESG-WARN (fi-kbl-7500u)
                pass       -> DMESG-WARN (fi-kbl-r)
Test gem_exec_basic:
        Subgroup gtt-blt:
                pass       -> DMESG-WARN (fi-skl-6770hq)
Test gem_exec_create:
        Subgroup basic:
                pass       -> DMESG-WARN (fi-kbl-7560u)
Test gem_exec_suspend:
        Subgroup basic-s4-devices:
                pass       -> INCOMPLETE (fi-skl-6260u)
                pass       -> INCOMPLETE (fi-skl-6600u)
                pass       -> INCOMPLETE (fi-skl-6700hq)
                pass       -> INCOMPLETE (fi-skl-6700k2)
                pass       -> INCOMPLETE (fi-skl-6770hq)
                pass       -> INCOMPLETE (fi-skl-guc)
                pass       -> INCOMPLETE (fi-skl-gvtdvm)
                pass       -> INCOMPLETE (fi-kbl-7500u)
                pass       -> INCOMPLETE (fi-kbl-7560u)
                pass       -> INCOMPLETE (fi-kbl-7567u)
                pass       -> INCOMPLETE (fi-kbl-r)
                pass       -> INCOMPLETE (fi-cfl-s2)

fdo#103989 https://bugs.freedesktop.org/show_bug.cgi?id=103989

fi-bdw-5557u     total:288  pass:267  dwarn:0   dfail:0   fail:0   skip:21  time:419s
fi-bdw-gvtdvm    total:288  pass:264  dwarn:0   dfail:0   fail:0   skip:24  time:429s
fi-blb-e6850     total:288  pass:223  dwarn:1   dfail:0   fail:0   skip:64  time:373s
fi-bsw-n3050     total:288  pass:242  dwarn:0   dfail:0   fail:0   skip:46  time:479s
fi-bwr-2160      total:288  pass:183  dwarn:0   dfail:0   fail:0   skip:105 time:281s
fi-bxt-dsi       total:288  pass:258  dwarn:0   dfail:0   fail:0   skip:30  time:487s
fi-bxt-j4205     total:288  pass:259  dwarn:0   dfail:0   fail:0   skip:29  time:482s
fi-byt-j1900     total:288  pass:253  dwarn:0   dfail:0   fail:0   skip:35  time:468s
fi-byt-n2820     total:288  pass:249  dwarn:0   dfail:0   fail:0   skip:39  time:451s
fi-cfl-s2        total:109  pass:97   dwarn:0   dfail:0   fail:0   skip:11 
fi-elk-e7500     total:224  pass:169  dwarn:9   dfail:0   fail:0   skip:45 
fi-gdg-551       total:288  pass:179  dwarn:0   dfail:0   fail:1   skip:108 time:276s
fi-glk-1         total:288  pass:260  dwarn:0   dfail:0   fail:0   skip:28  time:512s
fi-hsw-4770      total:288  pass:261  dwarn:0   dfail:0   fail:0   skip:27  time:390s
fi-hsw-4770r     total:288  pass:261  dwarn:0   dfail:0   fail:0   skip:27  time:399s
fi-ilk-650       total:288  pass:228  dwarn:0   dfail:0   fail:0   skip:60  time:412s
fi-ivb-3520m     total:288  pass:259  dwarn:0   dfail:0   fail:0   skip:29  time:459s
fi-ivb-3770      total:288  pass:255  dwarn:0   dfail:0   fail:0   skip:33  time:416s
fi-kbl-7500u     total:109  pass:96   dwarn:1   dfail:0   fail:0   skip:11 
fi-kbl-7560u     total:109  pass:104  dwarn:1   dfail:0   fail:0   skip:3  
fi-kbl-7567u     total:109  pass:105  dwarn:0   dfail:0   fail:0   skip:3  
fi-kbl-r         total:109  pass:96   dwarn:1   dfail:0   fail:0   skip:11 
fi-skl-6260u     total:109  pass:104  dwarn:1   dfail:0   fail:0   skip:3  
fi-skl-6600u     total:109  pass:97   dwarn:0   dfail:0   fail:0   skip:11 
fi-skl-6700hq    total:109  pass:97   dwarn:0   dfail:0   fail:0   skip:11 
fi-skl-6700k2    total:109  pass:97   dwarn:0   dfail:0   fail:0   skip:11 
fi-skl-6770hq    total:109  pass:104  dwarn:1   dfail:0   fail:0   skip:3  
fi-skl-guc       total:109  pass:97   dwarn:0   dfail:0   fail:0   skip:11 
fi-skl-gvtdvm    total:109  pass:104  dwarn:0   dfail:0   fail:0   skip:4  
fi-snb-2520m     total:288  pass:248  dwarn:0   dfail:0   fail:0   skip:40  time:527s
fi-snb-2600      total:288  pass:248  dwarn:0   dfail:0   fail:0   skip:40  time:403s

Blacklisted hosts:
fi-glk-dsi total:288 pass:258 dwarn:0 dfail:0 fail:0 skip:30 time:472s

0edf4cc4e7c163fa61312f1708fc92284b855f8e drm-tip: 2018y-01m-30d-18h-33m-48s UTC integration manifest 5422cb80b935 drm/i915: cpufreq based dynamic ring freqency table on Gen9

== Logs ==

For more details see: https://intel-gfx-ci.01.org/tree/drm-tip/Patchwork_7832/issues.html
_______________________________________________
Intel-gfx mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/intel-gfx