Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
Le 29/04/2011 02:56, Nigel Cunningham a écrit : Hi. On 29/04/11 04:35, Martin Peres wrote: Le 28/04/2011 20:29, Maxim Levitsky a écrit : On Thu, 2011-04-28 at 20:24 +0200, Martin Peres wrote: Le 28/04/2011 18:58, Maxim Levitsky a écrit : Interesting fact is that GPU temperatures rise to very high levels (~75C) even while doing CPU only work (like compiling kernel for example). Let me guess, you're on a laptop? The temperature of the case goes up when the processor is working and so, it cools the GPU less ;) Yes, but that doesn't happen while using nvidia to this extent. I think that GPU has many units running in endless loop doing nothing but consuming power (~5W) difference! I really need to have a look at this. I have a power meter now (acpi reports me funky power consumption figures because my battery is dead). Just been reading through your posts from last night, and wanted to let you know I'm interested and willing to test patches too. I have an 8600GTS based laptop, and have seen the same range of power usage (ie have seen 14W in the page - presumably when I was using the blob -, but can only get it down to 23W at the moment using Nouveau). Regards, Nigel 9W! Well, you can try downclocking the card, but this won't get you down to 14W. When fan management is done, I'll have a look at what the blob does and try to find some magic there. Martin ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
Hi. On 29/04/11 16:54, Martin Peres wrote: Le 29/04/2011 02:56, Nigel Cunningham a écrit : Hi. On 29/04/11 04:35, Martin Peres wrote: Le 28/04/2011 20:29, Maxim Levitsky a écrit : On Thu, 2011-04-28 at 20:24 +0200, Martin Peres wrote: Le 28/04/2011 18:58, Maxim Levitsky a écrit : Interesting fact is that GPU temperatures rise to very high levels (~75C) even while doing CPU only work (like compiling kernel for example). Let me guess, you're on a laptop? The temperature of the case goes up when the processor is working and so, it cools the GPU less ;) Yes, but that doesn't happen while using nvidia to this extent. I think that GPU has many units running in endless loop doing nothing but consuming power (~5W) difference! I really need to have a look at this. I have a power meter now (acpi reports me funky power consumption figures because my battery is dead). Just been reading through your posts from last night, and wanted to let you know I'm interested and willing to test patches too. I have an 8600GTS based laptop, and have seen the same range of power usage (ie have seen 14W in the page - presumably when I was using the blob -, but can only get it down to 23W at the moment using Nouveau). Regards, Nigel 9W! Well, you can try downclocking the card, but this won't get you down to 14W. When fan management is done, I'll have a look at what the blob does and try to find some magic there. Yeah - I find 9W a bit unbelievable too. I'll try to find some time to give the NVidia driver a run again, but I'm a chronic over-committer, so won't promise I'll do anything real soon now! Nigel -- Evolution (n): A hypothetical process whereby improbable events occur with alarming frequency, order arises from chaos, and no one is given credit. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
On Thu, 2011-04-28 at 01:58 +0200, Martin Peres wrote: Hi everyone, I would like everyone to test this set of patch as we'll need them quite soon for timing management on nv50. Please report success/failure by answering to this email. Thanks in advance, Martin I tested this. Patch seems to work except following problems: Sometimes reclocking fails like this: root@maxim-laptop:/home/maxim# echo 2 /sys/class/drm/card0/device/performance_level bash: echo: write error: Resource temporarily unavailable [ 630.371117] [drm] nouveau :01:00.0: setting performance level: performance_level_2 [ 630.412048] [drm] nouveau :01:00.0: PFIFO DMA_PUSH never depleted (0xd06091) Once it even hang the GPU. Easy to reproduce while running compiz and its benchmark overlay. Also, it seems not to restore perf level after resume from ram. Best regards Maxim Levitsky ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau -- Best regards, Maxim Levitsky Visit my blog: http://maximlevitsky.wordpress.com Warning: Above blog contains rants. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
Le 28/04/2011 12:32, Maxim Levitsky a écrit : On Thu, 2011-04-28 at 01:58 +0200, Martin Peres wrote: Hi everyone, I would like everyone to test this set of patch as we'll need them quite soon for timing management on nv50. Please report success/failure by answering to this email. Thanks in advance, Martin I tested this. Patch seems to work except following problems: Sometimes reclocking fails like this: root@maxim-laptop:/home/maxim# echo 2 /sys/class/drm/card0/device/performance_level bash: echo: write error: Resource temporarily unavailable This is expected. Unfortunately, I can't do anything better than this for the upcoming months. I need help from the people really into the command submission system. [ 630.371117] [drm] nouveau :01:00.0: setting performance level: performance_level_2 [ 630.412048] [drm] nouveau :01:00.0: PFIFO DMA_PUSH never depleted (0xd06091) I consider it not safe to reclock core if PFIFO DMA_PUSH isn't empty. Hence the reason why I abort there. Once it even hang the GPU. This shouldn't happen but I managed to hang it too once. Anyway, this isn't perfect but still way better than what is currently implemented. Easy to reproduce while running compiz and its benchmark overlay. Exactly, when playing open arena, the success ratio is very very low but I don't mind yet. I first want stability and then reliability. Also, it seems not to restore perf level after resume from ram. Good catch, I thought I had fixed that but it doesn't seem like. I'll fix this tonight and send an update. Thanks a lot. Best regards Maxim Levitsky Thanks a lot for testing it Maxim. Martin ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
On Thu, 2011-04-28 at 13:15 +0200, Martin Peres wrote: Le 28/04/2011 12:32, Maxim Levitsky a écrit : On Thu, 2011-04-28 at 01:58 +0200, Martin Peres wrote: Hi everyone, I would like everyone to test this set of patch as we'll need them quite soon for timing management on nv50. Please report success/failure by answering to this email. Thanks in advance, Martin I tested this. Patch seems to work except following problems: Sometimes reclocking fails like this: root@maxim-laptop:/home/maxim# echo 2 /sys/class/drm/card0/device/performance_level bash: echo: write error: Resource temporarily unavailable This is expected. Unfortunately, I can't do anything better than this for the upcoming months. I need help from the people really into the command submission system. [ 630.371117] [drm] nouveau :01:00.0: setting performance level: performance_level_2 [ 630.412048] [drm] nouveau :01:00.0: PFIFO DMA_PUSH never depleted (0xd06091) I consider it not safe to reclock core if PFIFO DMA_PUSH isn't empty. Hence the reason why I abort there. Once it even hang the GPU. This shouldn't happen but I managed to hang it too once. Anyway, this isn't perfect but still way better than what is currently implemented. Easy to reproduce while running compiz and its benchmark overlay. Exactly, when playing open arena, the success ratio is very very low but I don't mind yet. I first want stability and then reliability. Also, it seems not to restore perf level after resume from ram. Good catch, I thought I had fixed that but it doesn't seem like. I'll fix this tonight and send an update. Thanks a lot. Best regards Maxim Levitsky Thanks a lot for testing it Maxim. Martin Martin, one more thing, this is my observations regarding clocks I finished today: clock = (ref * N / M) (P 0x7) first line is blob, second nouveau level0: P NNMM 0x4008 - memory - 0018e200 2505 = 1 = 01. 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8001 0b01 = 11 / 1 / 2 = 05.5000 8002 1b02 = 27 / 2 / 4 = 03.3750 0x4028 - core - a012 1b04 = 27 / 4 / 4 = 01.6875 a002 1b04 = 27 / 4 / 4 = 01.6875 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8002 1b04 = 27 / 4 / 4 = 01.6875 level1: 0x4008 - memory - 809ae400 2505 = 37 / 5 / 4 = 01.8500 80026400 2505 = 37 / 5 / 4 = 01.8500 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8001 0b01 = 11 / 1 / 2 = 05.5000 0x4028 - core - a009 0b02 = 11 / 2 / 1 = 05.5000 a001 0b02 = 11 / 2 / 1 = 05.5000 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 0b02 = 11 / 2 / 2 = 05.5000 level2: 0x4008 - memory - 8059e400 2505 = 37 / 5 / 2 = 03.7000 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8000 1002 = 16 / 2 / 1 = 08. 0x4028 - core - a009 1002 = 16 / 2 / 2 = 04. a001 1002 = 16 / 2 / 2 = 04. 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 1202 = 18 / 2 / 2 = 06. Clocks are very different and we seems to overclock hard the unk_05 PLL And shader PLL is underclocked always, so another way to boost performance? (it didn't seem to affect perfomace much here, but yet...) -- Best regards, Maxim Levitsky Visit my blog: http://maximlevitsky.wordpress.com Warning: Above blog contains rants. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
Le 28/04/2011 13:43, Maxim Levitsky a écrit : Martin, one more thing, this is my observations regarding clocks I finished today: clock = (ref * N / M) (P 0x7) first line is blob, second nouveau level0: P NNMM 0x4008 - memory - 0018e200 2505 = 1 = 01. 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8001 0b01 = 11 / 1 / 2 = 05.5000 8002 1b02 = 27 / 2 / 4 = 03.3750 0x4028 - core - a012 1b04 = 27 / 4 / 4 = 01.6875 a002 1b04 = 27 / 4 / 4 = 01.6875 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8002 1b04 = 27 / 4 / 4 = 01.6875 level1: 0x4008 - memory - 809ae400 2505 = 37 / 5 / 4 = 01.8500 80026400 2505 = 37 / 5 / 4 = 01.8500 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8001 0b01 = 11 / 1 / 2 = 05.5000 0x4028 - core - a009 0b02 = 11 / 2 / 1 = 05.5000 a001 0b02 = 11 / 2 / 1 = 05.5000 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 0b02 = 11 / 2 / 2 = 05.5000 level2: 0x4008 - memory - 8059e400 2505 = 37 / 5 / 2 = 03.7000 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8000 1002 = 16 / 2 / 1 = 08. 0x4028 - core - a009 1002 = 16 / 2 / 2 = 04. a001 1002 = 16 / 2 / 2 = 04. 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 1202 = 18 / 2 / 2 = 06. Clocks are very different and we seems to overclock hard the unk_05 PLL And shader PLL is underclocked always, so another way to boost performance? (it didn't seem to affect perfomace much here, but yet...) Good work Maxim! Can you please tell us what card you use? For what range of card is this true? I can give you access to the vbios repo so as you can contact some people for further testing if you want to :) Anyway, I hope you are interested in looking deeper into this. Xexaxo and darktama did all the work on this. I have no time to put on this issue in the upcoming weeks as I'm already working on getting the PMS patches upstream, fan management and Fermi PM. Thanks :) Martin ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
On Thu, 2011-04-28 at 14:11 +0200, Martin Peres wrote: Le 28/04/2011 13:43, Maxim Levitsky a écrit : Martin, one more thing, this is my observations regarding clocks I finished today: clock = (ref * N / M) (P 0x7) first line is blob, second nouveau level0: P NNMM 0x4008 - memory - 0018e200 2505 = 1 = 01. 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8001 0b01 = 11 / 1 / 2 = 05.5000 8002 1b02 = 27 / 2 / 4 = 03.3750 0x4028 - core - a012 1b04 = 27 / 4 / 4 = 01.6875 a002 1b04 = 27 / 4 / 4 = 01.6875 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8002 1b04 = 27 / 4 / 4 = 01.6875 level1: 0x4008 - memory - 809ae400 2505 = 37 / 5 / 4 = 01.8500 80026400 2505 = 37 / 5 / 4 = 01.8500 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8001 0b01 = 11 / 1 / 2 = 05.5000 0x4028 - core - a009 0b02 = 11 / 2 / 1 = 05.5000 a001 0b02 = 11 / 2 / 1 = 05.5000 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 0b02 = 11 / 2 / 2 = 05.5000 level2: 0x4008 - memory - 8059e400 2505 = 37 / 5 / 2 = 03.7000 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8000 1002 = 16 / 2 / 1 = 08. 0x4028 - core - a009 1002 = 16 / 2 / 2 = 04. a001 1002 = 16 / 2 / 2 = 04. 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 1202 = 18 / 2 / 2 = 06. Clocks are very different and we seems to overclock hard the unk_05 PLL And shader PLL is underclocked always, so another way to boost performance? (it didn't seem to affect perfomace much here, but yet...) Good work Maxim! Can you please tell us what card you use? For what range of card is this true? I can give you access to the vbios repo so as you can contact some people for further testing if you want to :) I use NV86 Geforce 8400M GS card. Don't have much time though. Anyway, I hope you are interested in looking deeper into this. Xexaxo and darktama did all the work on this. I have no time to put on this issue in the upcoming weeks as I'm already working on getting the PMS patches upstream, fan management and Fermi PM. Thanks :) Martin No problem! Best regards, Maxim Levitsky -- Best regards, Maxim Levitsky Visit my blog: http://maximlevitsky.wordpress.com Warning: Above blog contains rants. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
Le 28/04/2011 14:32, Maxim Levitsky a écrit : On Thu, 2011-04-28 at 14:11 +0200, Martin Peres wrote: Good work Maxim! Can you please tell us what card you use? For what range of card is this true? I can give you access to the vbios repo so as you can contact some people for further testing if you want to :) I use NV86 Geforce 8400M GS card. Don't have much time though. Ok, cool. I got a 8600 GS from my father, I'll see if I can reproduce this issue. Too bad for your time, I'll see what I can do though. This could very well explain some instability I've been encountering. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
On Thu, 2011-04-28 at 14:35 +0200, Martin Peres wrote: Le 28/04/2011 14:32, Maxim Levitsky a écrit : On Thu, 2011-04-28 at 14:11 +0200, Martin Peres wrote: Good work Maxim! Can you please tell us what card you use? For what range of card is this true? I can give you access to the vbios repo so as you can contact some people for further testing if you want to :) I use NV86 Geforce 8400M GS card. Don't have much time though. Ok, cool. I got a 8600 GS from my father, I'll see if I can reproduce this issue. Too bad for your time, I'll see what I can do though. This could very well explain some instability I've been encountering. And one more thing, I did a hour ago. I documented power usage of nouveau vs nvidia. Same settings, same kernel, only GPU drivers differ: Result is very sad: nvidia: perf level0 : ~15.8-16.0W perf level2 : ~16.7-17.1W nouveau: perf level0 : ~20.4-20.8W perf level2 : ~23.0-23.5W This means 2 things. 1. There is major source of PM improvement somewhere else 2, Clocks don't affect things much (thats why somewhat wrong clocks set by nouveau don't show visible difference in performance/power usage. (Note: I was under false impression that when you force nvidia to maximum perf level, it lies to you and sets it only when device isn't idle. Well, according to registers 0x4000-0x4040 nvidia doesn't lie...) Also, interesting fact is that clock registers have many bits that nouveau doesn't touch, but they sure have some meaning. Well, reverse engineering is very tough job. In fact what nouveau developers did is already well beyond what believed to be possible, so no complains :-) Speaking of report on general PM regression reported by phoronix.com, its probably there but not as dramatic as they want it to appear. I remember being able (with help of some smoke and mirrors) to lower power consumption to 14W, so its higher a bit probably but not that much. -- Best regards, Maxim Levitsky Visit my blog: http://maximlevitsky.wordpress.com Warning: Above blog contains rants. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
On Thu, 28 Apr 2011 12:43:31 +0100, Maxim Levitsky maximlevit...@gmail.com wrote: Martin, one more thing, this is my observations regarding clocks I finished today: clock = (ref * N / M) (P 0x7) first line is blob, second nouveau level0: P NNMM 0x4008 - memory - 0018e200 2505 = 1 = 01. 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8001 0b01 = 11 / 1 / 2 = 05.5000 8002 1b02 = 27 / 2 / 4 = 03.3750 0x4028 - core - a012 1b04 = 27 / 4 / 4 = 01.6875 a002 1b04 = 27 / 4 / 4 = 01.6875 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8002 1b04 = 27 / 4 / 4 = 01.6875 level1: 0x4008 - memory - 809ae400 2505 = 37 / 5 / 4 = 01.8500 80026400 2505 = 37 / 5 / 4 = 01.8500 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8001 0b01 = 11 / 1 / 2 = 05.5000 0x4028 - core - a009 0b02 = 11 / 2 / 1 = 05.5000 a001 0b02 = 11 / 2 / 1 = 05.5000 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 0b02 = 11 / 2 / 2 = 05.5000 level2: 0x4008 - memory - 8059e400 2505 = 37 / 5 / 2 = 03.7000 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8000 1002 = 16 / 2 / 1 = 08. 0x4028 - core - a009 1002 = 16 / 2 / 2 = 04. a001 1002 = 16 / 2 / 2 = 04. 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 1202 = 18 / 2 / 2 = 06. Clocks are very different and we seems to overclock hard the unk_05 PLL And shader PLL is underclocked always, so another way to boost performance? (it didn't seem to affect perfomace much here, but yet...) Maxim most likely you are tired of hearing this, but the shader PLL in many cases is disabled(-ish) Either by nv_wr32(dev, (nv_rd32(dev, 0x4020) ~0x8000)), or by the 0xc040 register (don't remember the exact bit) Whereas for the unk_05 the logic of the blob has been changed in it's recent versions - i.e. blob v195.xx the numbers(P/M/N) where the same as the nouveau Cheers Emil (xexaxo) ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
On Thu, 2011-04-28 at 16:24 +0100, Emil Velikov wrote: On Thu, 28 Apr 2011 12:43:31 +0100, Maxim Levitsky maximlevit...@gmail.com wrote: Martin, one more thing, this is my observations regarding clocks I finished today: clock = (ref * N / M) (P 0x7) first line is blob, second nouveau level0: P NNMM 0x4008 - memory - 0018e200 2505 = 1 = 01. 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8001 0b01 = 11 / 1 / 2 = 05.5000 8002 1b02 = 27 / 2 / 4 = 03.3750 0x4028 - core - a012 1b04 = 27 / 4 / 4 = 01.6875 a002 1b04 = 27 / 4 / 4 = 01.6875 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8002 1b04 = 27 / 4 / 4 = 01.6875 level1: 0x4008 - memory - 809ae400 2505 = 37 / 5 / 4 = 01.8500 80026400 2505 = 37 / 5 / 4 = 01.8500 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8001 0b01 = 11 / 1 / 2 = 05.5000 0x4028 - core - a009 0b02 = 11 / 2 / 1 = 05.5000 a001 0b02 = 11 / 2 / 1 = 05.5000 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 0b02 = 11 / 2 / 2 = 05.5000 level2: 0x4008 - memory - 8059e400 2505 = 37 / 5 / 2 = 03.7000 80016400 2505 = 37 / 5 / 2 = 03.7000 0x4020 - shader - 8000 0b01 = 11 / 1 / 1 = 11. 8000 1002 = 16 / 2 / 1 = 08. 0x4028 - core - a009 1002 = 16 / 2 / 2 = 04. a001 1002 = 16 / 2 / 2 = 04. 0x4030 - unk_05 - a012 0b02 = 11 / 2 / 4 = 01.3750 8001 1202 = 18 / 2 / 2 = 06. Clocks are very different and we seems to overclock hard the unk_05 PLL And shader PLL is underclocked always, so another way to boost performance? (it didn't seem to affect perfomace much here, but yet...) Maxim most likely you are tired of hearing this, but the shader PLL in many cases is disabled(-ish) Either by nv_wr32(dev, (nv_rd32(dev, 0x4020) ~0x8000)), or by the 0xc040 register (don't remember the exact bit) Whereas for the unk_05 the logic of the blob has been changed in it's recent versions - i.e. blob v195.xx the numbers(P/M/N) where the same as the nouveau Cheers Emil (xexaxo) I sure didn't know about register 0xC040, but I know about bit 31 (0x8000) and took it into account. It isn't set only for memory clock in perf level 0 by the blob. Also clocking seems not to be the major power problem, there is something else. (see my post about observed power consumption). My gut feeling is that blob can decrease number of running shader cores, and in nouveau all are running anyway. Interesting fact is that GPU temperatures rise to very high levels (~75C) even while doing CPU only work (like compiling kernel for example). -- Best regards, Maxim Levitsky Visit my blog: http://maximlevitsky.wordpress.com Warning: Above blog contains rants. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
Le 28/04/2011 18:58, Maxim Levitsky a écrit : Interesting fact is that GPU temperatures rise to very high levels (~75C) even while doing CPU only work (like compiling kernel for example). Let me guess, you're on a laptop? The temperature of the case goes up when the processor is working and so, it cools the GPU less ;) ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
On Thu, 2011-04-28 at 20:24 +0200, Martin Peres wrote: Le 28/04/2011 18:58, Maxim Levitsky a écrit : Interesting fact is that GPU temperatures rise to very high levels (~75C) even while doing CPU only work (like compiling kernel for example). Let me guess, you're on a laptop? The temperature of the case goes up when the processor is working and so, it cools the GPU less ;) Yes, but that doesn't happen while using nvidia to this extent. I think that GPU has many units running in endless loop doing nothing but consuming power (~5W) difference! -- Best regards, Maxim Levitsky Visit my blog: http://maximlevitsky.wordpress.com Warning: Above blog contains rants. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
Le 28/04/2011 20:29, Maxim Levitsky a écrit : On Thu, 2011-04-28 at 20:24 +0200, Martin Peres wrote: Le 28/04/2011 18:58, Maxim Levitsky a écrit : Interesting fact is that GPU temperatures rise to very high levels (~75C) even while doing CPU only work (like compiling kernel for example). Let me guess, you're on a laptop? The temperature of the case goes up when the processor is working and so, it cools the GPU less ;) Yes, but that doesn't happen while using nvidia to this extent. I think that GPU has many units running in endless loop doing nothing but consuming power (~5W) difference! I really need to have a look at this. I have a power meter now (acpi reports me funky power consumption figures because my battery is dead). ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau
Re: [Nouveau] [PATCH 0/2] reclocking stability improvements
Hi. On 29/04/11 04:35, Martin Peres wrote: Le 28/04/2011 20:29, Maxim Levitsky a écrit : On Thu, 2011-04-28 at 20:24 +0200, Martin Peres wrote: Le 28/04/2011 18:58, Maxim Levitsky a écrit : Interesting fact is that GPU temperatures rise to very high levels (~75C) even while doing CPU only work (like compiling kernel for example). Let me guess, you're on a laptop? The temperature of the case goes up when the processor is working and so, it cools the GPU less ;) Yes, but that doesn't happen while using nvidia to this extent. I think that GPU has many units running in endless loop doing nothing but consuming power (~5W) difference! I really need to have a look at this. I have a power meter now (acpi reports me funky power consumption figures because my battery is dead). Just been reading through your posts from last night, and wanted to let you know I'm interested and willing to test patches too. I have an 8600GTS based laptop, and have seen the same range of power usage (ie have seen 14W in the page - presumably when I was using the blob -, but can only get it down to 23W at the moment using Nouveau). Regards, Nigel -- Evolution (n): A hypothetical process whereby improbable events occur with alarming frequency, order arises from chaos, and no one is given credit. ___ Nouveau mailing list Nouveau@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/nouveau