lanbonden Posted August 9, 2015 Share Posted August 9, 2015 I decided to try GPUIP for that Slowest GPUPI for CPU 1B part in the team cpu today and encounted a very wierd problem as the showed time in the benchmark doesnt match the real time it took to run it. This is a far from complete run but its already pretty far out of sync Id say. I know the screenshoot doesnt say much but I keep an external document running for comparing times as I was suprised the first rbatch only showed up as 12 mins when it feelt way longer. For example, I know the 00h 38mins 20 sec batch occured between 17:42 and 17:50 and the 01h 22mins 55sec batch occured between 19:10 and 19:19 The time differense between those 2 batches where ~45 mins but even the minimum time between my messurments is 17:50 to 19:10 or 80 mins. Next round 01h 35min 17sec, its just 13min after the batch before but it showed up between 19:39 and 19:46 so atleast 20 (19:19 to 19:39) mins after the batch before. Same thing happend for the batch after the screen shoot, 01h 48min 12sec. It showed up exactly 20:07 so atleast 21mins after even if GPUPI only says there is 13 mins differense. The clock in windows is staying synced so thats not the problem atleast. Is this a known problem? Any idea what could couse it? Its very anoying now when doing a slow run but I have a feeling it might be of great advantage when trying to do a fast run instead... Quote Link to comment Share on other sites More sharing options...
GENiEBEN Posted August 9, 2015 Share Posted August 9, 2015 Interesting, I am on my 80hrs run now, but I will test on the 100hrs run. Quote Link to comment Share on other sites More sharing options...
lanbonden Posted August 9, 2015 Share Posted August 9, 2015 Interesting, I am on my 80hrs run now, but I will test on the 100hrs run. on a run that long should it be rather easy to see, I meen Im close to twice the time compared to what the program reports. 1h 34mins as reported in the benchmark have taken ~2h50mins in reality so should be easy to see on a 80h run if your missing a few days in the ressult Quote Link to comment Share on other sites More sharing options...
trodas Posted August 9, 2015 Share Posted August 9, 2015 I envy people that can start the GPUPI Quote Link to comment Share on other sites More sharing options...
_mat_ Posted August 10, 2015 Author Share Posted August 10, 2015 lanbonden, the problem is with your OS timer. Have a look at your HPET frequency, it's way out of bounds. Seems like QueryPerformanceFrequency() returns a very wrong value on XP, that GPUPI can't handle at all. Which SP of XP is this? I will have to look into that. A quick solution for you: Disable your HPET timer in the BIOS and it should automatically fall back to RTC. Quote Link to comment Share on other sites More sharing options...
lanbonden Posted August 10, 2015 Share Posted August 10, 2015 lanbonden, the problem is with your OS timer. Have a look at your HPET frequency, it's way out of bounds. Seems like QueryPerformanceFrequency() returns a very wrong value on XP, that GPUPI can't handle at all. Which SP of XP is this? I will have to look into that. A quick solution for you: Disable your HPET timer in the BIOS and it should automatically fall back to RTC. This was run on XP pro, 32bit with sp3. As of the quick solution, there is nothing called HPET mentioned in bios or manual and from what Ive read its introduced around 2005 so after this motherboards are launched (dfi lanparty ut nf3 250gb). I was about to try and disabling HPET in windows to try and see if it worked tho but managed to corupt the instalation before I could do that. I can also say that the timer worked good (no noticible variation atleast) when I tried running at the normal frequency (200fsb) atleast, unfortunatly didnt get a chanse to test how higher FSB affected the timer as that gave the corrupt OS... Quote Link to comment Share on other sites More sharing options...
_mat_ Posted August 10, 2015 Author Share Posted August 10, 2015 skulstation - I wonder, what to do, when GPUPI just output the message "Could not start worker thread" ... I have debugged this issue and Intel OpenCL drivers are not compatible with Core2Duos and Core2Quads anymore - at least with the drivers version 15.1 and 14.2. It might work with older versions of the SDK, but they are simply not available on the internet. Btw, if you want to downgrade the Intel OpenCL drivers you have to manually delete the following registry key: HKEY_LOCAL_MACHINE\SOFTWARE\Intel\OpenCL Otherwise the installer will tell you that there is already a higher version installed, even if you have uninstalled it. But there is a solution: The AMD OpenCL drivers that are automatically installed with the Catalyst drivers DO WORK with these old Intel CPUs. Just go with them, it will work. You might have to plug in an AMD card to be able to install the Catalyst drivers though. Quote Link to comment Share on other sites More sharing options...
trodas Posted August 10, 2015 Share Posted August 10, 2015 (edited) So, after uninstall of the Intel OpenCL, I use this: REGEDIT4 [-HKEY_LOCAL_MACHINE\SOFTWARE\Intel\OpenCL] Done. Ah, thanks for the tips, _mat_ ! Did someone reported that bug to Intel yet? I, sadly, did not have any OpenCL supported AMD cards, so could you pretty please extract the OpenCL installer for me from the drivers, so I'm out of luck? Edited August 10, 2015 by trodas Quote Link to comment Share on other sites More sharing options...
_mat_ Posted August 10, 2015 Author Share Posted August 10, 2015 You can install the AMD APP SDK as well, it should work without an AMD graphics card. Btw, the graphics card is only needed to be able to install the Catalyst drivers. Otherwise the installation quits because there is "no supported hardware" found on the system. Quote Link to comment Share on other sites More sharing options...
trodas Posted August 10, 2015 Share Posted August 10, 2015 Sh...! That is a problem, because from the list of supported graphic cards: http://developer.amd.com/tools-and-sdks/opencl-zone/amd-accelerated-parallel-processing-app-sdk/system-requirements-driver-compatibility/ ...I have none. Lowest for OpenCL 1.1 is HD 5400 Series GPUs, for OpenCL 1.2 are Radeon HD 5800 and for OpenCL 2.0 it is Radeon HD 7790 Radeon 9600 XT does not qualify at all. Suxx. So I'm where I was started. Unless these OpenCL drivers cannot be somehow extracted from the drivers, then I'm stuck Quote Link to comment Share on other sites More sharing options...
GENiEBEN Posted August 10, 2015 Share Posted August 10, 2015 App SDK 2.9 does not ask for compatible hw to be installed in system. Quote Link to comment Share on other sites More sharing options...
_mat_ Posted August 10, 2015 Author Share Posted August 10, 2015 This was run on XP pro, 32bit with sp3. As of the quick solution, there is nothing called HPET mentioned in bios or manual and from what Ive read its introduced around 2005 so after this motherboards are launched (dfi lanparty ut nf3 250gb). I was about to try and disabling HPET in windows to try and see if it worked tho but managed to corupt the instalation before I could do that. I can also say that the timer worked good (no noticible variation atleast) when I tried running at the normal frequency (200fsb) atleast, unfortunatly didnt get a chanse to test how higher FSB affected the timer as that gave the corrupt OS... I have looked into this issue now and it seems like your mainboard and OS combination falls back to an invalid and likely bus dependend timer function. That's why the time is skewed. I have already implemented a more secure way to detect issues with the timer to be able to fall back to RTC on XP accordingly. It will be published with the next version of GPUPI. Quote Link to comment Share on other sites More sharing options...
_mat_ Posted August 10, 2015 Author Share Posted August 10, 2015 Ah, thanks for the tips, _mat_ ! Did someone reported that bug to Intel yet?I, sadly, did not have any OpenCL supported AMD cards, so could you pretty please extract the OpenCL installer for me from the drivers, so I'm out of luck? As GENIEBEN already posted, you should be able to install the AMD OpenCL drivers with the APP SDK 2.9.1. I will be at an Intel Technical Workshop in London soon, where one of the topics is parallel processing. Hopefully I can make some contacts to forward our findings to the Intel OpenCL driver team, so the bugs can get addressed properly. Quote Link to comment Share on other sites More sharing options...
trodas Posted August 10, 2015 Share Posted August 10, 2015 GENiEBEN - thanks for the tip! I installed the ADM SDK v2.9 and the install went fine. I just did not installed the samples, but that it is... However the GPUPI still failed to work at all: Error code -1001: could not query for available OpenCL devices! Damn. Should I try some settings or higher version... or... something? ... _mat_ - well, GENiEBEN specificaly mentioned v2.9, not 2.9.1 ... so I go on with the old version. I emailed the two responsible people (Russ Beutler - russel.r.beutler@intel.com and Kathy Farrel kathy.a.farrel@intel.com) and hopefully someone take a look soon to these failures Thanks for your efforts. Any debudging tools that could help me out? It does not look like the AMD OpenCL drivers accept my CPU as the OpenCL rendering unit.... Quote Link to comment Share on other sites More sharing options...
lanbonden Posted August 10, 2015 Share Posted August 10, 2015 Sh...! That is a problem, because from the list of supported graphic cards:http://developer.amd.com/tools-and-sdks/opencl-zone/amd-accelerated-parallel-processing-app-sdk/system-requirements-driver-compatibility/ ...I have none. Lowest for OpenCL 1.1 is HD 5400 Series GPUs, for OpenCL 1.2 are Radeon HD 5800 and for OpenCL 2.0 it is Radeon HD 7790 Radeon 9600 XT does not qualify at all. Suxx. So I'm where I was started. Unless these OpenCL drivers cannot be somehow extracted from the drivers, then I'm stuck It works without a problem to install a driver thats not suported by the card. I installed the driver catalyst 12.1 with a radeon 7000 card (not a HD7000 but a really really old card) and GPUPi worked after that (well, except for my timing problem but that didnt have anything with the driver to do probably). Quote Link to comment Share on other sites More sharing options...
trodas Posted August 10, 2015 Share Posted August 10, 2015 Hmmm, it keep pushing me v10.2 for R 9600 XT: http://support.amd.com/en-us/download/desktop/legacy?product=Legacy1&os=Windows%20Vista%20-%2064# So finding a Catalyst 12.1 could be harder... Hopefully this will be it: http://news.softpedia.com/news/Download-AMD-Catalyst-12-1-Display-Driver-248748.shtml Quote Link to comment Share on other sites More sharing options...
lanbonden Posted August 10, 2015 Share Posted August 10, 2015 Hmmm, it keep pushing me v10.2 for R 9600 XT:http://support.amd.com/en-us/download/desktop/legacy?product=Legacy1&os=Windows%20Vista%20-%2064# So finding a Catalyst 12.1 could be harder... Hopefully this will be it: http://news.softpedia.com/news/Download-AMD-Catalyst-12-1-Display-Driver-248748.shtml Think 12.1 was the driver you get when you search for drivers to a X800 card. Quote Link to comment Share on other sites More sharing options...
trodas Posted August 10, 2015 Share Posted August 10, 2015 (edited) Well, that works! Thanks! Removed the 6800 GT, put there R 9600 XT, installed the Catalysts 12.1 and voila! All is working: Thanks a lot, lanbonden for the hint. I did not search for X800 card, as I did not have it, lol. *** Now it makes me wonder, witch reduction size setting is fastest and witch is slowest on my HW. Looks like that reduction size 16 is 9sec slower that reduction 32... on HWbot guide is recommended 64, but what happens using 512? PS: answer is there: Reduction size 512 - 4:41.085 Reduction size 256 - 3:10.086 Reduction size 128 - 3.02.958 Reduction size 64 - 3:01.822 Reduction size 32 - 3:01.939 Reduction size 16 - 3:10.934 Edited August 10, 2015 by trodas Quote Link to comment Share on other sites More sharing options...
lanbonden Posted August 10, 2015 Share Posted August 10, 2015 I have looked into this issue now and it seems like your mainboard and OS combination falls back to an invalid and likely bus dependend timer function. That's why the time is skewed. I have already implemented a more secure way to detect issues with the timer to be able to fall back to RTC on XP accordingly. It will be published with the next version of GPUPI. I spent some more time experimenting after I repaired the OS. Lower fsb then standard makes the reported time to be to fast like previous mentioned. I tested overclocking and then the reported time is slower then it should be just like expected. I also tested to not be lazy and change fsb in bios as Ive only used SetFSB due to not being able to underclock in bios. The good thing for overclocking atleast is that higher fsb in bios doesn create the time drift and everything works like it should when doing it the real way. I also tested a fast run of superpi, and it wasnt effected by the time drift when using SetFSB. Thanks for the response and help and hope its fixed in next version Quote Link to comment Share on other sites More sharing options...
GENiEBEN Posted August 10, 2015 Share Posted August 10, 2015 As GENIEBEN already posted, you should be able to install the AMD OpenCL drivers with the APP SDK 2.9.1. I will be at an Intel Technical Workshop in London soon, where one of the topics is parallel processing. Hopefully I can make some contacts to forward our findings to the Intel OpenCL driver team, so the bugs can get addressed properly. Lmk when you're there mate so we can have a pint. Quote Link to comment Share on other sites More sharing options...
_mat_ Posted August 13, 2015 Author Share Posted August 13, 2015 GPUPI 2.2 is out now! Thanks to the feedback from you guys, I was able to fix a lot of things and improve many others. It will be mandatory to use this version in the future, because there are some important changes to make the bench even more bullet proof. But we will wait until the the currently started competitions have finished, including the Team Cup of course. Last but not least I would like to talk about our next plans with GPUPI. Thanks to HWBOT I can integrate CPU-Z into the next version, which will improve hardware detection a lot and allows several frequencies and even voltages to be automatically submitted to the database. Additionally we are going to include support for HWBOT competitions directly into the online submission dialog. I already have a prototype working, but I didn't want to rush anything. Full changelog and download here: https://www.overclockers.at/news/gpupi-2-2-english Quote Link to comment Share on other sites More sharing options...
trodas Posted August 13, 2015 Share Posted August 13, 2015 we will wait until the the currently started competitions have finished, including the Team Cup of course I hope so. Got only 2nd pass with 5h 32min... and I would hate to cancel the test running so nicely slow Quote Link to comment Share on other sites More sharing options...
_mat_ Posted August 13, 2015 Author Share Posted August 13, 2015 :banana: Quote Link to comment Share on other sites More sharing options...
trodas Posted August 13, 2015 Share Posted August 13, 2015 Well, I'm not that optimistic, since 1st score have 1 batch at 3h 14min and 2 batch at 7h 1min. Mine times for 1 batch are 2h 46 min and 2 batch at 5h 32min mark. That clearly put them way behind the 1st score by GENiEBEN. Sure it beat his older score ( http://hwbot.org/submission/2938964_ ) with goes to 2:25 and 4:50 ...but that was old score Beating old score is nothing earthshaking. But I get wild ideas what can I do with another mobo, if I did not manage to lower the FSB with ASRock 775i65G, witch freeze each time I change the FSB using SetFSB _AND_ fail to post, each time I set under 100MHz for FSB in bios, that allow FSB 95 as lowest... But the question is - will that score be ready in time to end the Team Cup? Quote Link to comment Share on other sites More sharing options...
_mat_ Posted August 13, 2015 Author Share Posted August 13, 2015 Slow it down by running another benchmark in parallel. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.