_mat_

GPUPI needs a lot of precision to output a valid score. Billions of smaller calculations are added up to a final sum. If your graphics card is unstable in any way, the chance is very high to get an invalid result.

Thanks for pointing that out. I have clarified this much butter (and in aggressive red ) now.

vcredist_x86.exe is the right one for the bench (and for any other 32 bit application running on 32 bit or 64 bit OS!).

Turrican's sister just posted a score with Karl's old daily system. GTX 295 with an old E8700 dual core! What else? Best part is, that the first few runs didn't work out this hardware, because there was some kind of memory error. Only be reducing the batch size, she was able to run it. Seems like Karl has benched the card pretty good!

Awesome model!

Awesome score! Regarding SSE2 and SSE3 as a minimum requirement for OpenCL: SSE3 is important because otherwise there isn't really a performance advantage running it in parallel. But the driver itself decides, what is 100% necessary to run the implementation. I didn't know it before, but the AMD drivers are fine with SSE2. Intel on the other hand needs SSE4.1 to make it work.

http://hwbot.org/newsflash/2781__turrican_memorial_charity_challenge_extended__additional_prizes_by_ek_water_blocks Frontpage news finally! Thanks for your support tiborrr, it's very much appreciated. Regarding the prizes, we have separated stage 1 and stage 2 so all the oldschool benchers won't need R9 290's to grab some loot (and the other way round). Without further ado, lets bench! Btw, the first shirts are being printed as we speak and will very soon be shipped.

It was my intention to build the benchmark as independent from the CPU as possible, so the overclocker can focus on the GPU only. But I thought about it and I may have a nice idea to be able to include both CPU and GPU into the calculation. It would be possible to let the user decide to offload a few batches to the CPU to speed up the benchmark furthermore. It won't be a big improvement to the final time, just a few seconds, but it would measure the performance of the whole system. Currently I am using my spare time to include multiple GPUs to the benchmark. Let's see what happens after that.

There would be no advantage in doing this except for better device compatibility. GPU is a lot faster than CPU.

I have no access to Radeon graphics cards at the moment, so I can't test it myself. Seems like between the kernel calls the memory goes back to idle mode. Definitely strange behaviour, that should cost some efficiency. Regarding stability testing, GPUPI is mainly using 64 bit integer calculations and some double precision math. Those parts of the hardware will be tested for stability of course, even the slightest miscalculation ends up in an invalid run. That's very different from game benchmarks, where only a big hiccup will stop the run (hardware lockup and the driver watchdog resets everything in about 5 seconds), small errors might be seen in certain shaders and different image quality, but not necessarily. But game benchmarks also test different parts of the hardware, because they mostly use shaders for floating point math, which is the fastest, most common use of graphics hardware. My guess for the additionally produced heat is the usage of more parts of the graphics pipeline and things like the geometry shaders (for tesselation) and so on. So I would recommend GPUPI as an additional GPU stability test that is focused on math precision and not on full utilization of all hardware features.

This should not be the case as far as I know. The newest drivers have had the most impact in the past. But I'm always curious, the bench is pretty new and tweaks/best practices have yet to be found.

It's mostly an OpenCL driver issue. If I may ask, why even bother with XP?

You are using a very old AMD APP driver. Installing the latest Catalyst drivers or the lastest AMD APP SDK should fix this. You can check the driver version in the benchmark itself where it says "AMD APP (1xxx) is ready". The number should be 1500+ in the latest versions.

What problems exactly? With OpenCL drivers? The shop will guide you through. You need a PayPal account for international shipping, that's for sure.

I will ask our delivery service, if it's possible and let you know. Regarding the VC redistributable error, I've never seen something like this. I guess one of your installed packages was damaged, so it was fixed by removing these. It's also good to know that the 2013 package is only needed in 32 bit, no matter if your OS is in 32 vor 64 bit.

Taloken, well explained. Thanks. saint, I coded the shop myself. How can I help?

CPU from 2004 will always win. But if you both have the same CPU, best score wins.

Nice, thanks!

First off a word about compatibility. I'm using the newest CUDA toolkit for best performance, but it doesn't support Windows XP. That's why GPUPI.exe and GPUPI_CUDA.exe won't work on XP and Server 2003. But the pure OpenCL version does. You just need the VS2013 redistributable and some OpenCL drivers of course. As for OpenCL itself, it will need SSE3 support, so compatibility goes back to Pentium 4 (since Prescott) on Socket 478 and old AMD Athlon 64 (since Venice) on Socket 754. Here is an outdated list of CPUs with SSE3: http://de.wikipedia.org/wiki/Streaming_SIMD_Extensions_3 That's part of the challenge, as well as finding a working driver.

Dear community, the most quiet time of the year is here, where we think about about our family and friends. This year we have lost a big part of the overclocking community, our dear Karl. So I would like to invite you to join us in honoring his name by participating in our donation campaign for an Austrian children hospital dedicated to researching cancer especially for young patients. The idea was born when you guys asked where you could donate to his family, but they wanted you to help the hospital instead. Now this idea is wrapped up the way Turrican would like it ... To cut things short, overclockers.at (my site), our friends from Noctua and HWBot have teamed up to present you the Turrican Memorial Challenge! It's a contest with two stages using GPUPI, the benchmark I dedicated to Turrican a few weeks ago. The first stage is called "Turrican's choice" and focuses on old hardware. Run GPUPI in 1B on the oldest CPU possible and overclock the hell out of it afterwards. The oldest chip with the fastest run wins! Yes, you know you wouldn't stand a chance if Karl could participate. The second stage is GPUPI in 1B for graphics cards all out. The fastest run wins, whatever it takes! We would like to see some crazy 290s on liquid nitrogen here, so be ready to refill your containers. But this contest should not be all about winning and prizes. Noctua, some of the guys there knew Turrican personally, will donate 1 Euro for every participant of this contest! So get ready to crunch some pi on your system, all of you can make this happen. The contest will start whenever HWBot is ready. It will be officially announced in the news. In addition to the contest we have prepared something special to take the donation campaign to another level. At overclockers.at we have been giving out "Hardware Terminator" shirts for the winners of our little overclocking contests. It's Turrican's nickname that Monstru gave him when he visited Austria for an OC event in 2009. It kind of stuck and we made a couple of shirts just for fun. They were not for sale, they had to be earned by liquid nitrogen, sweat and blood. Turrican wearing is his shirt surround by his most favorite toys. For this donation campaign we've redesigned this shirts a bit and made them available in our web shop in multiple cuts and colors. They will be printed on demand in small by a little company here in Austria and shipped by us as soon as they are ready. To be able to buy one, you have to donate at least 5 Euros to the donation campaign. We let the amount of the donation open to your choosing. Every Euro is appreciated and will be directly forwarded as it is. The proceedings of the whole campaign will be officially handed over to St. Anna Kinderspital in February next year. More details on that as soon as possible. I wish you all a merry christmas, have fun benching, get yourself a nice shirt and help us with this good cause! Links: English web shop with international shipping | Turrican Memory Challenge (starts soon)

that efficiency!

GPUPI 1.4.1 is now available for download and fixes a minor bug with a certain OpenCL CPU configuration. Thanks again to Genieben for his bug report! https://www.overclockers.at/news/gpupi-international-support-thread

The hwbot API can automate the uploading of submission results and already supports encryption. I don't know if some kind of validation is possible, but I guess not because the logic would have to be on the server's side. Well, if I can add something to the bench to actively prevent photoshop overclocking, I'll do it. But these are clearly bigger issues, that have to be decided by the hwbot staff. Edit: Owned. Well, if it's possible, I'll see to it. Genieben, do you know if there are C++ examples for hwbot automation?

Thanks for the video, with your help I could find a possible solution for the problem. Please test this version and see if it works now: Link removed It's just a temporary link, if this fixes the problem, I will update the download links. EDIT: GPUPI 1.4.1 is now available for download. https://www.overclockers.at/news/gpupi-international-support-thread