_mat_

I am currently working on GPUPI 3.0 with Volta support, did some testing on a Titan V as well. After looking into this score I am a bit puzzled about your CU count and the difference in bootup clock frequency of your Titan V. I am not able to reproduce this and the score doesn't add up as well. I have reported this score so we can further look into it. Please be so kind and provide any alterations you did on your card plus details of your setup like driver version. You can send it to me personally: matthias at hwbot.org

Impressive CPU score with an impressive CPU!

The Tensor Cores are for tensor mathematics and are not general purpose like the other cores. They won't play a part in the calculation of Pi. The issue is with the memory reduction to calculate the final sums of the billions of partial results. I don't know what's happening but the kernel seems to be incompatible. I already reported the issue to NVIDIA and asked to borrow me a sample. No answer as usual. Anybody has an NVIDIA contact in the US that actually knows what he's talking about?

GPUPI actually detects your 3rd core, each is displayed as one Compute Unit so in the header it states: "OpenCL CPU: AuthenticAMD ... (3 CUs, 3933 MHz). It's a submission problem and most problably fixable. Please send me a data file of your submission to matthias at hwbot.org and I will look into it.

The windows version doesn't matter for GPUPI, so that's not a surprise. What was the error with CUDA? Can you it to me please, I will look into it (matthias at hwbot.org).

Oh my god. It's even better than I thought. Why did you use the OpenCL driver? Was there a problem with the normal CUDA path? It should be significantly faster.

7.1? That's crazy!

Just wow!

I tested both files and they are valid. I uploaded them without problems via the HWBOT "submit score" site as well. As it seems that the upload was not successful, I guess this is related to problems with your internet connection or your browser settings/plugins. I recommend to try a different browser first.

Please send the result files to matthias [at] hwbot.org, I will have a look at them.

Probably uploaded in the wrong category, right? The error messages are not very clear on that. You have to be sure to upload CPU scores in the "GPUPI for CPU" categories and GPU scores into the "GPUPI" categories. It's best to submit the scores inside the benchmark, if your bench system is connected to the internet.

Ryzen is not split up in GPUPI. Each compatible OpenCL platform will show one Ryzen device. You can install multiple AMD OpenCL platforms to try out multiple OpenCL drivers, each will perform differently. Search for the AMD App SDK, version 3.0 will install the OpenCL 2.0 driver, version 2.9 should result in OpenCL 1.2.

What? Skylake X is not supported by Intel's official OpenCL drivers? Now I see why the current scores where made with the AMD OpenCL driver. That's really sad.

That's strange indeed. Please send one of these result files to matthias at hwbot org, I will have a look at it. Thank you.

The submission node is the root node of the HWBOT validation file. But the errors before that already make it very clear, that the data could not be decrypted correctly. Sry, that's bad news. Seems like the file was damaged at some point.

Check GPUPI.log there should be an extended error description which XML node makes the file invalid. Please post it here, might be a bug in the validation.

A little preview of some of the features of GPUPI 3.0 for my fellow overclockers. Command line version for Windows: Autoselection of the compute platform, Batch Size and Reduction Size depending by prebenching it for the user:

Thank you for your kind words, very much appreciated.

Windows 8 and above are effected by the RTC skewing bug when bclock is changed in Windows. I don't think that Skylake and Kaby Lake are any exception to this rule, but I haven't tested it myself yet. To circumvent HPET you have to use Windows 7. Edit: Rules of HWBOT allow the legacy benchmarks on SL and KL, so I guess it has been tested and it's not affecting the RTC timer. Well, with the next version GPUPI I will remove the HPET restriction on SL and KL.

It's not a good sign for your windows install, if the COM library can't be initialized. These are pretty basic methods to query WMI information. Disable the hardware detection for CPU and GPU on the submission dialog.

Hey Nick, overclockers.at (where the download is hosted) was temporarily down due to a new forum software update. It has been back a few hours later, so is the download as well. Sry for the inconvencience!

Wow! 5M/512?!

Way to go, congrats! Six of these puppies could actually do it!

Well done! GPUPI 1B on those big chips is a hell of a ride, congrats!

You need to install the Intel OpenCL drivers: https://software.intel.com/en-us/articles/opencl-drivers