Hello,

I tried to run GPU application from here on my GTX 680.

According GPU-Z here is about 10-15 seconds of 90% GPU usage and then 0% for long time. For example I canceled this task after long runtime: https://numberfields.asu.edu/NumberFields/workunit.php?wuid=97780201.

Where can be the problem?

2021.01.05 16:21:47 | | CUDA: NVIDIA GPU 0: GeForce GTX 680 (driver version 460.89, CUDA version 11.2, compute capability 3.0, 2048MB, 1672MB available, 3252 GFLOPS peak)
2021.01.05 16:21:47 | | OpenCL: NVIDIA GPU 0: GeForce GTX 680 (driver version 460.89, device version OpenCL 1.2 CUDA, 2048MB, 1672MB available, 3252 GFLOPS peak)
2021.01.05 16:21:47 | | OpenCL: Intel GPU 0: Intel(R) HD Graphics 4600 (driver version 20.19.15.4835, device version OpenCL 1.2, 1630MB, 1630MB available, 240 GFLOPS peak)
2021.01.05 16:21:47 | | OpenCL CPU: Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz (OpenCL driver vendor: Intel(R) Corporation, driver version 5.2.0.10094, device version OpenCL 1.2 (Build 10094))
2021.01.05 16:21:47 | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz [Family 6 Model 60 Stepping 3]
2021.01.05 16:21:47 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx smx tm2 pbe fsgsbase bmi1 smep bmi2
2021.01.05 16:21:47 | | OS: Microsoft Windows 10: Professional x64 Edition, (10.00.19042.00)
2021.01.05 16:21:47 | | Memory: 15.69 GB physical, 31.69 GB virtual


Thank You.

Sorry about that!

I think the problem is the age of the card, given the complexity of the application. I recently reduced the required compute capability from 3.5 down to 3.0 for another user. That may have been a mistake, for the reasons you are now seeing.

If memory serves, the other user had a 7xx or 8xx series card (with compcap 3.0). Their run times were longer than usual but did complete successfully, and as far as I know their gpu had high utilization. The fact that your utilization drops to zero tells me that the card doesn't have enough compute resources to complete the calculation, and therefore hangs. That's my best guess...

I think in your case, the CPU version would be much more efficient. In the next week or so I will be putting out a new openCL version of the app. One of the main design changes is to use many smaller kernels instead of several big kernels. This may help with the resource problem.

Thank You for information, Eric.

By the way, this card is not very bad, it is similar to GTX 960 from newer series. Maybe problem can be 2 GB of RAM, but according GPU-Z app using about 400 MB only.

Good luck :)

My gpu is NVIDIA GeForce GTX 1650

My gpu is NVIDIA GeForce GTX 1650

Your card should work well. Do you have a question?

Im using a R9 280x and a Radeon 7900. There was no load on the GPU and about half way through I got a computational error. The 3rd wu is still running.
I'll let it go for a while and see what happens.
Are mine too old also?

Edit - I think the answer is yes, those 2 cards are too old. They errored out whilst my Radeon VII breezed through it.
Oh well.

---