GPUPi (OPenCL Pi gpu Benchmark )

Anyone got any idea why this bench seems to think I'm running two 280Xs and not two 7970s?

 

My 7950 was also registered / put into the same category as the R9 280.

 

AMD's OpenCL driver only gives access to the codename of the GPU, which is the same for rebranded graphics cards.

 

Several diagnostics are still reporting "OpenCL 1.2". What's up with that?


GPU Caps Viewer:

Ver.: OpenCL 1.2 AMD-APP (1642.5)
-CL_DEVICE_VERSION:OpenCL 1.2 AMD-APP (1642.5)
Version/Profile: OpenCL 2.0 AMD-APP (1642.5) FULL_PROFILE

 

need to check with the guys who have done thoses diagnostics softwares ? even with DX i see some funny things.

 

what driver to use with 290x on this bench? my titan x runs it faster and that should not be

 

For be honest, i have not try this one since a good time, i see a new version is out ( 2.1.1 ). As for driver, sadly i dont know.

Personally i have install the separated complete AMD APP SDK 2.0, as i use the compilers for OpenCL accelerated render engine for Blender+Luxrender. (Luxcore + path OpenCL ). so i dont know if theres something who can change from driver to driver.

 

still thought it would be 10 second runs at 1300 or so on my 290x I have all the latest thing as you describe

 

Really unstable but i was able to run the benchmark ...

 

Another run on Win 10:
980 @ 1505 and 8ghz memory : 30.030s + 1.005s

OpenCL 1.2 CUDA 7.5.9 is ready. Timer: HPET (14.32 MHz)

OpenCL GPU: NVIDIA GeForce GTX 980 (16 CUs, 1367 MHz)
Compiling OpenCL kernels ... done.

Calculating 1.000.000.000th digit of PI. 20 iterations.

Allocated device memory : 335546368 Bytes
Batch Size : 20M
Reduction Size : 64

00h 00m 00.715s Batch 1 finished.
00h 00m 01.404s Batch 2 finished.
00h 00m 02.459s Batch 3 finished.
00h 00m 05.158s Batch 4 finished.
00h 00m 07.840s Batch 5 finished.
00h 00m 08.518s Batch 6 finished.
00h 00m 09.191s Batch 7 finished.
00h 00m 10.231s Batch 8 finished.
00h 00m 12.891s Batch 9 finished.
00h 00m 15.530s Batch 10 finished.
00h 00m 16.208s Batch 11 finished.
00h 00m 16.885s Batch 12 finished.
00h 00m 17.933s Batch 13 finished.
00h 00m 20.642s Batch 14 finished.
00h 00m 23.330s Batch 15 finished.
00h 00m 24.012s Batch 16 finished.
00h 00m 24.689s Batch 17 finished.
00h 00m 25.734s Batch 18 finished.
00h 00m 28.401s Batch 19 finished.
00h 00m 31.043s PI value output -> 5895585A0

Statistics

Calculation + Reduction time: 30.030s + 1.005s

 

Running Win 10

Error: High Precision Event Timer for time measurement not found!

This is mandatory for Windows 8 and higher to avoid skewed timing.
You'll find a guide on how to enable the HPET timer on your system
in our FAQ. Open about dialog in the menu above for the link.

--
Haven't disabled HPET, a Win 10 bug?

 

The value of useplatformclock should be "Yes". If it's not, you can fix this by running:
SHELL:
bcdedit /set useplatformclock yes

A reboot might be necessary afterwards.

 

Thanks, is there a command to check the status before changing it?

To see what it was defaulted to

EDIT - Found it: bcdedit /deletevalue useplatformclock

If you get an error (Which I did) it was disabled

---Working now---

OpenCL 1.2 CUDA 7.5.14 is ready. Timer: HPET (14.32 MHz)

OpenCL GPU: NVIDIA GeForce GTX 980 (16 CUs, 1367 MHz)
Compiling OpenCL kernels ... done.

Calculating 1.000.000.000th digit of PI. 20 iterations.

Allocated device memory : 335546368 Bytes
Batch Size : 20M
Reduction Size : 64

00h 00m 00.719s Batch 1 finished.
00h 00m 01.398s Batch 2 finished.
00h 00m 02.441s Batch 3 finished.
00h 00m 05.112s Batch 4 finished.
00h 00m 07.763s Batch 5 finished.
00h 00m 08.431s Batch 6 finished.
00h 00m 09.095s Batch 7 finished.
00h 00m 10.120s Batch 8 finished.
00h 00m 12.750s Batch 9 finished.
00h 00m 15.360s Batch 10 finished.
00h 00m 16.030s Batch 11 finished.
00h 00m 16.696s Batch 12 finished.
00h 00m 17.731s Batch 13 finished.
00h 00m 20.408s Batch 14 finished.
00h 00m 23.064s Batch 15 finished.
00h 00m 23.735s Batch 16 finished.
00h 00m 24.401s Batch 17 finished.
00h 00m 25.431s Batch 18 finished.
00h 00m 28.066s Batch 19 finished.
00h 00m 30.678s PI value output -> 5895585A0

Statistics

Calculation + Reduction time: 29.632s + 1.026s

 

980 @ 1661

Cuda - Calculation + Reduction time: 30.305s + 0.934s
OpenCL - Calculation + Reduction time: 27.160s + 0.886s

------------------------------------------------------------------------

CUDA GPU: GeForce GTX 980
Kernel 1, Batch Size: 20M, Blocks: 20480, Threads: 1024
Kernel 2, Batch Size: 20M, Blocks: 27307, Threads: 768

Calculating 1.000.000.000th digit of PI. 20 iterations.

Allocated device memory : 335549456 Bytes
Batch Size : 20M
Reduction Size : 64

00h 00m 00.677s Batch 1 finished.
00h 00m 01.318s Batch 2 finished.
00h 00m 02.556s Batch 3 finished.
00h 00m 05.259s Batch 4 finished.
00h 00m 07.893s Batch 5 finished.
00h 00m 08.541s Batch 6 finished.
00h 00m 09.182s Batch 7 finished.
00h 00m 10.408s Batch 8 finished.
00h 00m 13.053s Batch 9 finished.
00h 00m 15.632s Batch 10 finished.
00h 00m 16.281s Batch 11 finished.
00h 00m 16.923s Batch 12 finished.
00h 00m 18.162s Batch 13 finished.
00h 00m 20.866s Batch 14 finished.
00h 00m 23.503s Batch 15 finished.
00h 00m 24.151s Batch 16 finished.
00h 00m 24.794s Batch 17 finished.
00h 00m 26.019s Batch 18 finished.
00h 00m 28.666s Batch 19 finished.
00h 00m 31.247s PI value output -> 5895585A0

Statistics

Calculation + Reduction time: 30.305s + 0.934s

OpenCL 1.2 CUDA 7.5.15 is ready.

OpenCL GPU: NVIDIA GeForce GTX 980 (16 CUs, 1481 MHz)
Compiling OpenCL kernels ... done.

Calculating 1.000.000.000th digit of PI. 20 iterations.

Allocated device memory : 335546368 Bytes
Batch Size : 20M
Reduction Size : 64

00h 00m 00.627s Batch 1 finished.
00h 00m 01.226s Batch 2 finished.
00h 00m 02.167s Batch 3 finished.
00h 00m 04.627s Batch 4 finished.
00h 00m 07.066s Batch 5 finished.
00h 00m 07.672s Batch 6 finished.
00h 00m 08.272s Batch 7 finished.
00h 00m 09.208s Batch 8 finished.
00h 00m 11.631s Batch 9 finished.
00h 00m 14.031s Batch 10 finished.
00h 00m 14.638s Batch 11 finished.
00h 00m 15.238s Batch 12 finished.
00h 00m 16.180s Batch 13 finished.
00h 00m 18.642s Batch 14 finished.
00h 00m 21.084s Batch 15 finished.
00h 00m 21.691s Batch 16 finished.
00h 00m 22.292s Batch 17 finished.
00h 00m 23.229s Batch 18 finished.
00h 00m 25.653s Batch 19 finished.
00h 00m 28.055s PI value output -> 5895585A0

Statistics

Calculation + Reduction time: 27.160s + 0.886s

 

HD7870 + r7_240 gives 65 seconds @ default settings. (doesnt use oc setting for a reason)