I have a device providing the peak GFLOPS specs and I want to measure how far my program is away from it. Since all the data I used was double precision, should I multiply the number of ops by 2 to get the GLOPS value and do the comparison?
Asked
Active
Viewed 2,150 times
1
-
What device is this? In most cases dp-FLOPS is half of the sp-FLOPS. – Mysticial Mar 19 '13 at 18:45
-
nVidia graphic cards. – Hailiang Zhang Mar 19 '13 at 18:46
-
Oh. Then that's different. I'm not too familiar with GPUs. – Mysticial Mar 19 '13 at 18:48
-
Zhang, what is the generation of your Nvidia cards (or just its model)? Do you use CUDA or something else? – osgx Mar 19 '13 at 18:50
-
Its tesla 2090. I just found its gflops for double precision, which is just twice that of single precision. Sorry about the confusion. – Hailiang Zhang Mar 19 '13 at 19:11
2 Answers
3
No. 1 double-precision floating-point operation is still one floating-point operation.
Most GPUs process double-precision data slower than single-precision, so there should be two specifications of peak GFLOPS. One peak single-precision GFLOPS spec, and one peak double-precision GFLOPS spec. Sometimes it is broken done further, so that (for example) peak division performance is listed separately from peak addition performance.
Ben Voigt
- 277,958
- 43
- 419
- 720
0
" ... , should I multiply the number of ops by 2 to get the GLOPS value and do the comparison?"
No, not for any (but one) of these Cards: http://www.geeks3d.com/20140305/amd-radeon-and-nvidia-geforce-fp32-fp64-gflops-table-computing/ .
Note that the ratio varies from 1/24th to as good as 1/3 in most cases, also note that the 'Workstation Graphics Card' has a ratio 1/2 - it is specifically designed that way to improve DP performance.
You need to read the Specs for the Hardware in your Card and determine what performance hit you should expect from switching to DP from SP. There will be a small additional amount of overhead to load the additional precision into the Registers (Memory where the Hardware will perform the Operation on) and to retrieve the additional precision after each Operation.
Rob
- 1,487
- 2
- 25
- 29