Skip to content
This repository has been archived by the owner on Aug 5, 2022. It is now read-only.

why HD Graphics 630 run classification slower than i5 7500 CPU? #279

Open
mahaoyanghb opened this issue May 7, 2019 · 0 comments
Open

Comments

@mahaoyanghb
Copy link

mahaoyanghb commented May 7, 2019

my environment:
windows 10 enterprise
cpu: i5 7500 3.4GHz 4Core
graphics: HD Graphics 630

I download this code ,and compile it by instruction in this website.
I add some code to measure forward time by clock() function.
clock_t start, finish;

start = clock();
for (int i = 0; i < 10;i++)
{
classifier.Classify(img);
}
finish = clock();
std::cout << "cost " << difftime(finish,start)/10<<" ms"<< std::endl;

after I compared GPU mode and CPU mode, I found that classification.exe runs 10 times slower in graphics mode than in CPU mode .
CPU : 8ms
HD Graphics: 80ms

HD graphics card should run faster than CPU , shouldn't it ?
why it runs much more slow?

remark:
OpenCL is downloaded from intel
I have checked caffe recognize the true opencl dir.
the only change is I remove the HAS_HALF_SUPPORT definition , because clblas.h don't have half precision functions.

does anybody know the reason?
Thank you very much!

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant