topic Re:Vtune question for memory bound problem on GPU in Analyzers

Vtune question for memory bound problem on GPU

LaurentPlagne — Mon, 17 Aug 2020 08:26:16 GMT

Hi,

I wonder about the vtune diagnostic for memory bound problem on GPU.

I measure the observed bandwidth of a vector kernel (MemoryBoundKernel.hpp) for large (2>>27) vector of floats a and b :

h.parallel_for(global_range, [=](id<1> i) { acc_b[i]+=acc_a[i]; }); }); // end submit

I obtain 19 GB/s on my laptop (9300H with UHD630) which I suspect to be close to the maximal bandwidth on this machine.

What I found surprising is that the vtune GPU analysis emphasis (in red) on EU occupancy and does not (not in red) emphasis on the RAM bandwidth saturation.

Do I miss something obvious ?

Re: Vtune question for memory bound problem on GPU

GouthamK_Intel — Tue, 18 Aug 2020 05:30:59 GMT

Hi,

Thanks for reaching out to us!

As your query is related to Vtune, we are redirecting your post to the Vtune forum so that Vtune experts can guide you better.

Thanks & Regards

Goutham

Re: Vtune question for memory bound problem on GPU

Plagne__Laurent — Wed, 19 Aug 2020 06:10:04 GMT

Self answer : I have replaced my kernel (20 GB/s)

q.submit([&](auto &h) {// Submit command group for execution auto acc_a = buf_a.template get_access<access::mode::read>(h);// Create accessors auto acc_b = buf_b.template get_access<access::mode::write>(h); auto global_range = range<1>(vsize);// Define local and global range h.parallel_for(global_range, [=](id<1> i) { acc_b[i]+=alpha*acc_a[i]; }); }); // end submit

by oneAPI MKL axpy (26 GB/s)

mkl::blas::axpy(q, vsize, alpha, buf_a, 1, buf_b, 1);

and now vtune correctly emphasizes (in red) the DRAM bandwidth bound (83.5 %).

Although I don't know how my kernel could be enhanced, Vtune correctly identifies that there was room for improvement.

BTW, what does exactly mean the figure 83.5% ?

Re: Vtune question for memory bound problem on GPU

Plagne__Laurent — Wed, 19 Aug 2020 06:27:19 GMT

The metric description is accurately described in the doc (except on how the default threshold values (low-medium-high) are computed.

https://software.intel.com/content/www/us/en/develop/documentation/vtune-help/top/reference/cpu-metrics-reference/memory-bound/dram-bound/dram-bandwidth-bound.html

Re:Vtune question for memory bound problem on GPU

Adweidh_Intel — Wed, 19 Aug 2020 06:33:02 GMT

Hi Plagne,

We are checking on this with our SME, will get back to you soon.

Thanks,

Adweidh

Re: Vtune question for memory bound problem on GPU

Kevin_O_Intel1 — Wed, 19 Aug 2020 16:44:45 GMT

The low medium and high thresholds are just the default values. You can change these by moving the sliders at the bottom of the graph. The defaults are evenly distributed.

Re: Vtune question for memory bound problem on GPU

LaurentPlagne — Thu, 20 Aug 2020 06:51:18 GMT

Thank you !

BTW do you know how the max system GPU bandwidth is evaluated ?

I use the GPU version of oneMKL saxpy (on 2<<27 sized arrays) and obtain 26 GB/s on my laptop (repeated 100 times to eliminate the device/host communication) while vtune put the default max gpu bandwidth to 35 GB/s.

Is is OK to assume that oneMKL saxpy should saturate the available bandwidth ?