Performance scale of the Intel Phi MIC


The attached is plot of execution time on Intel Phi with varying number of threads. The same program runs in native and offload modes.

The Phi device has 60 cores.

1) Why the timing steps don't occur at multiples of number of cores (i.e., multiple of 60s)?

2) Why the time drops substantially around 248 threads and increases again? (i.e., > 4x60)

adding offload pragma , performance drops

Hello ,

I am running a code in openMP which is like this:


#pragma omp parallel for default( none ) shared( X , Y ,V ,V ,H , W ,N ) private ( i,x,y ,Kx,Ky,initD ,T ) 

		for ( y = 0; y < H; y++ )
			for ( x = 0; x < W; x++ )

				initD = aValue;
				for ( i = 0; i < N; i++ )
				V[ x + y * Width ] = T;


Now , I want to run it on mic card , so when I just add the line:

internal error: bad pointer

my code is this: ---------------------------------------------------------------------------------------------- #include class TEST{ public: double *A; public: TEST(double * _A){ A = _A; #pragma offload_transfer target(mic:0) nocopy(this : alloc_if(1) free_if(0)) in(A:length(2*3) alloc_if(1) free_if(0)) } void run(){ A[1] = 0; // double *B = A; std::cout<

timing is different each time

Hello ,

I wrote a simple application on cpu and I am using offload pragmas for the pieces I want to run on the coprocessors.

Since I am compiling on cpu and I use offloads , I am using :

<code>export MIC_ENV_PREFIX=MIC

in order to specify the threads number.

My problems:

1) Running the code , shows always 40 threads been used.

2) Running again and again the code without compiling , I am getting different time results.

Videos - Parallel Programming with Intel Xeon Phi Coprocessors

Here is a list of recently published videos from Colfax International on Intel(R) Xeon Phi(TM) Coprocessors.

In this video we will discuss software tools needed and recommended for developing applications for Intel Xeon Phi coprocessors. We will begin with software that is necessary to boot coprocessors and to run pre-compiled executables on them.

Iscriversi a Azienda