Hi
Could anybody help me the following?
I have a vector v , which is 1 row by 100 columns.
I want to extract 10 of 1 row by 10 columns vectors v1, v2, .., v10 from v.
And for each vector v1, v2, ..., v10, I would like to apply magma_sgemm function simultaneously.
How could I extract the vectors from the original vector v efficiently using MAGMA function?
How could I design the parallel magma_sgemm call for each vectors?
I am using CUDA version of MAGMA.
Any help is much appreciated.
Thank you,
-Hiroki
Parallel MAGMA function call
Re: Parallel MAGMA function call
Can you specify more details about the next magma_sgemm call? Why are you using gemm for matrix vector multiplication?
What do you mean by extracting vectors? You could just use the starting pointer eg. v1 = v[0], v2 = v[10], .... and length = 10 to specify each vector.
What do you mean by extracting vectors? You could just use the starting pointer eg. v1 = v[0], v2 = v[10], .... and length = 10 to specify each vector.
Re: Parallel MAGMA function call
Thank you for your reply, hsahasra.
Yes you are right.
But please forget about this post for now.
I need to organize my thought.
But thank you.
-Hiroki
Yes you are right.
But please forget about this post for now.
I need to organize my thought.
But thank you.
-Hiroki