Search found 11 matches

by mtacconi
Wed Feb 29, 2012 12:53 pm
Forum: User discussion
Topic: Error making libmorse_quark
Replies: 3
Views: 1581

Re: Error making libmorse_quark

by using the plasma 2.4.0 I was able to work around the "CORE_zaxy_quark undefined" thing. It seems that the support to the zaxy family of BLAS routines simply disappears from the plasma-2.4.5 release. However plasma-2.4.0 has those routines and by using it I finally got the libmagma_mgpu.a and libm...
by mtacconi
Wed Feb 29, 2012 7:02 am
Forum: User discussion
Topic: Error making libmorse_quark
Replies: 3
Views: 1581

Re: Error making libmorse_quark

I am experimenting exactly the same problem. It seems to me that my installation of plasmalib is somewhat broken as the core_zblas.h header doesn't contain the symbol CORE_zaxpy_quark. Further investigations with nm of the .a files shows also that that symbol is not defined within the whole plasma l...
by mtacconi
Tue Mar 15, 2011 9:14 am
Forum: User discussion
Topic: magma_dsytrd questions
Replies: 25
Views: 8879

Re: magma_dsytrd questions

I don't believe it is all about the cublasAlloc: as far as I saw from my tests on the magma_dsytrd you can obtain about 40 GFlop/s (asymptotic) from a Tesla C2050 if: 1- call magmablas_dsymv6_fermi instead of the cublasDsymv 2- use magmablas_dsyr2k (you can obtain this routine from the single precis...
by mtacconi
Thu Feb 24, 2011 1:05 pm
Forum: User discussion
Topic: magma_dsytrd questions
Replies: 25
Views: 8879

Re: magma_dsytrd questions

Following the suggestion made by Stan, I modified the dsytrd routine to call the magma_dsymv or the "expert driver" magma_dsymv6_fermi which needs some basic GPU memory management. Here the results: MKL_NUM_THREADS=1 cublasDsymv: M N CPU GFlop/s CPU etime GPU GFlop/s GPU etime ======================...
by mtacconi
Wed Feb 23, 2011 12:20 pm
Forum: User discussion
Topic: magma_dsytrd questions
Replies: 25
Views: 8879

Re: magma_dsytrd questions

[...] probably in the paper they compare 1 CPU core/thread against 1 GPU. [...] Anyway, are you using a multithreaded LAPACK library CPU side? The paper claims they are using "MKL's parallel BLAS" with "MKL 10.0". I'm using MKL as well. I'd think that hindering the multi-core BLAS would hurt the GP...
by mtacconi
Wed Feb 23, 2011 11:32 am
Forum: User discussion
Topic: magma_dsytrd questions
Replies: 25
Views: 8879

Re: magma_dsytrd questions

The paper also claims a worse GPU (their GTX280 vs my C2050) and a better CPU (their Xeon vs my Desktop PC). So I'm not sure what I'm doing wrong! Perhaps the numbers in the paper were theoretical throughput but mistakenly presented as observed results? probably in the paper they compare 1 CPU core...
by mtacconi
Wed Feb 23, 2011 7:51 am
Forum: User discussion
Topic: magma_dsytrd questions
Replies: 25
Views: 8879

Re: magma_dsytrd questions

I'm seeing similar results. I had to edit your example, though, removing the d_work parameter. magma_dsytrd(uplo, N, h_R, lda, h_DR, h_ER, h_TAUR, h_WORKR, &LWORKR, d_work, &info); to magma_dsytrd(uplo, N, h_R, lda, h_DR, h_ER, h_TAUR, h_WORKR, &LWORKR, &info); I had modified the magma_dsytrd routi...
by mtacconi
Tue Feb 22, 2011 11:38 am
Forum: User discussion
Topic: magma_dsytrd questions
Replies: 25
Views: 8879

Re: magma_dsytrd questions

The use is as in testing_dsytrd.cpp I could not find this file in the testing directory. However I ran some test on the dsytrd subroutine using the following code: * @the testing_dgetrf source code has been used as a template. * **/ // includes, system #include <stdlib.h> #include <stdio.h> #includ...
by mtacconi
Fri Dec 10, 2010 9:27 am
Forum: User discussion
Topic: MAGMA 1.0
Replies: 16
Views: 19127

Re: MAGMA 1.0

Good news, really!
It is going to be a happy number-crunching holyday season :)
by mtacconi
Thu Dec 09, 2010 6:13 am
Forum: User discussion
Topic: MAGMA 1.0
Replies: 16
Views: 19127

Re: MAGMA 1.0

It seems that some important (at least for me :) ) routines of the 0.2 release (xGEHRD for example) haven't been included in the 1.0RC1. Do you plan to include again the xGEHRD in the final 1.0 release? I am also looking forward to experiment with the symmetric eigensolver, I wonder if we have any c...