Search found 14 matches

by Allan Menezes
Tue Sep 14, 2010 12:07 am
Forum: User discussion
Topic: MAGMA 0.3 for Fermi and CUDA 3.2 RC2 results
Replies: 1
Views: 1793

MAGMA 0.3 for Fermi and CUDA 3.2 RC2 results

Here are the results of Magma 0.3 for Fermi versus Cublas 3.2 RC2 on a GTX-480: This are MAGMA 0.3 DGEMM and SGEMM Routines for Fermi GPUs. In this version matrix sizes have to be divisible by 64 Usage: ./testing_dgemm N N magmablas0.3 GFLops/s cudablas-3.2 GFlops/s error ===========================...
by Allan Menezes
Sun Sep 12, 2010 11:51 pm
Forum: User discussion
Topic: MAGMA GEMM Sources for Fermi Released
Replies: 21
Views: 12847

Re: MAGMA GEMM Sources for Fermi Released

Dear Stan, As this is just pointer arithmetic and used in only a few places it does not change the perfomance much at all as per my experiment below. Just for fun I changed fermi_dgemm.cu and fermi_sgemm.cu with a single #define on top as #define __mul24(a,b) ((a)*(b)) and there was no significant d...
by Allan Menezes
Tue Jul 20, 2010 7:52 am
Forum: User discussion
Topic: CPU vs GPU speed
Replies: 5
Views: 5659

Re: CPU vs GPU speed

Dear Gaurav, Here are the specs of the Tesla C1060 at URL http://www.nvidia.com/object/product_tesla_c1060_us.html Here are the specs of the Intel E5530 at URL http://ark.intel.com/Product.aspx?id=37103 As you can see it can perform at only 78 GFlops double precision peak maximum and you are getting...
by Allan Menezes
Fri Jul 16, 2010 10:46 pm
Forum: User discussion
Topic: CPU vs GPU speed
Replies: 5
Views: 5659

Re: CPU vs GPU speed

Can you please specify in the second and first tables how many CPUs and which ones (INTEL VS AMD and number of cores) and the number and type of GPUs?
Allan
by Allan Menezes
Sat May 22, 2010 3:33 am
Forum: User discussion
Topic: magma_sgetrf_gpu seg faults
Replies: 1
Views: 8790

Re: magma_sgetrf_gpu seg faults

Hi, See my post for the results of tests I ran with a recompiled version of magma 0.2 source with CUDA3.0 and the GTX480 which is similar to the FERMI Tesla C2050. device 0: GeForce GTX 480, 1401.0 MHz clock, 1535.2 MB memory device 1: GeForce GTX 470, 1215.0 MHz clock, 1279.7 MB memory Usage: testi...
by Allan Menezes
Tue Apr 20, 2010 8:18 pm
Forum: User discussion
Topic: Magma 0.2 with Nvidia Fermi GTX470 and GotoBLAS2 results
Replies: 5
Views: 16345

Re: Magma 0.2 with Nvidia Fermi GTX470 and GotoBLAS2 results

Dear Stan, Thank you for your response! Wow 300GFlops/s sounds interesting. But one stupid question: Is it double precision? And you probably would not get it with the GTX480 just the Tesla versions of Fermi. I tried in zgetrf.cpp and zgetrf_gpu.cpp adding at the beginning as you suggested:#define m...
by Allan Menezes
Tue Apr 20, 2010 1:42 am
Forum: User discussion
Topic: Magma 0.2 with Nvidia Fermi GTX470 and GotoBLAS2 results
Replies: 5
Views: 16345

Magma 0.2 with Fermi GTX480, GTX470 and GotoBLAS2 results

Dear All, The makefiles are the same and the cuda toolkit and drivers are rev level 3.0 as per my previous post with the GTX470 and GTX260 results. Below is the performance of the NVIDIA Fermi GTX480 as device 0 and GTX470 as device 1 with CUDA 3.0 and all else same as above as it was run and is pre...
by Allan Menezes
Sat Apr 17, 2010 5:31 pm
Forum: User discussion
Topic: Magma 0.2 with Nvidia Fermi GTX470 and GotoBLAS2 results
Replies: 5
Views: 16345

Re: Magma 0.2 with the Nvidia GTX260,and GotoBLAS2 results

Dear All, The makefiles are the same and the cuda toolkit and drivers are rev level 3.0 as per my previous post with the GTX470 results. Below is the performance of the NVIDIA GTX260 with CUDA 3.0 and all else same as above of compute level 1.3 of magma 0.2 as it was run and is presented as is: I th...
by Allan Menezes
Sat Apr 17, 2010 1:52 pm
Forum: User discussion
Topic: Magma 0.2 with Nvidia Fermi GTX470 and GotoBLAS2 results
Replies: 5
Views: 16345

Magma 0.2 with Nvidia Fermi GTX470 and GotoBLAS2 results

Dear All, I tried magma 0.2 recompiled by me for fedora core 12 x86_64 with the nvidia fermi GTX470 and here are the make.inc.goto and the results from the testing directory. Note that in the following make.inc.goto the place of the libgoto2.a is hard coded. You will have to modify that line for you...
by Allan Menezes
Mon Mar 29, 2010 6:01 pm
Forum: User discussion
Topic: MAGMA for NVIDIA FERMI
Replies: 1
Views: 9131

MAGMA for NVIDIA FERMI

Dear Stan,
When do you propose to add support for the new Nvidia GPU offerings such as the GTX 470,480 and CUDA 3.0, Please?
What about the corresponding AMD offerings with stream software?
Thank you,
Best wishes,
Allan Menezes