Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-07 Thread Namik Karovic
Hi Karl, Sounds good. Just to make sure we're taking about the same: 17:00 UTC+1 is 16:00 > UTC, which is 18:00 CEST? Yeah, sounds about right. =) Ragards, Namik On Thu, Aug 7, 2014 at 8:53 AM, Karl Rupp wrote: > Hi Namik, > > > > How's tomorrow at 17:00 UTC+1 ? > > Sounds good. Just to mak

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-06 Thread Karl Rupp
Hi Namik, > How's tomorrow at 17:00 UTC+1 ? Sounds good. Just to make sure we're taking about the same: 17:00 UTC+1 is 16:00 UTC, which is 18:00 CEST? Best regards, Karli > > On Wed, Aug 6, 2014 at 9:52 PM, Karl Rupp > wrote: > > Hi, > > > > @Nami

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-06 Thread Namik Karovic
Hi, How's tomorrow at 17:00 UTC+1 ? Regards, Namik On Wed, Aug 6, 2014 at 9:52 PM, Karl Rupp wrote: > Hi, > > > > @Namik: Can you please make the number of benchmark runs (i.e. the > >> preprocessor define BENCHMARK_RUNS) as well as the vector and matrix >> sizes (BENCHMARK_VECTOR

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-06 Thread Karl Rupp
Hi, > @Namik: Can you please make the number of benchmark runs (i.e. the > preprocessor define BENCHMARK_RUNS) as well as the vector and matrix > sizes (BENCHMARK_VECTOR_SIZE, BLAS3_MATRIX_SIZE) runtime parameters > and make them available through the GUI (expert tab)? Similarly,

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-06 Thread Namik Karovic
Hey, Just compute the memory consumption: The matrix has 10^8 rows, with 5 > entries per row. Each element takes 8 bytes of data in double precision, so > we need 4*10^9 bytes. In other words: 4 GB ;-) Damn. Time to get a new GPU. x) The respective sparse benchmarks don't include any runs of u

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-06 Thread Karl Rupp
Hi again, > argl, this is crappy uBLAS again. I'll kick this out now, it only > creates problems rather than solving them. here we go: https://github.com/viennacl/viennacl-benchmark-gui/commit/86461f18f1c13758ca13fb7fc4739bd6197c484f The respective sparse benchmarks don't include any runs of uB

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-06 Thread Karl Rupp
Hey, > After cutting the matrix size down to 1000 the program didn't crash, but > the procedure took way too long (since I'm using ublas). Cutting the > size to 100 makes it ok, but there's a different problem now. > > As soon as the benchmark starts doing tests with the generated matrix, > it cr

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-06 Thread Karl Rupp
Hi, > whoops, that results in a matrix with 10^8 (=10k*10k) rows and > columns - just too big for GPU RAM. > > > Is it really that big? My GPU has 3GB of RAM, I can't believe 10^8 > matrix entries is too much for it. Just compute the memory consumption: The matrix has 10^8 rows, with 5

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-05 Thread Namik Karovic
A quick update on the matrix generation: After cutting the matrix size down to 1000 the program didn't crash, but the procedure took way too long (since I'm using ublas). Cutting the size to 100 makes it ok, but there's a different problem now. As soon as the benchmark starts doing tests with the

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-05 Thread Namik Karovic
Hey Karl, whoops, that results in a matrix with 10^8 (=10k*10k) rows and columns - > just too big for GPU RAM. Is it really that big? My GPU has 3GB of RAM, I can't believe 10^8 matrix entries is too much for it. Yes, this makes sense. It's enough to use the same value for xPoints and > yPoint

Re: [ViennaCL-devel] Benchmark Matrix Usage Questions

2014-08-05 Thread Karl Rupp
Hi Namik, > I've just attempted to use Karl's on-the-fly generation in the Benchmark > GUI. Here's what I tried: > >viennacl::vcl_size_t xPoints = 1; >viennacl::vcl_size_t yPoints = 1; whoops, that results in a matrix with 10^8 (=10k*10k) rows and columns - just too big for GPU