Re: [R] R performance: different CPUs
Hi, 64bit CPUs, such as opterons, help significantly with large databases or if you are running multiple processes. But there is a speed penalty if you are not. Some packages can make use of multiple processors, such as my rgenoud (genetic optimization using derivatives) and Matching packages, but most do not. For these packages the speed up is significant. There are also multithreaded BLAS which can be used reliably under LINUX, but the speed benefit is usually small. You may want to check out some benchmarks at: http://sekhon.berkeley.edu/macosx/ (linux does very well). Cheers, JS. === Jasjeet S. Sekhon Associate Professor Survey Research Center UC Berkeley http://sekhon.berkeley.edu/ V: 510-642-9974 F: 617-507-5524 === Toby Muhlhofer writes: Hello! I need to purchase a new box, which I would like to optimize for good R performance. For the record, I will run Fedora Core 5 as and OS, and I wanted to know if anyone has experience with how the following affects R performance: - Is there a big advantage to having a 64-bit CPU over having a 32-bit? - Does an Opteron offer any advantages over an Athlon, and if yes, does it justify an investment of about US $75 more for equivalent listed speeds? - Have people successfully multithreaded R computations, such as to justify a dual-core CPU? I understand R itself does not multithread, but of course it should be possible to write code that paralellizes computations and I wanted to know if anyone has experience doing so and gained large speed advantages by it. Thanks, Toby Muhlhfoer __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R performance: different CPUs
On Wed, 5 Apr 2006, Jasjeet Singh Sekhon wrote: Hi, 64bit CPUs, such as opterons, help significantly with large databases or if you are running multiple processes. But there is a speed penalty if you are not. This would be true of 64-bit builds of R, not 64-bit CPUs. On a 64-bit processor you can usually run either 64-bit or 32-bit builds of R, and the 64-bit one will be able to access more memory but will be slower. This doesn't mean that a 32-bit build of R on a 64-bit processor will be slower than a 32-bit build of R on a 32-bit processor. -thomas Thomas Lumley Assoc. Professor, Biostatistics [EMAIL PROTECTED] University of Washington, Seattle __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R performance: different CPUs
This would be true of 64-bit builds of R, not 64-bit CPUs. [...] This doesn't mean that a 32-bit build of R on a 64-bit processor will be slower than a 32-bit build of R on a 32-bit processor. There is the issue, however, of running a 32bit application on a 64bit OS. Under RedHat and SuSE this works transparently and I've not noticed a performance issue, but under Debian's or Ubuntu's chroot setup there is in my experience a measurable performance hit. Of course, one could simply run 32bit Debian along with 32bit R on an Opteron. Cheers, Jas. === Jasjeet S. Sekhon Associate Professor Travers Department of Political Science Survey Research Center UC Berkeley http://sekhon.berkeley.edu/ V: 510-642-9974 F: 617-507-5524 === Thomas Lumley writes: On Wed, 5 Apr 2006, Jasjeet Singh Sekhon wrote: Hi, 64bit CPUs, such as opterons, help significantly with large databases or if you are running multiple processes. But there is a speed penalty if you are not. This would be true of 64-bit builds of R, not 64-bit CPUs. On a 64-bit processor you can usually run either 64-bit or 32-bit builds of R, and the 64-bit one will be able to access more memory but will be slower. This doesn't mean that a 32-bit build of R on a 64-bit processor will be slower than a 32-bit build of R on a 32-bit processor. -thomas Thomas LumleyAssoc. Professor, Biostatistics [EMAIL PROTECTED]University of Washington, Seattle __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
Re: [R] R performance: different CPUs
From: Jasjeet Singh Sekhon This would be true of 64-bit builds of R, not 64-bit CPUs. [...] This doesn't mean that a 32-bit build of R on a 64-bit processor will be slower than a 32-bit build of R on a 32-bit processor. There is the issue, however, of running a 32bit application on a 64bit OS. Under RedHat and SuSE this works transparently and I've not noticed a performance issue, but under Debian's or Ubuntu's chroot setup there is in my experience a measurable performance hit. Of course, one could simply run 32bit Debian along with 32bit R on an Opteron. It's my recollection that about a year ago there was still fierce debate about Debian on x86_64: One group got a pure 64-bit version to a usable state sooner than another group that was trying to make a standard conforming version that would run both 64- and 32-bit apps transparently. I don't know how that was resolved in the end, but that basically was enough for me to steer clear of any Debian-based distro for x86_64. Andy Cheers, Jas. === Jasjeet S. Sekhon Associate Professor Travers Department of Political Science Survey Research Center UC Berkeley http://sekhon.berkeley.edu/ V: 510-642-9974 F: 617-507-5524 === Thomas Lumley writes: On Wed, 5 Apr 2006, Jasjeet Singh Sekhon wrote: Hi, 64bit CPUs, such as opterons, help significantly with large databasesor if you are running multiple processes. But there is a speedpenalty if you are not. This would be true of 64-bit builds of R, not 64-bit CPUs. On a 64-bit processor you can usually run either 64-bit or 32-bit builds of R, and the 64-bit one will be able to access more memory but will be slower. This doesn't mean that a 32-bit build of R on a 64-bit processor will be slower than a 32-bit build of R on a 32-bit processor. -thomas Thomas Lumley Assoc. Professor, Biostatistics [EMAIL PROTECTED] University of Washington, Seattle __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html __ R-help@stat.math.ethz.ch mailing list https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
