"Lance Palmer" <lance.palmer at="" stonybrook.edu=""> writes: > I will get to purchase a new computer and have about $3500. Ijust wanted > some advice on a couple options. Ideally I would want something that would > be optimal for things such as Bioconductor and Genespring. > > > > In general, I will probably get a system with 4G of RAM and will probably > use the 32 bit Windows XP Pro OS. Are you sure that Windows will be able to use that 4GB of RAM? If you read posts to this list and to R-help you will see many threads discussing how to get a Windows system to access more than 1.5GB or so. No such threads for Linux; its memory system is much more useful for scientific computing. > If I get the systems lower on the list I could get a 256MB graphics card, > instead of 128. How big of a difference will this make compared to the > difference in processor power? Also I could get two monitors with some of > the cheaper processors towards the bottom. > > > > Also I believe the systems towards the top have more maximum memory > available (if I go to 64 bit system) > > > > Any thoughts? Go with a 64-bit system. Consider running a 64-bit flavor of Linux. Graphics card memory will almost certainly be irrelevant. + seth

> Most computer purchases here are through Dell, but I may be able to go > through HP if I wanted to get an AMD. > > > > Would there be a tremendous amount of difference in computing power for > things like Bioconductor and Genespring between, say > > > > Dual core Xeon (2.8 GHz) > > Dual processor Opteron 248 (2.2 GHz) > > Pentium Extreme 955EE(3.4 GHz) > > Dual Processor Xeon (2.8 GHz) > > Pentium D 940 (3.2 GHz) > > > > The pentiums have 2x2Mb L2 Cache while the opteron is 2x1Mb > > You might want to consider waiting a few months, if you can. Performance-wise, right now is not the most advantageous time to buy and any top of the line system you build or order will be easily outperformed in a few months. In a few months, Intel will introduce its Conroe-based processors which, in pre-production tests, have been shown to significantly outperform other current and upcoming chips (both Intel's own and the competition). With PC hardware, there's always something better just around the corner, but this move represents much more than the typical incremental "speed bumps". A good preview look at Conroe and Intel's new "Core" architecture: http://www.anandtech.com/tradeshows/showdoc.aspx?i=2713

Seth, Can you explain what effect that the Conroe-based processor will have on performance in Bioconductor? Thanks, Lana >You might want to consider waiting a few months, if you can. >Performance-wise, right now is not the most advantageous time to buy and >any top of the line system you build or order will be easily >outperformed in a few months. In a few months, Intel will introduce its >Conroe-based processors which, in pre-production tests, have been shown >to significantly outperform other current and upcoming chips (both >Intel's own and the competition).

"Lana Schaffer" <schaffer at="" scripps.edu=""> writes: > Seth, > Can you explain what effect that the Conroe-based processor > will have on performance in Bioconductor? Improve it :-) I'm not a hardware expert and don't have any detailed knowledge of the conroe technology. In general, faster chips lead to faster computations. The notion of "performance in Bioconductor" is not all that meaningful; it really depends on what computations you are interested in. Hope that helps some, + seth

yodersj at moffitt.usf.edu said: > antyhing you do for an analysis workstation. I'm running a Dell Precision > 370 with an intel EM64T with 4 GB RAM. I went with 64 bit for future Windows > 64 uses, but with the intensions of running 64-bit linux now. We just > ordered another $2500 Dell with similar specs. I went with the dual core > processor since it was not an expensive upgrade, but I don't expect it to be > of much help...more of a marketing ploy. The 64-bit Windows (Server 2003) is working fine here; Let's you address more than 4GB RAM. We're running it on Opteron-based Sun servers. Another OS that does BioC well is Solaris-x86, in either 32- or 64-bit mode. Sun has long experience with getting the most out of multi-core/multi-CPU hardware, and Solaris-10 with its (now free) Sun compilers is known to get more performance out the the same hardware than some other OS's. Plus it's free. > You WILL NOT be able to access 4 GB of RAM in Windows for some reason...I can > only get it to recognize 3 GB, and as Seth mentions, most applications will > only access about 1.5GB of RAM at the most...despite what else you have going > on in the background or your Windows memory settings. There is an issue with the way the BIOS maps hardware (PCI) devices into the memory address space, which can make some of your RAM unavailable when you get close to the 4GB range. Google around for "BIOS memory hole"; Some modern BIOS's have ways around this. Regards, Marion

Seth is right. Unless you have other uses for heavy duty video, a cheapy card will do fine for GeneSpring and BioC as well as pretty much antyhing you do for an analysis workstation. I'm running a Dell Precision 370 with an intel EM64T with 4 GB RAM. I went with 64 bit for future Windows 64 uses, but with the intensions of running 64-bit linux now. We just ordered another $2500 Dell with similar specs. I went with the dual core processor since it was not an expensive upgrade, but I don't expect it to be of much help...more of a marketing ploy. You WILL NOT be able to access 4 GB of RAM in Windows for some reason...I can only get it to recognize 3 GB, and as Seth mentions, most applications will only access about 1.5GB of RAM at the most...despite what else you have going on in the background or your Windows memory settings. I'd recommend doing a dual boot or having switchable hard drives and booting into 64-bit Linux with 64-bit BioC for large analyses. I had a dual boot setup on this Dell with one HDD and the HDD crashed within a month and it was a lot of work to get things up and running again... Oh, and a dual monitor setup is great for miroarray or sequence analysis because you can have your code or your sequences on the extended desktop and you don't have to swap back and forth when you copy and paste, etc. I think having two 19" LCDs is much more valuable than upgrading from a 3.0 to 3.2 GHz CPU, for example...plus it makes you look really important :-) Good luck, Sean Sean Yoder, M.S. Research Associate Microarray Core Facility Moffitt Cancer Center & Research Institute E-mail: yodersj at moffitt.usf.edu Fax: 813-979-7265 Phone: 813-745-1917 -----Original Message----- From: bioconductor-bounces@stat.math.ethz.ch [mailto:bioconductor-bounces at stat.math.ethz.ch] On Behalf Of Seth Falcon Sent: Tuesday, April 25, 2006 10:57 AM To: bioconductor at stat.math.ethz.ch Subject: Re: [BioC] Thoughts on computer purchase "Lance Palmer" <lance.palmer at="" stonybrook.edu=""> writes: > I will get to purchase a new computer and have about $3500. Ijust wanted > some advice on a couple options. Ideally I would want something that would > be optimal for things such as Bioconductor and Genespring. > > > > In general, I will probably get a system with 4G of RAM and will probably > use the 32 bit Windows XP Pro OS. Are you sure that Windows will be able to use that 4GB of RAM? If you read posts to this list and to R-help you will see many threads discussing how to get a Windows system to access more than 1.5GB or so. No such threads for Linux; its memory system is much more useful for scientific computing. > If I get the systems lower on the list I could get a 256MB graphics card, > instead of 128. How big of a difference will this make compared to the > difference in processor power? Also I could get two monitors with some of > the cheaper processors towards the bottom. > > > > Also I believe the systems towards the top have more maximum memory > available (if I go to 64 bit system) > > > > Any thoughts? Go with a 64-bit system. Consider running a 64-bit flavor of Linux. Graphics card memory will almost certainly be irrelevant. + seth ----------------------------------------- ################################################################### ########## This transmission may be confidential or protected from disclosure and is only for review and use by the intended recipient. Access by anyone else is unauthorized. Any unauthorized reader is hereby notified that any review, use, dissemination, disclosure or copying of this information, or any act or omission taken in reliance on it, is prohibited and may be unlawful. If you received this transmission in error, please notify the sender immediately. Thank you. ################################################################### ##########

Yes, the 64-bit version of Windows would not have a memory problem. However, the original poster is planning to use 32-bit Windows XP. He's also ordering a Dell, so I'd imagine he'll run into the same problems that I have on my Dell... Thanks for the info on the bios...I've tried a few things in the past to address the problem, but had no resolution. -----Original Message----- From: Marion Hakanson [mailto:hakansom@ohsu.edu] Sent: Friday, April 28, 2006 1:23 PM To: Yoder, Sean J. Cc: bioconductor at stat.math.ethz.ch Subject: Re: [BioC] Thoughts on computer purchase The 64-bit Windows (Server 2003) is working fine here; Let's you address more than 4GB RAM. We're running it on Opteron-based Sun servers. Another OS that does BioC well is Solaris-x86, in either 32- or 64-bit mode. Sun has long experience with getting the most out of multi-core/multi-CPU hardware, and Solaris-10 with its (now free) Sun compilers is known to get more performance out the the same hardware than some other OS's. Plus it's free. > You WILL NOT be able to access 4 GB of RAM in Windows for some reason... There is an issue with the way the BIOS maps hardware (PCI) devices into the memory address space, which can make some of your RAM unavailable when you get close to the 4GB range. Google around for "BIOS memory hole"; Some modern BIOS's have ways around this. Regards, Marion ----------------------------------------- ################################################################### ########## This transmission may be confidential or protected from disclosure and is only for review and use by the intended recipient. Access by anyone else is unauthorized. Any unauthorized reader is hereby notified that any review, use, dissemination, disclosure or copying of this information, or any act or omission taken in reliance on it, is prohibited and may be unlawful. If you received this transmission in error, please notify the sender immediately. Thank you. ################################################################### ##########