Re: [luau] hyperthreading facts
I don't know much about Hyperthreading, but do the applications need to be specially optimized and/or recompiled for you to get performance increases, or will standard applications see these types of improvements just due to the CPU alone? Thanks! Rob --- "R. Scott Belford" <[EMAIL PROTECTED]> wrote: > http://www-106.ibm.com/developerworks/linux/library/l-htl/?ca=dgr-lnxw01HyperThread#resources > > "Intel Xeon Hyper-Threading is definitely having a positive impact > on Linux > kernel and multithreaded applications. The speed-up from > Hyper-Threading > could be as high as 30% in stock kernel 2.4.19, to 51% in kernel > 2.5.32 due > to drastic changes in the scheduler run queue's support and > Hyper-Threading > awareness. ..." > > This article was linked from Slashdot. > > scott > ___ > LUAU mailing list > [EMAIL PROTECTED] > http://videl.ics.hawaii.edu/mailman/listinfo/luau __ Do you Yahoo!? Yahoo! Mail Plus - Powerful. Affordable. Sign up now. http://mailplus.yahoo.com
Re: [luau] hyperthreading facts
Robert Green wrote: I don't know much about Hyperthreading, but do the applications need to be specially optimized and/or recompiled for you to get performance increases, or will standard applications see these types of improvements just due to the CPU alone? Thanks! Rob Anything that would benefit from SMP can potentially benefit from HyperThreading. In other words, if your app is threaded (and the threads are doing different things, using different functions of the CPU), it can benefit. You can also benefit if you're running more than one process (you are). The kernel sees the HyperThreaded CPU as 2 CPUs and schedules between them appropriately. However, the 2.5 series that's HyperThreading aware knows the limitations and benefits of hyperthreading (shared cache, etc). Therefore, it will schedule according to this (especially useful in situations where you have more than one HyperThreading CPU). --MonMotha -- Optimist: The glass is half full. | PGP Key: 0x1B0390E0 Pessimist: The glass is half empty.| Outgoing mail signed Engineer: The glass is twice as big as it needs to be. | [EMAIL PROTECTED] pgp62z5bIrAeN.pgp Description: PGP signature
RE: [luau] hyperthreading facts
YES!! Ask Brian Low also. We are putting together P4-2.53 with Intel boards with HT that do the job regardless What else you may hear. -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of Robert Green Sent: Tuesday, January 14, 2003 10:54 AM To: [EMAIL PROTECTED] Subject: Re: [luau] hyperthreading facts I don't know much about Hyperthreading, but do the applications need to be specially optimized and/or recompiled for you to get performance increases, or will standard applications see these types of improvements just due to the CPU alone? Thanks! Rob --- "R. Scott Belford" <[EMAIL PROTECTED]> wrote: > http://www-106.ibm.com/developerworks/linux/library/l-htl/?ca=dgr-lnxw01Hype rThread#resources > > "Intel Xeon Hyper-Threading is definitely having a positive impact > on Linux > kernel and multithreaded applications. The speed-up from > Hyper-Threading > could be as high as 30% in stock kernel 2.4.19, to 51% in kernel > 2.5.32 due > to drastic changes in the scheduler run queue's support and > Hyper-Threading > awareness. ..." > > This article was linked from Slashdot. > > scott > ___ > LUAU mailing list > [EMAIL PROTECTED] > http://videl.ics.hawaii.edu/mailman/listinfo/luau __ Do you Yahoo!? Yahoo! Mail Plus - Powerful. Affordable. Sign up now. http://mailplus.yahoo.com ___ LUAU mailing list [EMAIL PROTECTED] http://videl.ics.hawaii.edu/mailman/listinfo/luau
Re: [luau] hyperthreading facts
--- MonMotha <[EMAIL PROTECTED]> wrote: > Robert Green wrote: > > I don't know much about Hyperthreading, but do the applications > need > > to be specially optimized and/or recompiled for you to get > > performance increases, or will standard applications see these > types > > of improvements just due to the CPU alone? > > > > Thanks! > > > > Rob > > Anything that would benefit from SMP can potentially benefit from > HyperThreading. In other words, if your app is threaded (and the > threads are > doing different things, using different functions of the CPU), it > can benefit. > > You can also benefit if you're running more than one process (you > are). The > kernel sees the HyperThreaded CPU as 2 CPUs and schedules between > them > appropriately. However, the 2.5 series that's HyperThreading aware > knows the > limitations and benefits of hyperthreading (shared cache, etc). > Therefore, it > will schedule according to this (especially useful in situations > where you have > more than one HyperThreading CPU). > > --MonMotha I am primarily concerned with applications like MUDs (chatroom like envirmonment) and web servers - I take it these would get an excellent benefit from Hyperthreading, then? __ Do you Yahoo!? Yahoo! Mail Plus - Powerful. Affordable. Sign up now. http://mailplus.yahoo.com
RE: [luau] hyperthreading facts
THANK YOU SCOTT THAT'S WHAT IVE BEEN STATING ALL ALONG AND GETTING NEGATIVE FEEDBACK. Some people are so opinionated that they don't read the facts. I have been using this for over a month with even better results. Dan -Original Message- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of R. Scott Belford Sent: Tuesday, January 14, 2003 10:00 AM To: [EMAIL PROTECTED] Subject: [luau] hyperthreading facts http://www-106.ibm.com/developerworks/linux/library/l-htl/?ca=dgr-lnxw01Hype rThread#resources "Intel Xeon Hyper-Threading is definitely having a positive impact on Linux kernel and multithreaded applications. The speed-up from Hyper-Threading could be as high as 30% in stock kernel 2.4.19, to 51% in kernel 2.5.32 due to drastic changes in the scheduler run queue's support and Hyper-Threading awareness. ..." This article was linked from Slashdot. scott ___ LUAU mailing list [EMAIL PROTECTED] http://videl.ics.hawaii.edu/mailman/listinfo/luau
Re: [luau] hyperthreading facts
LinuxDan wrote: THANK YOU SCOTT THAT'S WHAT IVE BEEN STATING ALL ALONG AND GETTING NEGATIVE FEEDBACK. Some people are so opinionated that they don't read the facts. I have been using this for over a month with even better results. Dan You were talking about 200% performance improvements which I said rightly was impossible. Hyperthreading does not ever approach SMP in performance gains. I said that certain things will go slower with Hyperthreading which the benchmarks do show, but other things do indeed go faster. I had not seen 2.5.x kernel benchmarks before, but yes those numbers do indeed indicate more significant gains. Warren
RE: [luau] hyperthreading facts
>>You were talking about 200% performance improvements which I said >>rightly was impossible. Hyperthreading does not ever approach SMP in >>performance gains. >> >>I said that certain things will go slower with Hyperthreading which the >>benchmarks do show, but other things do indeed go faster. I had not >>seen 2.5.x kernel benchmarks before, but yes those numbers do indeed >>indicate more significant gains. >> >>Warren I concur. I was not disagreeing with you LinuxDan but never seen any benchmarks that established a 200% improvement. HT also makes certain apps run slower from the benchmarks I've read. I apologize if you thought I was giving opinionated statements, guess I should have stated some facts. Randall