tag:blogger.com,1999:blog-7434008.post111661322022745674..comments2023-11-02T09:04:29.998-07:00Comments on Adrian Cockcroft's Blog: Performance monitoring with HyperthreadingAdrian Cockcrofthttp://www.blogger.com/profile/14695336135416848505noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-7434008.post-74987797938156662852009-06-16T22:37:20.728-07:002009-06-16T22:37:20.728-07:00Americans everywhere humor A detention wow gold n...Americans everywhere humor A detention <a href="http://www.wowgold-wow.com/" rel="nofollow">wow gold</a> notice was written like this: a <a href="http://www.wowgold-powerleveling.com/" rel="nofollow">wow power leveling</a> police car with stones, to win <a href="http://www.cheap-lotrogold.com/" rel="nofollow">wow gold</a> the detention center for seven <a href="http://www.wowpowerleveling.me/" rel="nofollow">wow power leveling</a> days all-inclusive accommodation <a href="http://www.watchrolexshop.com/" rel="nofollow">replica rolex</a> Tour Value; hit send 2 a beautiful bracelet, <a href="http://www.wowpowerleveling.me/" rel="nofollow">wow power level</a> fashionsuit, police transport; more more surprises , the former can enjoy free shaved 10; before the 100 can play with <a href="http://www.wowgold-powerleveling.com/" rel="nofollow">power leveling</a> the dogs, the guests were presented massage sticks, electric shocks to CHEAPEST <a href="http://www.wowpowerleveling.me/" rel="nofollow">power leveling</a> the dead skin beauty care services.<br />asdasdasdAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-7434008.post-1119566366220807482005-06-23T15:39:00.000-07:002005-06-23T15:39:00.000-07:00Great post!Some comments:You wrote:"Hyperthreading...Great post!<BR/><BR/>Some comments:<BR/>You wrote:<BR/>"Hyperthreading should be disabled for anything other than Windows XP"<BR/>->actually, Windows 2003 uses the same hyper-threading-aware logic as XP, so it is an excellent choice.<BR/><BR/>"From a hardware point of view, the benefit of Hyperthreading increases as CPU clock rates and pipeline lengths increase."<BR/>-> you can look at it as 'the faster the CPU is versus the RAM access time, the more the CPU is doing nothing, so the more taking complex steps to make sure the CPU is busy gets useful"<BR/><BR/>->Also, one point to make about CPU utilization/capacity planning is: if you have hyperthreading enabled, CPU utilization is more than just the utilization %.<BR/><BR/>You can find more details in my recent <A HREF="http://msdn.microsoft.com/msdnmag/issues/05/06/HyperThreading/default.aspx" REL="nofollow">MSDN Magazine </A> article discussing hyperthreading performance (June 05).<BR/><BR/>// <A HREF="http://www.yanivpessach.com" REL="nofollow">Yaniv Pessach</A>Anonymoushttps://www.blogger.com/profile/12519481626365841087noreply@blogger.comtag:blogger.com,1999:blog-7434008.post-1118170715168285192005-06-07T11:58:00.000-07:002005-06-07T11:58:00.000-07:00hi any news on the exacct source? just curious! ta...hi any news on the exacct source? just curious! take careIt's Mr Gintonic to you!https://www.blogger.com/profile/00699311552850150297noreply@blogger.comtag:blogger.com,1999:blog-7434008.post-1117737979865334402005-06-02T11:46:00.000-07:002005-06-02T11:46:00.000-07:00I got a comment via email from Jaime Cardoso, I've...I got a comment via email from Jaime Cardoso, I've added my own response, and italicized his questions:<BR/><BR/><I><BR/>I think I understood your point and, it makes absolutely sense but, do<BR/>you think this will also happen with CPUs with heavy TLP?<BR/><BR/>I'll try to explain my question.<BR/><BR/>I never saw Solaris running on an Intel with HT but, assuming Sun took<BR/>the same aproach than it did with US-IV (dual core), Solaris only shows<BR/>you 1 CPU per each US-IV (and not one per core).<BR/></I><BR/><BR/>From a performance and capacity metrics viewpoint, Solaris shows you TWO CPUs per US-IV. The licencing argument is important but doesn't do anything to hide the two cores, which look exactly like two US-III CPUs by all measures. There is a command that licence managers are supposed to use which will report the number of CPU chips/sockets rather than cores.<BR/><I><BR/>This was made to gain grounds with Oracle's CPU licencing (yeah, I too<BR/>thought the excuse lame but, it's what Sun told us).<BR/></I><BR/>Some vendors do charge per chip, others still charge per schedulable entity. In theory this could lead to Oracle charging twice as much with Hyperthreading enabled. I'm not sure what their actual policy is.<BR/><I><BR/>Now, if I have something like vmstat running on a solaris X86 machine in<BR/>a box with an HT processor, would your thoughts about observability<BR/>still hold up?<BR/></I><BR/>Yes, vmstat will show you the Hyperthreads as if they were CPUs, it doesn't matter whether you run Solaris or Linux you see twice as many CPUs with HT turned on.<BR/><I><BR/>In an Intel CPU with HP, things could be misleading but, in an Niagara,<BR/>this error in observability can be critical<BR/></I><BR/>There is an interesting discussion of Niagara <A HREF="http://www.theinquirer.net/?article=19423" REL="nofollow">from The Inquirer, late last year.</A> The OS will see it as 32 CPUs, its capacity to utilization characteristic will also be non-linear so my comments on Hyperthreading will apply.<BR/><I><BR/>I'm guessing that vmstat (and sar, and iostat and, ...) should be "TLP<BR/>aware" and report the CPU loads accordingly.<BR/><BR/>--<BR/>Jaime Cardoso<BR/></I><BR/>These tools get their data from the same kernel statistic, it reports on every schedulable entity it can see, which means that every Hyperthread looks like a CPU.Adrian Cockcrofthttps://www.blogger.com/profile/14695336135416848505noreply@blogger.com