I think their kernel just doesn't supports that (like some linux distro). all the test i've done demonstrates that turbo boost isn't working and no tests demonstrates the opposite, now everyone can link all the websites they wants speaking about TB technology, but numbers are numbers, and words are words. There are very slight changes in frequency readings (a couple of megabytes), so that means rightmark is currently measuring cpu frequency over time correctly.Īs you can see osx is continuously changing cpu workload between the various cpus, instead of turning unused ones off and pushing up frequency of the others. There are NO changes of the cpu frquency for the whole calculation, and cpu usage is stuck at 100% here's the picture: This is again a proof that turbo boost isn't working at all.Īgain i can't say it for sure, cause it should be vmware that doesn't recognize cpu frequency changes, but considering that the cpu is virtualized i highly doubt it's wrong.īasically i've installed a win xp over vmware, i have assigned ony 1 core to it, then i've launched rightmark, that checks cpu load and frequency every second, and then i've launched a superpi calculation while osx was nearly idle (only cpupalette running)
#TURBO BOOST SWITCHER WINDOWS 10 DESKTOP PRO#
hope it works for you.Īnd for the happiness of mac pro nehalem owners here is another test!!!!
#TURBO BOOST SWITCHER WINDOWS 10 DESKTOP FREE#
Please fell free to demonstrate that i'm wrong, i'll be very happy if i'm wrong, by the way for the application i use daily for my work it looks like it does NOT work.
I have done more tests in 10.5.x with some audio applications including Logic.Īll tests confirm that Turbo Boost it's just a dream on osx.įinally, i really can't understand how anyone has noticed it after nearly 1 YEAR, what the **** you do with your mac other than wanking on youporn and spamming on facebook ?īy the way the lack of any benchmark and monitoring tool on OSX does not help. So we can identify if it's an hardware limitation, or an OSX missing feature. If anyone can check if that feature is working on a bootcamped windows, it's quite easy to check, even on Intel website there are some tools. There is still a test to do, does anyone has bootcamp ?
What we can deduce now? we have 2 options:ġ-Turbo Boost is working but the performance improvement is very very low (i don't believe that!)Ģ - TURBO BOOST IS NOT WORKING, yeah, that's the right one!!!Ģ - remove all false advertising you have on your papers. There are 2 results in the window, the first with HT, the second without.
I've launched 4 tasks of LAME simultaneously (well, i have to press enter 4 times so it's not exactly simultaneously 🙂 )Īs you can see the encoder took about 4 seconds more to encode, but it's due to the fact that HD is working 4th times more, and also L2 cache, RAM and bus are shared so it's normal that 4 tasks take a bit longer than just 1. So to demonstarte that Turbo Boost isn't working i've done another test.
So basically until now we can only say that Third one is with HT off and only 2 CPU activeġ-single threaded applications works better when hyperthreading is off, and we all already know that.Ģ -running with only 2 cpus instead of 4 does not affect performanceģ -running with only 1 cpu get even worst results (that is normal cause in the same cpu there are also system tasks)Ĥ - There are no changes in performance even if i disable CPUs manually. Look at the image there are 4 tests conescutively.įirst encoding is done with everything default, 4cpu + hyperthread I'm using also "processor" and "cpupalette" taken from the xcode tools. I 've tested encoding a 800MB wav to mp3 using LAME mp3 encoder, that is a single thread application I have just installed osx 10.6.2 on a fresh new hard disk and updated everything to the lastest version Ok i'll show you the last test i've just done.
0 kommentar(er)
