ok i need help about hyperthreading

fazz33 · Aug 25, 2003

hi everyone i am not a p4 owner but my friend is and his computer aperently has hyperthreadin enabled and he doesn't know how to disable it do any of you know how to or atleast where to disable it. sorry i dont have to many details for you since its not my computer other than itz got a p4 2.4c 800mhz fsb and an asus mobo

Blunt · Aug 25, 2003

why do you want to disable hyperthreading?

Mac · Aug 25, 2003

no point in disabling it, and i doubt you can disable it anyway.

n00bular · Aug 25, 2003

Go to the system bios and turn of hyperthreading. To enter the bios, press the Delete key when the computer is starting up.

gamefoo21 · Aug 25, 2003

the only decent reason for wanting to shut off hypertheading are:

1. he is running win9x

if neither of these apply tell him its a good thing and for him not to screw with the bios. just my thoughts on the matter. XP fully supports HT and gets a real very noticable boost from it.

Edit: Sorry seen benches that show that ht doesn't provide any extra boost in 2k and actually slows it down a lil compared to the huge boost it gives XP.

fazz33 · Aug 25, 2003

hyperthreading

hyperthreading is spliting 1 processor into 2 yes? well from what i hear having 2 (virtual or not) processors does not increase speed of a computer but will let you run more programs at once because for every program you run the processor time is split in half for both the processors. sooo if you have a 2.4ghz wit hyperthreadin enabled will that not decease the max speed to two 1.2ghz p4, my friend want to turn it off because he isnt like encoding videos and playin a game or what ever, so he doesnt need hyperthreadin enabled now and wantz his virtual processors to be 1 so overall speed of runnin 1 program will be better. (please correct me if i am wrong bout this stuff)

batboy · Aug 25, 2003

Must mobos allow the HT to be disabled in the BIOS. Most reviews I've seen usually gave a slight performance gain in many benchmarks with it enabled. I don't really see any big advantage to disabling it. No, HT does not make the processor run like two 1.2 gig processors, sheesh.

Gamefoo, you are incorrect about Win2k, at least about the Win2k Pro version anyway. I have Win 2000 Pro and it fully supports hyperthreading.

substratus · Aug 25, 2003

Just to offer my input, I tried disabling hyperthreading in my bios, figured it was worth a try since prime95 was crapping all over me at 3.4ghz and 1.7vcore. So I gave the whole idea the benefit of the doubt and disabled hyperthreading, and to my surprise, it is still running stable at 3.4+ghz. Seems to me that hyperthreading does allow you to run more programs (an inordinately higher number than I usually run at the same time, no doubt), but other than that I can't see much benefit. Could give you the extra edge you need in keeping your system stable at higher speeds. Then again, it might make no difference at all. These chips, though standardized to a point, all seem to have their own little quirks and preferences.

TC · Aug 25, 2003

There's no reason to disable HT unless you're going for the highest possible overclocked speed for a screen shot or something of that nature. And I have no idea why you would want to disable it if you're running win2k - that's what I've been running for months and it works just fine. I briefly installed xp pro on a p4c box and couldn't find any difference in performance.

substratus · Aug 26, 2003

After doing some more research I decided it was worth it to keep hyperthreading, I ran 3Dmark03 and it seems that my chip had better performance at 3.2ghz with hyperthreading enabled than 3.4ghz with hyperthreading disabled. I can do 3.2 with stock voltage, so if I max out at 3.3 or something with hyperthreading and vcore as high as 1.7, then I'll just go back to 3.2 and 1.525... it's just not worth the extra 100mhz to have to increase the voltage that much.

edit - with hyperthreading enabled, prime95 seems to immediately fail at the precise moment that it passes one series of tests with a certain FFT length and begins the next series with a different FFT length. Does this mean anything besides the obvious?

fazz33 · Aug 26, 2003

well ok this is what happened my friend diabled hyperthreading and then he ran 3dmark 2001 and scored like a meger 100points better, so in the long run he is gonna turn on hyperthreading and leave it like that.

batboy · Aug 26, 2003

Seems strange that a video card benchmark program would have a performance gain by disabling HT. I'm not at home right now, but I'm tempted to try doing that myself to see what happens (except I have 3Dmark2003). Anyone else wanna try it too?

NookieN · Aug 26, 2003

substratus said:
edit - with hyperthreading enabled, prime95 seems to immediately fail at the precise moment that it passes one series of tests with a certain FFT length and begins the next series with a different FFT length. Does this mean anything besides the obvious?

That could indicate a memory or CPU cache problem. Does it always fail at the same FFT bondary, or just any boundary? You should be able to adjust the FFT range so that it starts the torture test with FFT it's failing on. Which version of Prime are you running? If you have v23.6 you should be able to control the FFT size through the torture test menu. If you have an older version, you would need to edit the prime.ini.

Originally posted by batboy
I'm not at home right now, but I'm tempted to try doing that myself to see what happens (except I have 3Dmark2003). Anyone else wanna try it too?

I tried it and I only see a 14 point decrease with HT disabled vs. enabled in 3dmark2001. Since that's well within the margin of error of 3Dmark, I would say HT has no impact on the program.

Since87 · Aug 26, 2003

NookieN said:
That could indicate a memory or CPU cache problem. Does it always fail at the same FFT bondary, or just any boundary? You should be able to adjust the FFT range so that it starts the torture test with FFT it's failing on. Which version of Prime are you running? If you have v23.6 you should be able to control the FFT size through the torture test menu. If you have an older version, you would need to edit the prime.ini.

I've noticed that when I've got my system pushed near the edge, it tends to fail about 5 hours into Prime95. I haven't paid attention to the particular test that was running though.

Do you think that knowing which test is failing could give any useful insight into what the limiting factor is?

Atari · Aug 26, 2003

Man you scared me by sayin Win2k does not support HT. lol

Glad thats not true... My chip comes next week.

NookieN · Aug 26, 2003

Since87 said:
Do you think that knowing which test is failing could give any useful insight into what the limiting factor is?

Not exactly. If you were failing 256k for example, that wouldn't necessarily pinpoint the problem.

If Prime fails in very small FFTs (say <32k), but passes large FFTs (>640k), then it's possible the CPU's cache is struggling. On the other hand, if you pass small FFTs, but fail large FFTs, then main memory or the chipset is most likely the cause.

If Prime fails for only one FFT size, or a small range of FFTs, that could indicate a very specific hardware problem. For example, I had an old Athlon setup that could not pass Prime at 192K. Every other FFT was fine. Swapping the memory and CPU made no difference, so I decided that the chipset was at fault. Of course, Abit didn't see it that way.

The moral of the story is, by tweaking the FFT ranges you _might_ be able to get more info about what's failing.

batboy · Aug 26, 2003

NookieN said:
I tried it and I only see a 14 point decrease with HT disabled vs. enabled in 3dmark2001. Since that's well within the margin of error of 3Dmark, I would say HT has no impact on the program.

Thanks Nookie, that's just as I suspected.

gamefoo21 · Aug 26, 2003

Sorry I just seen old benches from when HT first came out that showed it did damage to Win2k's performance when enabled. I learn something new everyday

.

substratus · Aug 27, 2003

Batboy - I am talking about the two CPU tests in 3Dmark03, not the first four game tests. There is a score for 3DMarks and a separate score for CPUMarks. Sorry, should have been more specific

I posted the scores somewhere on this forum, but I'm too lazy to go dig them up.

batboy · Aug 27, 2003

I have Win2k Pro and I've done benchmarking with HT enabled and with it disabled. There is generally a least minor improvement in most benchmarks with it enabled. I can also do an incredible amount of multitasking with it enabled. Win2k Pro works great with HT from my experience. I think TC and Ol' Man will back me up on that fact.

ok i need help about hyperthreading

Member

Member

Member

Member

Member

Member

Senior Moment

Member

Senior Seti Addict

Member

Member

Senior Moment

Member

Member

Member

Member

Senior Moment

Member

Member

Senior Moment

Similar threads