View Full Version : HyperThreading +50% performance?
I tried SuperPI on my friend's computer. He has Pentium 4 2.4C and 512MB RAM.
At first I ran SuperPI 1M one at a time. It finished in 55 seconds.
Then I ran 2 SuperPI 1M at a time, one forced to CPU no.1 and other to CPU no.2, both with Realtime priority.
They finished in 76 seconds. Both. :slobber:
Then I turned off HT, ran 2 SuperPI again and they finished in 120 seconds.
It means that there is nearly 50% bigger performance with HT enabled?
Does the same happens with higher frequencies? To the Prescott ?
I'm VERY curious, please you with NW / PS do some tests :)
BTW: 76/2 is 38, which is my time @ 2.4 GHz on my A64 with FSB240!!!
P4 HT ties A64 with 2 threads ? :stick:
craig588
10-28-2004, 03:21 PM
It's P4C2400!
grimREEFER
10-28-2004, 03:29 PM
I tried SuperPI on my friend's computer. He has Pentium 4 2.4C and 512MB RAM.
At first I ran SuperPI 1M one at a time. It finished in 55 seconds.
Then I ran 2 SuperPI 1M at a time, one forced to CPU no.1 and other to CPU no.2, both with Realtime priority.
They finished in 76 seconds. Both. :slobber:
Then I turned off HT, ran 2 SuperPI again and they finished in 120 seconds.
It means that there is nearly 50% bigger performance with HT enabled?
Does the same happens with higher frequencies? To the Prescott ?
I'm VERY curious, please you with NW / PS do some tests :)
BTW: 76/2 is 38, which is my time @ 2.4 GHz on my A64 with FSB240!!!
P4 HT ties A64 with 2 threads ? :stick:
u dont own an amd, do u?.....
u dont own an amd, do u?.....
I have Athlon 64 3200+ NC.
AMD Athlon 64 3200+ NewCastle
MSI K8N Neo Platinum 1.52 Beta BIOS
2x 512 MB DDR500 A-DATA Vitesta Hynix @ 250 1:1 2.5-3-3-1T
Sapphire Radeon 9800 Pro R350
Enermax EG365AX-VE FMA 3.3V@32A 5V@32A 12V@26A
DVD-RW TEAC
Seagate 200 GB SATA
Audigy 2
Mouse Logitech MX310
Monitor LCD Prestigio P171
Lithan
10-28-2004, 08:49 PM
Superpi is heavily Intel-leaning in the first place. Running 2x with HT just lets it work more efficiently. So technically, an intel without HT would be as fast as AMD with both running a single example, but because of heavy cache misses and other problems caused by the p4 architecture, the p4 needs HT to let it run closer to full potential on superpi. This really demonstrates just how intel bias superpi is if p4mhz=A64mhz in it.
Can someone with 3 GHz+ Northwood and Prescott do the same tests ? I want to see the results. ;)
Maximus
10-29-2004, 02:40 PM
Bone stock 3.2GHz P-4C, on Asus 865 board w/HT enabled = 43sec.
w/HT DISabled = 45sec
below A64 3.4GHz system = 36sec
Bone stock 3.2GHz P-4C, on Asus 865 board w/HT enabled = 43sec.
w/HT DISabled = 45sec
below A64 3.4GHz system = 36sec
Could you please do a SuperPI test with 2 threads ? Assign one SuperPI to CPU0 and the other to CPU1 and bench.
And someone with Prescott could you test this too? I want to see if Prescott has better HT than Northwood. :D
redstang308
10-29-2004, 03:04 PM
I have a prescott and I achieved a 40 on 1mb see sig
I have a prescott and I achieved a 40 on 1mb see sig
Could you please do a SuperPI test with 2 threads ? Assign one SuperPI to CPU0 and the other to CPU1 and bench.
Then do a normal 1M test and post results, thanks :toast:
saaya
10-30-2004, 02:12 PM
P4 HT ties A64 with 2 threads ? :stick:
in theory, yes :D
in most real world applications... no
rick_fx
10-30-2004, 02:24 PM
Intel = :owned:
Beenthere
10-30-2004, 03:14 PM
Why do you think they call it "HYPE" ???
Few applications use hyperthreading. Unless you have apps that specifically are written for HYPE, there is a performance penalty if HYPE is enabled.
After all why do you think it's called "HYPErthreading" ???
;)
My feel of a friends computer with HT was that it is SLIGHLY faster than my A64. So i think it helps a BIT, not a lot. Of course i don't think that P4 matches A64 ALWAYS with HT (in practice never), but in 2 threads with superpi it happens.
I wonder if A64 dualcore will be better in multitasking that P4 with HT? What do you think?
I think yes, but i want a confirmation of my thoughts :)
jjcom
10-30-2004, 07:29 PM
a dual core Athlon 64 would be better at multi tasking than a single core P4 with HT.
Falkentyne
10-31-2004, 02:24 AM
a dual core Athlon 64 would be better at multi tasking than a single core P4 with HT.
Water is wet.....did you know that?
And the sun is hot....
Lithan
10-31-2004, 03:06 AM
Water is wet.....did you know that?
And the sun is hot....
Fire... hot?
jjcom
10-31-2004, 07:42 AM
I was just answering a question....
water...wet
fire....hot..
overclock...fire...fire...hot...water...wet...wate r..put out fire...
stardust
10-31-2004, 06:11 PM
What's with the cat avatars on this forum?!
Well, my old 3.2ghz Northy with HT did 41 sec and my A64 @ 3200+ does 38 sec.
Falkentyne
10-31-2004, 06:17 PM
I was just answering a question....
water...wet
fire....hot..
overclock...fire...fire...hot...water...wet...wate r..put out fire...
I'm joking.
I actually meant the reply for the guy who asked the question but it was too late.....