[ChankastRules]

I run this program that does advanced calculations on my intel pentium 3.0 ghz. When it runs it says only 50% of the CPU is used.. fine it's the bug in windows XP with hyper threading.. nothing to worry about right? WRONG. I sent the program over to my friends AMD XP 2.2 GHZ, sure enough his computer did the calculation about 20-25% faster.

Well, that's weird.. hmm but one test can surely not be enough to determine things like this, so we did lots of benchmark, and it turns out aslong as a program doesn't have support for hyperthreading indeed only 50% of the CPU is used.

Thanx a lot Intel, well now it's AMD forever, so I just want to warn everyone never to buy a intel proccessor again. It's so sad that I have to send him my calculations because he can do them faster even if his computer is two years older, and besides he got it for a cheaper price.

 

[ChankastRules]

we did not do the benchmark with only that program ( even if it mostly contains lots of additions, which in fact should be Intels strong point ), we also tested with other benchmark including for floating point operations and so on.

If we launch two programs at the same time though, my computer wins with about 30% or more on the very same application. ( The good thing is that most new programs support HT though, and on those mine is also about 30% faster )

 

[ChankastRules]

the math program is a lab project so I can't even share the name of it.

Here are some of the benchmarks attached though.

Results were:
my P3.0GHZ
1bench: 54484 ms
2bench: 69265 ms

Both at the same time:
1bench: 55321 ms
2bench: 74636 ms
total: 129957 ms

AMD XP 2.2 GHZ ( or AMD 3200 as someone said is a better name )
1bench: 42323 ms
2bench: 46186 ms

Both at the same time:
1bench: 71152 ms
2bench: 99412 ms
total: 170564 ms

All values are approximate values and may vary with +/- 200 ms.

For benchmarks with programs that support HT it's best to check ( http://www.tomshardware.com/cpu/20030513/index.html ) yes the P3.0 GHZ is much better here..

 

[ChankastRules]

Mine is C ( 800 MHZ FSB ), and 3dmark has nothing to do with my issue here as it has HT support. Try the benchmark I attached they don't have it.

I'm not irritated because his AMD is faster, because it's a lot slower in games and apps with HT support, but I'm extremely irritated that 40-50% of my CPU is being idle while I wait for said math program above to finish a computation. The only solution is to start up another process with the math program and do two computations at once, but at many times I don't want to do this.

 

[ChankastRules]

yes, you have to re-install windows, and it'll slow things down, HT is used all the time in windows XP, it increases performance quite a lot if you run 2 or more heavy load applications at the same time.

I guess you guys didn't read what we did in the benchmarks at all, but oo well, I got to rant about it anyhow, so I feel better now :rotflmao:

 

[ChankastRules]

In modern games it's not a problem, however doom3 ( openGL ) doesn't like ATI cards.. tell your friend to go buy a NVIDIA card LOL

it's sure not easy to be a PC owner these days, maybe it's time to buy a NextBox or Xbox 360 or whatever it will be named :S

 

[ChankastRules]

Single or multi-threaded?
Float or integer math?
Which compiler?
Optimization settings for the compiler?

Single-threaded
float math ( some integer as well but float would be the heavy part )
MS VC 2003 ( optimized for x86 , not intel or AMD specific )

 

[ChankastRules]

The number is probably how long it takes to run the benchmark.

Yes, ms is milleseconds :O, it's not a score... it really makes more sense to see how long it takes to run computations, since that's what is interesting.. atleast assuming both computer calculates everything correctly LOL!

That's probably why you're getting better results with the AMD.

Yep, as you can see the P3 is much better at running many things at the same time, while the AMD is better in the single-thread case for this scenario of computations, which is in the first place why I'm dissapointed I got an intel , and is why I started this thread

 

[ChankastRules]

No, that's a display issue with the Windows Task Manager.

Well, if you run two threads ( the programs at the same time ) you can see that the increase it takes to calculate the result is not much longer, compared to running them separated, that's why I say it doesn't use much more than 50% of it's potential computational power on process such as the benchmark I sent.

 

[ChankastRules]

It's only pointless if you're running one program without any threads