a just benchmarking program!

[tking]

Hi There,

I'm looking for a benchmarking program that gives an even handed evaluation of processor power.

It has been said that Sandra is designed to make P4s look good.

Is there something that does basic engineering tests and produces a lot of raw data?

I'm trying to get to the bottom of the cpu architecture philosophies and a just benchmarking program would let me do some independent testing.

Thanks for any help.

T

[jad1097]

RC5 benchmarks raw computing power. It's not really a benchmarking app though. I don't think it is what you are looking for though.

http://www.lostcircuits.com/cpu/p4_2400/6.shtml

http://www.distributed.net/

[grimfandango]

have a look at Pc mark 2002!!!

www.madonion.com

:t

[$1500-P4 gamer]

"It has been said that Sandra is designed to make P4s look good."

Gee, I wonder what type of person said that.

:rolleyes: :t

[grimfandango]

yea I heard that too..Lot of people say that ..
I wonder if that is true!!!

[tking]

Originally posted by $1500-P4 gamer
"It has been said that Sandra is designed to make P4s look good."
Gee, I wonder what type of person said that.:rolleyes: :t

Sigh, you know it.:rolleyes

I'm a firm believer that everything can be quantified. I think every successful company is ethically flawed, it's the only way to survive in the modern business world, so I don't go for this Evil Giant, Little Hero mentality: it's all just marketing.

That aside, I would like to learn the engineering behind the machinery. I've heard people talk about pipelines and processor speed vs. that esoteric thing that is more valuable than processor speed. What I'm looking for is a benchmarking program that shows me the strengths and weaknesses of technologies and lets me understand the mechanics.

I don't care if it's real world or app specific, I just want to see something that exploits the architecture and lets me see the strengths and weaknesses.

Links to articles, a benchmarking app, whatever you think might do it for me.

Thanks for the replies,
T:

[jad1097]

"that esoteric thing that is more valuable than processor speed"

IPC (instructions per clock cycle)?

[tking]

Originally posted by jad1097
IPC (instructions per clock cycle)?

This would be the width of the pipeline as opposed to its length?
I've heard these terms bandied about but I've never heard a single, straightforward explanation.

Athlons have a wider pipeline? (more IPCs) Pentiums have a faster, longer, narrower pipeline? (less IPCs, more cycles)

T

[$1500-P4 gamer]

If you want to exploit the strengths of a P4 just do some encoding into divx! What the auther said would take his P3 800mhz (which has a stronger IPC like the XP) 15hours I did in 4hours with my wimmpy Williamette 1.5gig! P4 is really a multi-media giant. Its made for running photosuite, renedering progs., encoding decoding etc. Games run about = on both. I'm sorry I dont think a 1-10 fps is a big dif. Maybe its just me? Now if you wanna exploit the strengths in a XP just do some RAW mode stuff like seti@home and all that totally antiquated code stuff! Run some really old dos games! I'm just j/k on the last one! The point is XP is better at raw mode, while anything (which is alot!) with sse or sse2 is way faster on the P4. Hope that helped some and didnt confuse you all the more.

Try reading this it should help ya,

"However, the ability to scale the Pentium 4 to before-unseen clock speeds carries with it an inherent disadvantage. Mainly, the average number of instructions successfully executed per clock cycle (referred to as IPC) has decreased. Therein lies AMD's window of opportunity. While their Athlon is not as scalable in terms of frequency, its IPC is greater.

So, even though Intel has sat merrily at 1.5GHz for the past four months and AMD has taken a breather at 1.2GHz for the past five, the performance delta between the competing processor lines has been very small. Generally, Intel has been dominating floating-point intensive applications, while AMD has taken most of the victories in integer-based programs."

From here: http://www.sharkyextreme.com/hardware/reviews/cpu/thunderbird_1-33ghz/

I like the release price of the Athalon T-bird 1.33gig
$350 in volumes of 1,000- thats not cheap!
Hehe I know it was first release, just picking!



:) :t

[tking]

Originally posted by $1500-P4 gamer
If you want to exploit the strengths of a P4 just do some encoding into divx! ... Now if you wanna exploit the strengths in a XP just do some RAW mode stuff like seti@home and all that totally antiquated code stuff!

The Intel CPU has more built in instructions but less ability to process raw data. The AMD CPU has less built in instructions and a wider IPC. One is specialized to certain code and the other isn't. Yes?

If this continues is it possible that one day Intel CPUs would come complete with Intel code and Intel OS? Is there enough of a benefit in building a jack of all trades that AMD and Intel can each go their own way and find markets that don't overlap?

T

[$1500-P4 gamer]

"The Intel CPU has more built in instructions but less ability to process raw data. The AMD CPU has less built in instructions and a wider IPC. One is specialized to certain code and the other isn't. Yes?"

Right, but thats a per mhz scale. See the P4 has the longer pipeline Which allows it to scale to a much higer freq. In this case mhz DOES matter due to the higher mhz the more cycles per sec. This allows it to scale to a point where it is = to XP in raw mode and the newest way overpriced (as all flagship chips amd and Intel) p4 even wins. But the price doesnt justify it if you are doing work thats raw mode oriented. But if you have sse and sse2 optimized apps and gaems which there are alot of then the P4 has a stronger fpu due to optimisation.

So as you can see when looking within the same architecture not crossplatform, mhz does matter as all chips scale this way as it reflects the actuall cycles per sec. Now useing mhz to compare a P4 against a P3 or a XP is totally inacurate as the dif. in optimisations and pure architecural dif. themselves like the pipeline length and etc.

Did that help ya?!;) :t

[tking]

Originally posted by $1500-P4 gamer
See the P4 has the longer pipeline Which allows it to scale to a much higer freq.

Thanks PCG, it did help. My only question is why does a longer pipeline allow a higher frequency?

I'm trying to fit it to something I know well. I used to be an auto mechanic. I keep thinking of the Athlon as a torquey V8 and the Pentium as a motorcycle engine. Both are very powerful but in very different ways. The bike uses RPMs (like clock cycles?) to make power while the big V8 uses shear displacement and lower RPMS to make power. Bad analogy?

T

[De Joker]

Well now I’m a little confused because I read the reviews (tests) at tomshardware, and according to the test results (for divx encoding) the XP’s beat the P4's at comparable clock speeds.

By comparable clock speed I mean a XP 1900 (which is actually 1600 MHz) versus a P4 1900.

So which is it then, are the XP’s faster at encoding divx or not.:confused:

[$1500-P4 gamer]

Depends. Most will run way faster on the P4. Hey its a multi-media beast, thats what its deseigned for. If it has sse or sse2 optimisation it will run the fastest on P4. Any good encoder will have sse or sse2. I'm just going by what the auther said in reference to his P3 sys-not a AMD one. But you see the P3 and XP mhz for mhz are about=. This is what I'm going by. Just saying its alot faster at encoding than given credit for. Heres a example. I have alot of dvd's. But my living room pc has no dvd player so I rip them to vcd as to be able to watch them in living room on reg. cd player. It should take T-mpeg inc. about 10-12hours on a good P3 sys. Alas it takes me 4hours! I was impressed I assure you!;) :t

And mine is only a P4 1.5 williamette. The northwoods are way faster at encoding than this one due to twice the L2 cache. I'm also useing rdram which is a HUGE help.

[$1500-P4 gamer]

The longer pipeline kills alot of heat issues. You see by cooling a Cpu from the top you cant kill all the heat within the chip. I mean internally certain parts are hotter than others. This reduces the heat where its a real issue. Thus allowing a higer mhz overall. This is also why it does less work per cycle. But more cycles. This is also why the new P4 2.5 533fsb beats the fastest XP. Its scaled to the point where it ineffeciencys have been over comed. Intel made a good bet, ala s I always had faith in them. But some people just think that a multi-billion dolar co. is dumb enuff to not know what they are doing. Whos the dummy now?!?