Sign in with
Sign up | Sign in
Your question

the athlon 1.2 or the p4 1.5 are not the fastest

Last response: in CPUs
November 28, 2000 1:54:38 PM

Well the dual p3 1ghz blows them both away i got a vp6 and @(1ghz)p3 coming in.

More about : athlon fastest

a b à CPUs
November 28, 2000 3:45:56 PM

That really doesn't seem quite right to me. A dual P3 system is, indeed, pretty fast in general (assuming you're running an SMP OS like Win NT or Win 2K), but even then, your performance gains are nowhere near perfectly scalable. In something like 3D Max you'll get good scaling, but even then not perfect (like 2x 1000). I'm still inclined to say that a 1.2 T-Bird will either top the 2 P3's or tie them. Of course, I'm anxiously awaiting dual T-Birds to construct my dream graphics workstation! :) 

a b à CPUs
November 28, 2000 4:34:54 PM

Give me a number, say FlaskMPEG score, since it is the hotest topic in the forum ;) 
Related resources
November 28, 2000 7:05:27 PM

Two 1GHz P3's would, theoretically run at around 1.6GHz, but I still think one 1.2GHz TBird, or even dual 1GHz- TBird's could beat it.
a b à CPUs
November 28, 2000 9:19:57 PM

Uh, that's the stupidest thing I ever heard. 2 P3s would probably beat an equally clocked Athlon, but how can you compare 2 processors to one? Why don't you just compare the Athlon to a Dual Alpha server? Or a G4? Let's try to use some common sense here.
a b à CPUs
November 29, 2000 12:52:21 AM

Neh.. Dual P3 1 Ghz = Uni P3 1 Ghz in almost any circumstances, unless the program is being written multithreaded. The only program is so well multithreaded is RC5 Client as far as I know. Q3A written multithreaded, but can't show it real potential in any Video card as far as I know.

It is like saying, you need 10 min to convert a WAV to MP3, it will still take 10 min to convert a WAV to MP3 on dual processor combo. Except, when you convert two WAV to MP3, it will also take 10 min on dual processor combo ;) 

Speed up? Neh.. very rare.. unless specifically written in multithreaded.

Do more stuff in the same amount of time? Yes.
November 29, 2000 5:30:30 AM

It is pretty obvious that if one opts for the dual solution then they will also be runinng an smp enabled os, or maybe the case of an "amd puppy" who does not know anything 'bout smp.
November 29, 2000 1:48:56 PM

oh course i have smp OS

1. win2k
2. beos
3. linux
a b à CPUs
December 3, 2000 10:32:47 AM

Rats, from the subject, I thought this was going to be about all the real processors out there like IBM Power or HP PA-RISC, not these X86 toys... ;) 
a b à CPUs
December 3, 2000 11:02:49 AM

Toys?! Sure these cpus can't compare to an IBM or HP high end part on a 1 to 1 basis but they are NOT toys.

You start clustering some of these "toys" and you end up having a pretty powerful computer that can beat the pants off of these high end parts. These clusters can also be put together a hell of a lot more cost effectively than any low production super chip from IBM or HP.

Not toys my boy, just super computers waiting to happen.

"in the end it's all about holding your breath"
December 5, 2000 1:43:59 AM

I Imagine that you get a marginal performance gain with the two p3's. I mean it might be a good gain but, I also imagine you payed quite an exurbanite amount of money to get the marginal gain. Give me an idea of the cost/per performance and I will tell you if it is worth it to me. I know it is worth it to you as you have it already!

a b à CPUs
December 5, 2000 2:25:38 AM

hey, at least I'll admit it: If you need this kind of power, then you'll rule the world! ;) 

You must admit guys that saying that you have a DUAL pIII 1ghz is a pretty cool prospect!
December 5, 2000 3:51:48 AM

Who cares about cost/performance ratios. If you buy such a fast CPU you aren't buying it for value for money!

What is the rest of the system? Any chance of you telling us the results of Q3 demo001?
a b à CPUs
December 5, 2000 7:25:43 AM

3D max actually scales perfectly with 2 processors 198% compared to one processor when rendering. Of course your subsystems need to be on par.
But too many programs dont scale at all.

December 5, 2000 2:09:18 PM


16-BIT = 181 FPS
32-BIT = 162 FPS

16-BIT = 150 FPS
32-BIT = 145 FPS

16-BIT = 120 FPS
32-BIT = 109 FPS

2x P3 700 @ 933
384MB PC150
2X 18GB 10K SCSI3 (133MB BURST)

December 5, 2000 2:31:23 PM

Wow! How about at 1200*1600 highest everything?
December 5, 2000 2:41:10 PM

1200 x 1600
16-BIT = 60 FPS
32-BIT = 55 FPS

1028x768 32BIT
3DMARKS - 6781
a b à CPUs
December 5, 2000 7:01:13 PM

I dont think that a 32mb radeon is capable of those kind of numbers, regardless of the system it is hooked up to. I would say even with only 1 1Ghz PIII the radeon would be the limiting factor. If you got those numbers then i'm impressed, or maybe the Video card fairy put a Gf2 Ultra in there while you werent looking :( 
a b à CPUs
December 5, 2000 8:07:15 PM

Very shortly the 760MP motherboard will be released and AMD WILL have an duel CPU system available. Also Micron is releasing a multiple (2+ CPU's) motherboard for the Athlon and unlike Intel's duel CPU motherboards that run a shared front side bus, The AMD multiprocessor chipsets will NOT share a front side bus, each CPU will have its own 266 Mhz front side bus :) 

December 6, 2000 1:11:21 AM

My point was two 900 dollar processors for how much of a performance gain over one 500 dollar AMD Processor. I'm thinking that the performance gain would be pretty good but would it be worth the extra 1300 dollars? Hey you folks decide.

a b à CPUs
December 6, 2000 2:22:51 AM

.... Dual PIII has a gain of less than 1% over single PIII unless the apps. is seriously optimized multithreaded.

.... Remember when Athlon first came out and Anand had benchmarks showing a single Athlon outperforming Dual PIII's. And a single PIII outperforming dual Celerons.

.... Next month we get dual AMD T-Birds or Palominos outperforming any dual intel setup.
December 7, 2000 1:06:05 AM

I look forward to the dual AMD review next month. I still wouldn't pay for twice the CPU cost and not get significant power increase. I of course am not seriously optimized either. Get me the AMD and let someone else keep Intel in business buying dual PIII's.

a b à CPUs
January 8, 2001 12:41:46 AM

True, so long as your workload scales nicely in a Beowulf cluster... The hardware is probably cheaper, but TCO likely favors the "real" stuff-- would you rather have a staff administer 50 commodity systems or 2 High nodes?

I was mainly venting frustration about how every time an advance comes to the pee cee world, Intel or Microsoft crow about it like it is the first time any of this has been done... pretty frustrating to folks who have worked on real stuff for years....

Also wanted to see more of a flame war, but nobody took me up on it... ;) 
a b à CPUs
January 8, 2001 2:45:50 AM

What I can't understand is why you bought a VP6 when you could have got a BP6 with better performance and compatability, comparing the BX chipset with the VIA ANTYTHING! Or did you decide that overclocking your AGP bus was not a good idea? (your second mark against performance)
a b à CPUs
January 9, 2001 8:11:16 AM

>Two 1GHz P3's would, theoretically run at around 1.6GHz,
>but I still think one 1.2GHz TBird, or even dual 1GHz-TBird's
>could beat it.

Your theory is not 100% true.
Every SMP OS is not created equal.
Dual 1GHz P-III is equal to 1.9GHz on BeOS (
in Linux I heard about 60-80% (1.6 - 1.8GHz)
in Win2000 I heard the number to be 40-50% (1.4 - 1.5)
And WinNT 4.0 is worse.

Anyway, I'm not hardware/software reviewer, so I can't verify this number.
I'm really hoping that Tom will do this dual processor review with lot's of
OS (and lots of Linux kernel too).

A side note from me.
A real SMP OS is the one that can take the benefit of more than 1 processor
natively for *EVERY* application without special compilation/design.
And it's only 1 true SMP OS that I know... BeOS (

Get BeOS (
The Best ever Operating System