fsb vs cpu Mhz

titling this thread was the most difficult task. Here's the question.

Which would be faster on an Asus p2b mobo?

a celeron 1,3ghz tualatin at 100Mhz

or

a 1 ghz PIII at 133 Mhz.

I have them both. Which one would be faster?

the p3, it has more cache which the pentiums love and the faster fsb gives it more memory bandwidth.

yeah, to me also it's obvious about the memory bandwidth, but don't they both have the same 256kb cache?

Don't mean to seem flippant, but have you considered dropping 'em in, & running some benchies, since you've already got the chips?

Not flippant at all. The thought had crossed my mind I was trying to avoid the task by asking a wuation and hopefully someone would know off-hand.

Basically, does the increase in memory bandwidth justify sacrificing cpu mhz?

if no one replies with a decent answer, I guess I just might do that. It's a question of time, really.

If I asked which was faster a celeron slot I 300Mhz chip or a Pentium III 600Mhz cpu , I'd get some quick knowing responses.

If the answer to my original question is so in doubt, then the difference must be minimal.

I think that Celerons are quite fast, and 300 Mhz would make it a bit faster than p3. But when it comes to overclocking things could change.

I think those Celeron Tully's are pretty good. I think they have 256K cache. In any case with a Celeron vs. P/// I'd probably take the P/// unless the Celeron is like 1Ghz faster or something. Either CPU you choose, I'm sure that you'll get some good performance. Good luck

the celly will win....since both have 256kb cache

Bardos -

Glad you didn't take my post the wrong way. The reason I was advocating actually using the chips, is each s/w app seems to have it's own affinity for cpu architecture and system attributes.

I do a lot of Distributed Computing.

Some apps, like SETI, are very Front-End execution cycle sensitive. Give 'em a short pipeline, or a longer pipline @ a fast FSB, & they will respond well, as long as they have a reasonably moderate sys mem b/w. That's why AMD generally ruled the mainstream desktop SETI roost, before the P4C's HT & 2 client config, shifted the SETI paradigm. But yet, when not using HT, and using only single client instances, AMD procs, even w/single channel DDR, or even sometimes SDR, will still generally rise to the top of the pack.

OTOH, I also run Seventeen-Or-Bust, which is a subset of GIMPS. This client appears to respond extremely well to increases in memory b/w. While it's true S||B gets a nice boost from the SSE2 SIMD set, if you do anything to open up the mem b/w, you'll be rewarded w/a nice increase in crunching rate.

Also, IIRC, the Tuallies have some slight architectural changes involving prefetch & better branch prediction, which the non-Tuallie PIII's do not have. This should make the Tuallies better at executing branchy code like office apps & source code compilation.

I'm in these fora 'cuz I'm an OC'er, but I crunch DC for another fairly large team. The differences in client/cpu architecture affinity can be so striking, that our 'project sub-teams' do machine swaps, to improve our over-all 'super-team' performance. IOW, user #1's farm may be running some AMD machines on a P4 friendly project, while user #2's farm is running some P4's on an AMD friendly project. Each user enjoys thier respective project, and wants to remain on it. Even though on different projects & sub-teams, they will run thier machines under each other's accounts & project sub-teams. They've essentially done a virtual machine swap, w/o physically exchanging machines. It's a machine-for-machine swap, and each guy wins, along w/the project sub-teams, and the super-team, in general. I'm usually not in the Overclockers.com DC forums, so I'm not sure if this is being done here, too.

I guess this was a long way of saying, if you ever find the time to try both chips, using your target system & apps, along w/some tweaking, you'll find the optimal chip for your use.

Good Luck,

Strat

I've never compared a 1.3G, but I did compare a 1.4G Celeron and a 1G P3 133 in the same computer. The Celeron was impressive in the CPU marks, but fell behind in the RAM scores. In 3DMark 2001SE, the 1G P3 had a very slight edge. So I would say the 1G P3 would be better overall than a 1.3 Celeron but the margin wouldn't be that large of a difference.

These benchmarks were on a Soyo 815eb motherboard that supports 133fsb. You will need to run the P3 at 133fsb with PC133 ram for it to outperform the Celeron running at 100fsb.

<pls disregard - erroneous double post>

OC wise, the tualatin is the best choice because you can push it up easily. Remember P3 of 1 GHz were about their maximum potencial with default speed.

Are you shure that the mobo is ok for these cpu's?