Try here? The Ultimate 686 Benchmark Comparison
It's very extensive and both Coppermine and Katmai were used.

Tetrium wrote:

Try here? The Ultimate 686 Benchmark Comparison
It's very extensive and both Coppermine and Katmai were used.

Unfortunately it probably won't help, since the only way to find out which one is better is to do a fair test between the two cores. Means, both CPUs have to be of the same frequency and have the same FSB, but I guess nobody has done anything like that and people just refer to hardware sites.

If it's any help, the Mendocino Celeron 300 (Celeron 300A) was regarded by some as being equal to or faster than a Pentium II 450MHz (Deschutes) in some applications, possibly due to its on-die full speed cache. Of course, this meant running the CPU out of spec - 100MHz instead of the rated 66MHz FSB. This was, in fact, fairly common practice for a time due to the high price of the Pentium cartridges.

Though indeed, you will only find out by testing unless I get around to it first - which is not likely - let us know if you do though, because I have always wanted to run this test myself and have never found time or a good enough excuse to spend the money.

Dreamer_of_the_past wrote:

Means, both CPUs have to be of the same frequency and have the same FSB, but I guess nobody has done anything like that and people just refer to hardware sites.

I wanted to know this when I did my V2 CPU scaling project.

It's a 65 pages document with 875 benchmark results and on page 33 you can find information about Katmai vs. Coppermine 😀

http://www.philscomputerlab.com/uploads/3/7/2 … ing_project.pdf

Keep in mind these are V2 benchmarks only.

HighTreason wrote:

If it's any help, the Mendocino Celeron 300 (Celeron 300A) was regarded by some as being equal to or faster than a Pentium II 450MHz (Deschutes) in some applications, possibly due to its on-die full speed cache. Of course, this meant running the CPU out of spec - 100MHz instead of the rated 66MHz FSB. This was, in fact, fairly common practice for a time due to the high price of the Pentium cartridges.

My experience was different back in those days. Celeron was like a turtle compare to Pentium II. I see, well cost is no longer a problem.

HighTreason wrote:

Though indeed, you will only find out by testing unless I get around to it first - which is not likely - let us know if you do though, because I have always wanted to run this test myself and have never found time or a good enough excuse to spend the money.

Yeah, unfortunately time is also a big factor here and I just don't have equal CPUs based on both cores. In addition to it, I want to build a system on i820 first, because I have this strange feeling that i820 is highly underestimated over i815.

PhilsComputerLab wrote:

I wanted to know this when I did my V2 CPU scaling project. […]
Show full quote

Dreamer_of_the_past wrote:

Means, both CPUs have to be of the same frequency and have the same FSB, but I guess nobody has done anything like that and people just refer to hardware sites.

I wanted to know this when I did my V2 CPU scaling project.

It's a 65 pages document with 875 benchmark results and on page 33 you can find information about Katmai vs. Coppermine 😀

http://www.philscomputerlab.com/uploads/3/7/2 … ing_project.pdf

Keep in mind these are V2 benchmarks only.

Thanks Phil, I'll take a look at it right now. Great job, it seems that the only major advantage of getting Coppermine over Katmai is lower electricity consumption.

/\ Did you happen to have one of those dreadful cacheless Celerons? Those are supposed to be worse than Pentium MMXs they were supposed to replace.

alexanrs wrote:

/\ Did you happen to have one of those dreadful cacheless Celerons? Those are supposed to be worse than Pentium MMXs they were supposed to replace.

Well, it was a long time ago, so I don't even remember, but probably so. Since then I stay away from Celerons.

Dreamer_of_the_past wrote:

alexanrs wrote:

/\ Did you happen to have one of those dreadful cacheless Celerons? Those are supposed to be worse than Pentium MMXs they were supposed to replace.

Well, it was a long time ago, so I don't even remember, but probably so. Since then I stay away from Celerons.

The Mendocino ones are actually quite nice. About as fast (or almost as fast) as a similarly clocked Pentium II while running at a lower FSB. These chips are often the only ones that will work on s370 LX boards.
The 800MHz+ ones are also not that bad, but not as good compared to similarly clocked Coppermines as the Mendocinos were compared to the Pentium IIs. Just stay away from the 66MHz FSB Coppermine ones as those run at a by then very low 66MHz FSB and as there are simply much better chips available for the boards those Celerons are compatible with.

I don't know about later Celerons, but I suppose the ones past Netburst might've been alright?

Considering the fact that a big chunk of the 300A supply were installed and run 450 mhz without even trying hard, it's no surprise it was the way to go back then.

Dreamer_of_the_past wrote:

Thanks Phil, I'll take a look at it right now. Great job, it seems that the only major advantage of getting Coppermine over Katmai is lower electricity consumption.

That and the fact that Katmai doesn't scale much over 600mhz, whereas Coppermine does about 1ghz.

In games of that period Katmai was not faster.

Dreamer_of_the_past wrote on 2015-12-23, 22:50:

Have any of you actually personally tested performance of Katmai vs Coppermine? Especially in games. I heard some rumors that Katmai with its slower, but bigger 512KB cache is actually faster than Coppermine with its faster, but smaller 256KB cache. Yes, I am aware that majority of hardware sites if not all say otherwise which is why I would like to see some personal benchmarks to be sure.

I've just tested it.
450 MHz Katmai/100 MHz RAM - 36 FPS in RTCW
500 MHz underclocked Coppermine/100 MHz ram - 45 FPS

Even when you count with 50 MHz difference, it still looks better for coppermine.
When I unlock it to original 667 MHz and 133 MHz RAM, it's like 3 times faster in games.

So, yes, it was a myth, Coppermine is clearly better, unless you have some rare Katmai with the best available clock or you have it highly clocked.

But in days when Coppermine started with lower clocks, it could be questionable if it's a good upgrade, but after clocks got over 600 MHz, coppermine was clearly much better, especially with 133 MHz RAM.