p3 667 or celeron 1.2 in desktop?

[dsigma6]

i have an old emachines t1220 with a 1.2ghz celeron, 100mhz fsb...i dont know about the cache. my friend gave me his old mobo w/667mhz p3 with 133mhz fsb.

ive been told that the celeron can only handle one process at a time, or something like that- so the p3 would be able to handle more, but not as fast?

the p3 mobo also has more memory slots, pci slots etc. im probably going to assemble it just for fun, just wanted to know if its a waste of time.

[leoblob]

That Celeron 1200 is faster. It's a "Tualatin" Celeron, which is actually a true PIII with a full-size, full-speed 256K cache... running with a 100MHz FSB.

That PIII 667 does have the faster 133MHz FSB, but since the clock rate is only about half that of the Celeron, it will be slower.

[dsigma6]

thanks leoblob. i think i might use the p3/mobo to build my folks a basic email checking computer.

[AlphaKilo470]

As far as desktop Celerons are concerned, I'd avoid the 66mhz bus versions such as the 500 and 700. Starting with the 800 Intel decided to give the Celeron a 100mhz bus speed and for most applications, the L2 deficiency when compared to the PIII counterpart makes for negligible difference unless you run tasks that are taxing on the CPU.

[dsigma6]

i don't consider anything i do to be taxing on the cpu, however the celeron stays at 100% cpu even while viewing a simple thing like 'the fan' on comcast.net.

my r40, which isnt the most powerful computer, hardly notices it.

if thats the case, i think the p3 would do a better job. i think ill just try and see what happens.

[leoblob]

The later PIII Celerons (Tualatins) all had the same 256K cache as the "real" PIIIs did.

At the end of the PIII era, Intel had some pretty weird naming conventions. Initially, all the "PIIIs" with 100MHz buses had 256K caches and all the "Celerons" had 128K caches. Once Intel switched to the Tualatin manufacturing process, they stopped making processors with 128K caches. Instead, all the 100MHz PIIIs with 256K caches were named "Celerons," and the "PIII" name ceased to be used on any 100MHz processors.

On the 133MHz parts, the last one to carry the "PIII" name was the 1.2GHz version with a 256K cache. The faster 133MHz PIIIs (at 1.26GHz and 1.4 GHz) were dubbed "PIII-S" to allude to server applications... they had 512K caches. (There was also a 1.13GHz PIII-S.)

When the P4 Celerons made their debut (at 1.7GHz), Intel went back to a 128K cache.

[draco2527]

That Celeron 1200 is faster. It's a "Tualatin" Celeron, which is actually a true PIII with a full-size, full-speed 256K cache... running with a 100MHz FSB.

That PIII 667 does have the faster 133MHz FSB, but since the clock rate is only about half that of the Celeron, it will be slower.

Yes and no! Even do they have the same cache size as the PIII's the "instruction" sets within the processor are different (A celeron has less cache pipes-even if they have the same cache size as a TRUE PIII). A Celeron is a Celeron for more then the cache alone.....


EDIT: For the "older" celerons I would call them celery procs-almost worthless IMHO-. They have gotten a lot better and for the savings in price for the "newer" celerons most people will not see a difference.

[AlphaKilo470]

Celerons of the PIII era feature the same cores as PIIIs. You will only see marginal perforance difference between a 100mhz bus PIII and a 100mhz bus Celeron equivalent. I'm actually typing this post on an i1200 with a 500mhz Celeron and it performs for me every bit as well as my old PIII 500 desktop did in non cpu hungry apps.

If you really think a 667mhz PIII will outperform a 1.2ghz Celeron, then go ahead and try but I'm almost willing to bet my good left arm that the 1.2ghz Celeron will have better performance.

[tfflivemb2]

...I'm almost willing to bet my good left arm...

Do you have a bad left arm?

[AlphaKilo470]

Well, not "bad" per say but I have a few untested spares in my desk drawer still in the shrinkwrap. Reason I can't guarantee a "good" status is because of the unknown origin and the transport and original owner records are missing.

[draco2527]

If you really think a 667mhz PIII will outperform a 1.2ghz Celeron, then go ahead and try but I'm almost willing to bet my good left arm that the 1.2ghz Celeron will have better performance.

I am just saying compare apples to apples!! The celeron PIII era is NOT a PIII! That is all, I am sure the performance of the 1.2 Celeron is better then that of the 667MHZ PIII.

EDIT: An I was comparing same speeds on the post above, in refence to Celerons being PIII or whatever other proc with less cache!

A Celeron M that currently has 1MB of cache is NOT a BANIAS proc repackaged!!

[leoblob]

FWIW...

I ran the benchmarking utility CliBench MK III on the 1.4GHz PIII-S in the PC300GL that's in my signature. Then, I also ran it on the 1.4GHz (Tualatin) Celeron that's in a virtually identical PG300GL that I also own.

First number is for the PIII-S, second number is for the Celeron:

Dhrystone 2.1: 2917, 2912
Whetstone: 719, 718
8 Queen's problem: 4939, 4933
Matrix operations: 43884, 37393
Number crunch: 100599, 100556
Floating point: 10977, 10956
Memory Thruput: 175027, 143432 (PC133, PC100)

[dsigma6]

so the p3-s outperformed the celeron, but just barely in most cases. this is a good reference point, thanks for taking the time to do it.

since those numbers are so close at 1.4ghz each, i imagine the p3 667 would be destroyed (ok maybe not that bad) by the 1.2 celeron.

[AlphaKilo470]

I'd put my money on the 1.2 Celeron against PIII 667 just on the bare fact that you're facing a Tualatin core against a Coppermine core. Oh, and thanks for the benchmarks Leoblob, those are nice to have.

I'm thinking later on, I'll try running benchmarks on my P4 desktop and on my sisters Celeron desktop since both of our computers have similar CPU cores.

[leoblob]

Except for the matrix operations (and memory throughput, of course), I think the differences in the other benchmarks are within "measuring tolerance"... in other words, I would assume that the results are equal.

I don't understand very well what the "matrix operations" benchmark is testing... but I have seen that it's influenced by the (main) memory throughput. When I had CL3 PC133 SDRAM in that PIII-S, the matrix operations benchmark (as well as the memory throughput) both were lower... when I installed CL2 PC133 SDRAM, both went up.

My subjective comparison of doing standard office tasks on the PIII-S vs. the 1400C is that the PIII-S system is slightly faster. But I suspect the main reason in this case (office applications) is the faster memory... ??

I've often heard that the late Tualatin-based processors were faster than many of the earliest P4s ... ??

[pianowizard]

i have an old emachines t1220 with a 1.2ghz celeron, 100mhz fsb...i dont know about the cache. my friend gave me his old mobo w/667mhz p3 with 133mhz fsb.

That's a no-brainer: the 1.2GHz Celeron would be much faster than the PIII! If you were comparing a 1.2GHz Celeron with a 1GHz 133FSB PIII, then I **might** go for the PIII.