CPU gurus: would like advice on Celeron and Tom's Hware article please

[Rigoletto]

I was reading this http://www6.tomshardware.com/cpu/00q4/001113/smackdown-05.html and idly considering overclocking a Celeron on a BE6 I have, e.g. 633>950. Elsewhere in good ol' Toms he gives the impression such an overclocked celeron would perform like a PIII 800.
I was wondering in what situations would the clock speed of the celeron win out over the neutered cache? I don't know what memlatency and overhead really mean. Would a chess kernel, which I believe runs on integer maths, and is a fairly small program, not be handicapped by the cache changes?
And what kind of T-bird or Duron does a Celeron@950 perform like?
I must say I think the Tom's article a bit stupid:
a) he has chosen different clock speeds of both CPUs
b) he hasn't overclocked either of them
c) he has chosen the budget lines but the fastest of them. Inconsistent. Surely more logical would be to compare slowest budget CPUs with each other, and fastest premium CPUs with each other. DUH?😱

 

[whizler]

I sort of doubt that a Celeron 633 would even post at 950, much less run any appreciable length of time.

The comparison Tom makes between the Duron 800 and Celeron 766 is entirely valid. Both are their respective manufacturers' highest end CPUs in the "budget segment."

As for overclocking, what's the point? Everyone knows the Duron could reach 1Ghz+. You'd be straining a PIII if you tried that.

 

[Rigoletto]

at overclockers.com the database says celerons are indeed reaching 950, not uncommon.

 

[Diesel21]

Some people on this BBS has indeed done the celeron 633@950. Personally I am 566@850. it can go to 900 with no problems, but I don't need the extra megahertz so I will keep it at 850.

 

[Killer Ape]

I didn't read Tom's article, but from my own benchmarking I can say that on my BX6r2/128MB SDRAM/GF2 64MB a C533A@825 gave me almost identical Q3A, 3DMark 2000, and Sandra benchmarks as a P!!!700 (non-overclocked) that I got a hold of for an afternoon.

If you extrapolate that in VERY general terms, you need an extra ~100-150MHz of Celeron clock speed to equal the performance of a P!!! at the SAME (or very close in my case) FSB speed. So a C633 @ 950 would roughly perform like a P!!!800. Bear in mind this is based on an N=1 sampling of C2 vs. P!!! benchmarks in a relatively uncontrolled environment.

 

[squirrel dog]

I have two cbo cel2 633's.Both do 950,with no sweat.In fact I think most of the cbo cel 2 633's will do 950.Thats with stock cooling and at 1.70-1.75 vlts.Now the rest of the question.They will not compare to a simularly o/c duron,and a tbird would'smoke' a cel.On the syn.benchmarks they give some high marks,and in some of my games,they are real smooth.If you have a bx board,they are a reasonable choice at their price point.If you do not have a bx board,well I would go amd.I have 4 bx boards,so I will use them untill I get totally bored with them.My 633's came from onvia.com,they were 102,with a 20 off coupon that I got from fatwallet.com.

 

[Rigoletto]

thanks squirreldog. I knew the celerons were spanked by AMD but I wondered what the 950celeron was like...maybe an 800 Duron? Still, those durons often go to 1gig >>sigh<<
yep, if I had to buy a setup all over again it would be a duron.

 

[DaddyG]

I am not a fan of TOM but I though his artcile on the Celery vs Duron was pretty fair. He's been on an anti-Intel kick for a while so you have to ignore some of the spin. Its not unreasonable to consider the price of the processor when compiling these articles. Tom HAS done lots of Benchmarks with standard and OCed CPUs so I don't think he is misleading the results of the OC'd Celery. Durons will REALLY begin to sell when fully integreated mobos begin to appear. The big OEM's will jump on the Duron band wagon when the integrated mobos appear.

 

[Rigoletto]

yes I wrote on that in another thread... this should be so obvious but either AMD or their chums VIA can't get integrated solutions together can they?