PIII 550 or Celeron 700

[bronibbear]

I am not particularly familiar with Intel CPUs but have recently had the opportunity to build system for a friend using second hand parts. I have a choice between a coppermine pIII 550 (100mhz FSB) and a celeron 700. Which is the better CPU? I know that its not necessarily the celeron as although its clock speed is quite a bit higher its small L2 cache and 66mhz FSB drag it down (particularly where gaming is concerned). However my friend is more concerned with using it for the internet in which case the celeron might perform better?

Your feedback would be appreciated.

 

[upec]

I would get the P3 500 and overclock it to 666MHz

 

[imgod2u]

It would depend on which P3 550. There's the Katmai based and the CuMine based. The later being better. You could easily overclock a 550E P3 to 733 or beyond and it would be significantly better than a Celeron 700.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[bronibbear]

Thanks for all your help. I'm going to get a new motherboard which is "Tualatin ready" as the motherboard I have is an old Pcchips one (a brand I've never had much faith in anyway). So if my friend isn't happy with which ever chip I give him he has the option of buying a Tualatin down the track. I'm opting for a sis630ST chipset as I need integrated graphics to save costs and the sis630 graphics engine is apparently almost as good as a geforce2.

 

[jclw]

Actually a Katmai 550 would outperform a Coppermine 550 in most apps, due to it's 512l L2 cache.

I'd take the PIII over the celeron.

- JW

 

[bronibbear]

Thanks for your imput, I think I'll try both processors and run a few benchmarks. I looked up the serial number of the PIII and it isn't a Katmai having only 256k L2 cache.

 

[imgod2u]

The Katmai's 512KB of L2 cache is running at half the CPU frequency and is off-die, giving it greater latency not to mention it only has a 64-bit data path to the CPU. The CuMine's 256KB of L2 cache is on-die and runs full speed at the CPU's frequency, not to mention has a 7 cycle access latency with data transfering every other clock. It also has a 256-bit datapath to the CPU. So it is significantly better than the Katmai.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[Crashman]

I don't think so, that 512k was running at half speed and with probably at least 1 cycle of latency.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[Crashman]

The Coppermine is a great core, a 550 will run at 733/133 under no stress. A user here named rcf84 had his 500E running at 1GHz! Normally the limit for most Coppermines is around 980MHz before you need extreme voltage and cooling.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>

 

[NurseMSIC]

I have a friend who really does know about CPUs to a frightening level. When i was looking to upgrade my AMD K6/2, he suggested me getting a second hand Katmai (512k) over a newer coppermine for my video editing because of the larger cache.
But if you have to buy yourself a new motherboard anyway - why not go for a P4 and even a second hand older 478 pin P4? You have lots of upgradability, and truth is a low end P4 will cost you the same as the P3s you are looking at anyway. The only extra cost would probably be RAM depending on what motherboard you get.
Basic motherboard, second hand P4 (478) and 256 MB RAM cost, i reckon around 200 dollars / 130 Pounds.

<A HREF="http://www.anandtech.com/mysystemrig.html?id=19557" target="_new">http://www.anandtech.com/mysystemrig.html?id=19557</A>

 

[imgod2u]

Well, video editing is a pretty streaming intensive so it's not very cache-intensive. The small amount of instructions usually used can easily fit into 256KB of L2 cache while the mass amounts of data are streamed directly from memory. I don't know exactly how well-educated your friend is but in this particular case, cache speed is more important by far.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[NurseMSIC]

Fair enough. I got a P4 Northwood in the end anyway, which is also what i recommend for the above guy.
It's an interesting point though about video editing - i dont know enough to argue about cache size vs cache speed, but video editing is primarily involved with two issues, one is the editing process itself which involves fast access to massive files (usually GBs at a time) and then there is the rendering issue which involves lots of sitting and waiting whilst the CPU computes its ass off. So one chip MIGHT (maybe?) be better at one task than the other.

<A HREF="http://www.anandtech.com/mysystemrig.html?id=19557" target="_new">http://www.anandtech.com/mysystemrig.html?id=19557</A>

 

[jclw]

SETI is a good example of an application that runs considerably faster on a Katmai then a Coppermine (clock for clock).

- JW

 

[bronibbear]

A P4 upgrade would be nice but I'm on a budget and because I live in Australia my dollar buys only half as much as yours! I already have the coppermine chips and sdram as left overs from upgrades I did for other people. So all I needed was a good mboard and I can build a computer for my friend as a present as he can't afford to buy one. If I got a P4 I would not only have to buy the mboard but CPU, a better power supply and probably rdram (as P4s apparently perform dismally with sdram).

 

[imgod2u]

As far as I know, Seti@home workloads are very small, easily fitting into 256KB of cache. Such a thing makes sense when you look at the Athlon vs P4 in Seti workload times (the Athlon having 256KB vs the P4's 512KB). If indeed Seti used workloads that would fit into 512KB but not 256KB, the P4 should be murdering the Athlon in Seti workload times having 256-bit, per-cycle access to pure cache and no memory access.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[jclw]

SETI workunits are 384k. SETI is pure floating point operations, which is why the Athlons chew through them.

- JW

 

[imgod2u]

It doesn't matter really. The Athlon can achieve at best, twice the peak FP throughput as the P4 in the best case for the Athlon (1 whole FP instruction per clock) and worst case for the P4 (half an FP instruction per clock). In the case of workloads fitting entirely into cache, that would be the best case for the P4. And with workloads not fitting into cache, it would be semi-worst case for the Athlon. Since you do not have room left for prefetch nor do you have enough room to store seti workloads. Now, while the Athlon's total cache is 384KB, keep in mind that 64KB of that is for instructions only. Seti workloads are data which means it only gets 320 KB at best, and even then it's not true because you need to reserve a little room for other processes such as OS functions.
You have the P4's FP unit filled to the breem with FP instructions allowing it to maintain almost a full 1 FP instruction per clock (not counting SSE/SSE2) vs an Athlon who has to wait countless clocks for fetching from memory the data that it needs. Even if the Athlon could achieve more than 1 FP instruction per clock (which is very very rare in any x86 code), it'd still have all those idle cycles.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.

 

[Dinski]

If you are ready to overclock, get PIII 550. If you are not going to overclock, you should consider exactly which kind of applications you will use. As you could suspect, in some application PIII 550MHz is faster.

<font color=orange>ÃÎËßÌ ÇÀËÚÊ ËÀÏÍÈ, ÃÎËßÌÀ ÙÀÍÃÀ ÂÄÈÃÍÈ!</font color=orange>