Whats the difference between

[Apex]

a "Clawhammer" skt 754 3200 and a "Newcastle" skt 754 3200 ?

[Firelord]

3200+ Clawhammer = 1MB Cache 2.0GHz
3200+ Newcastle = 512KB Cache 2.2GHz

[Apex]

Wud the clawhammer be a good buy ?

[GrahamGarside]

Depends, in some cases the L2 cache makes little differance thanks to the on chip memory controller, and so chips of the same clock speed perform about the same regardless of L2 cache, which means a newcastle 3200+, running at 2.2ghz compared with just 2ghz for the clawhammer would have the advantage.
But the are exceptions, especially when the cache is 1mb where it has advantages. all in all I would say the newcastle was teh better chip, at stock you get a faster clock, and the lower L2 cache helps with overclocking.

This page shows the relative performance differance between teh same clocked cpu's with differant L2 sizes

http://www.xbitlabs.com/articles/cpu...on-2600_5.html

[Firelord]

I prefer 1MB Cached CPUs and do find them quicker in everything I do.

I would say get the 3200 Clawhammer as long as its a CG Stepping as they have the newcastle memory controllers but 1MB Cache

I have had a 3200 C0(1MB), 3000 CG(512KB) and 3400 CG(1MB) and the 3400 is the best clocker i've had

Great chip

[Apex]

Well theres no way for me to check the core stepping, though the guy does state they are the 1mb cache version.

supose i could email and ask what the core the chips hes got are.

[Firelord]

Get the codes off the top

This is a CG ADA3200AEP5AR
This is a C0 ADA3200AEP5AP

[Apex]

I'll ask him for some picks and see what happens

either way it will still be a nice small upgrade for me from my sn45g v2 and 2500 xp-m

[StormPC]

In almost every case ClawHammers eat Newcastles alive. In other words, CPU cache is good! Ask any P4EE owner what a bit of cache can do for a CPU.

[GrahamGarside]

Yeah it makes a decent differance on the P4 where there is limitted bandwidth, but with an A64 having an onboard memory controller it makes little to no differance, 8% is about the most you will see and thats mostly in games at settings where the gpu isn't being the bottleneck
In things like video encoding, compression etc.. it's more like 5% differance. And thats at equal clock speeds, you add another 200mhz on which is what amd does for the same rating newcastles and you will match a clawhammer in those apps it previously came out top and exceed it in those that were equal.

[StormPC]

Yeah it makes a decent differance on the P4 where there is limitted bandwidth, but with an A64 having an onboard memory controller it makes little to no differance, 8% is about the most you will see and thats mostly in games at settings where the gpu isn't being the bottleneck
In things like video encoding, compression etc.. it's more like 5% differance. And thats at equal clock speeds, you add another 200mhz on which is what amd does for the same rating newcastles and you will match a clawhammer in those apps it previously came out top and exceed it in those that were equal.

"Little to no"???? How would you like an 8% pay cut?

Only 8%? You do realize there are people who have taken a blowtorch to their motherboards and videocards for less than 8%. By the way, 8% is about the average for ALL applications. In CPU intensive applications it's usually more like 15-20%.

If you've ever overclocked an A64 you know that 200MHz equals a HUGE performance increase. So the fact that AMD has to clock 512k chips 200MHz higher to get a rating equal to that of a 1MB chip should tell you something.

Give it up man. Cache rules!

[GrahamGarside]

I guess I'm fussy over practicle gains I don't tend to bother overclocking unless I can get around 20% and I'm not truly happy unless I see 50%, I guess I was spoilt back in the days of AXIA tbirds

[StormPC]

You must understand that ALL A64's scale like crazy. When you overclock a 1MB chip vs a 512k chip the effects are basically doubled. All of the newer chips LOVE cache though!

[Firelord]

2.6GHz 512KB chip - 28500 3D Mark 2001SE
2.6GHz 1MB Chip - touch under 30000 3D Mark 2001SE

Both where running same clocks (10*260) and same memory etc

Encoding took alot less as well

[madwelshboy]

RE: 2.6GHz 512KB chip - 28500 3D Mark 2001SE
2.6GHz 1MB Chip - touch under 30000 3D Mark 2001SE

Yes, but 3D Mark Scores are not true indications of real world applications.

E.G: 2.6GHz 512KB chip = 45.5fps (DOOM3)
2.6GHz 1MB Chip = 47.2fps (DOOM3)

Now thats about 8%, but I'll be damned if you could notice that difference with the naked eye !

It all comes down to cost and upgradeability and whether you plan to overclock I suppose. If you can afford the 1m cache chip and plan to overclock go for it. If you cant afford it or dont plan to overclock, go for the 512kb chip ?!?

[Apex]

So then if i can get 1mb cache clawhammer skt 754 i should

wud make a nice nice little shuttle box for me