Athlon XP 2500 (Barton) vs. Athlon XP 2400

[nogard]

Hi all,

I'm in the market for a new system and have pretty much settled on the nForce2 platform. Since I'm on a bit of budget and am looking at the "sweet spot", I had pretty much decided on the XP 2400 (266 MHz FSB) since there is such a huge jump in price right now to the 2600 (333 MHz FSB).

Then yesterday I browsed to newegg.com and noticed that that there is a 2500 Barton available for around $25 more than the 2400. But it is actually clocked at 1.87 GHz vs. 2.00 GHz for the 2400. Of course it has the Barton core advances which we all know about...

I don't suppose anyone will have any definitive answers on which is the better choice until we see a head-to-head (Anand? But I would welcome any opinions... Will the 333 MHz FSB make a difference at this "lower" clock speed? Is the 512K cache enough to make up for the lost clock speed?

cheers!

 

[coolred]

I have no technical data to support this, but the Mhz differance is only 133, but with the barton your getting the faster FSB speed and more cache, I think the barton will be at least as fast, but prolly quite a bit faster then the Tbred

 

[paralazarguer]

I dunno, the Athlon XP 2800+ (T-bredB) was faster in 10 ouf of 15 tests than the 3000+ (barton) so I would imagine that the 2400+ would be faster than the 2500+ aswell.

 

[BillStuck]

I am researching the same decision. Tomshardware.com has some benchmarks but the site seems to be down.

If I remember correctly they showed the 2400+ is a bit faster in then the 2500+, but they also overclocked the Barton. If you overclock...even mildly it looks like the additional L2 cache and the 333fsb of the barton really makes it pull ahead. I'm still undecided but I think I may end up paying $40 more for the Barton 2500+ and giving it a slight overclock. Since its allready running a 166fsb and has a larger surface area to disappate heat it should overclock to 2.1 with no problems.

But if you and I decide overclocking isn't the answer i think the 2400+ might be a better bet.

 

[paralazarguer]

Since its allready running a 166fsb and has a larger surface area to disappate heat it should overclock to 2.1 with no problems.

Umm...no. Actually, the extra cache and higher FSB makes it a much worse overclocker.

 

[butch84]

You must also remember that while the 2800+ tbred did beat out the barton 3000+, the 2800 was running on a 333fsb as well. The tbred 2400, however, does not. So, some of its advantage in clock speed will be negated right there - to be honest im not sure what i would do.

butch

 

[BillStuck]

well thats not what toms benchmarks are saying. The overclocked barton 3000+ is beating the P4 3.06 in allot of cases. They got 300-400mhz overclocks on 166fsb. Maybe you can't overclock it like crazy i.e. 1700+ but you keep a sync'ed memory cpu bus and it has a better chance of remaining stable because only the multiplier is changed.

On the barton 3000+ they took it from 13x166=2158mhz to 15*166=2490. I'd say thats not to shabby and although they didn't say I don't think they messed with the voltage either and were using stock air cooling in their benchmarks with this overclock.

 

[paralazarguer]

well thats not what toms benchmarks are saying. The overclocked barton 3000+ is beating the P4 3.06 in allot of cases. They got 300-400mhz overclocks on 166fsb.

Sure, but what kind of cooling do you think that Toms was using? Just don't expect to get your Barton 2500+ up to 2.1Ghz.

 

[BillStuck]

I wish his site would come back up. I really do think he said stock cooling but I can't remember. Why don't you think the barton can make 2.1ghz?

Thats only a change from 11x166 to 13x166....The other XP processors are working at far more aggressive settings then this.

 

[paralazarguer]

Everything I've read indicates that the Bartons are slightly worse overclockers than the T-bred B's. Keep in mind that toms had a handpicked CPU for review. This is not indicative of what you'll see.

 

[paralazarguer]

By the way, Tomshardware works fine for me. Must be a local problem for you.

 

[paralazarguer]

I believe this is the quote you're looking for. He has a hand picked CPU though:

One overclocking footnote: our test CPU was ideal for overclocking, remaining stable at 2500 MHz with conventional air cooling.

 

[BillStuck]

The site works now. Yea I didn't think of that...he has a hand picked cpu. Any links to where I can read up more on OC'ing bartons? I'm not finding to much in the archives.

 

[nogard]

Thanks for all the input...

I am not planning on overclocking the chip whichever way I go, since I'm just not as into squeezing the last drop of performance out of my machine as I once was. I think at this point I'm leaning towards the 2400 and saving the $50 or so (can get cheaper memory that way too).

Hopefully someone will benchmark them head-to-head though.

 

[nogard]

Whoops, didn't realize that Tom did to head-to-head benchmarks... great!

From a non-overclockers standpoint, its definitely a mixed bag. The 2500 seemed to do better in most office application/game benchmarks, but fell behind in encoding or other straight clockspeed dependant benchmarks as you would expect.

It may just have been enough for me to start leaning back towards the 2500 though...

 

[BillStuck]

it really is very close though. If you compare the 2400+ and the 2500+ barton in tom's benchmarks. Look at the frame rate differences for Quake and Unreal 2003. Percentage wise they are very very similiar. Its hard to say but for $50 less I'm leaning toward the 2400+