PDA

View Full Version : How Barton Stacks Up


WDHUFFMAN
02-12-03, 10:55 AM
http://www.anandtech.com/cpu/showdoc.html?i=1783&p=1

Its good on some but not so good on others.. Check it out.

DH

StealthHawk
02-12-03, 04:13 PM
Barton is about ~5% faster clock for clock than tbred. i think we can all agree the PR rating of Barton may also be a bit high.

-=DVS=-
02-12-03, 04:38 PM
I am dissapointed actually by Barton core , AMD could have increased they FSB better not cache , cache bearly gives any performance advantages :(

fastguy94416
02-12-03, 04:59 PM
Originally posted by -=DVS=-
I am dissapointed actually by Barton core , AMD could have increased they FSB better not cache , cache bearly gives any performance advantages :(


That may be, but cache is undeniably sexy ;)

StealthHawk
02-12-03, 05:09 PM
they should have made their cache more efficient.

then again, i'm sure AMD is concentrating more on Hammer. Barton is just a stop gap, just as i predicted.

Nemesis
02-13-03, 06:59 AM
Upping the FSB to 333Mhz is a stop gap as well, the Athlon core has run out of steam, and AMD are desparately wringing every last ounce out of it by any means they can. They should have released Barton with a 400Mhz FSB, and stuffed the P4. But of course, a 400Mhz FSB Barton makes the Clawhammer look bad, IMHO AMD are cooking the books with the PR ratings, holding them back.
I just hope the Clawhammer is out on time in September, because if it's not AMD are in trouble, even if Prescott is late, as rumoured.

Sazar
02-13-03, 08:17 AM
Originally posted by Nemesis
Upping the FSB to 333Mhz is a stop gap as well, the Athlon core has run out of steam, and AMD are desparately wringing every last ounce out of it by any means they can. They should have released Barton with a 400Mhz FSB, and stuffed the P4. But of course, a 400Mhz FSB Barton makes the Clawhammer look bad, IMHO AMD are cooking the books with the PR ratings, holding them back.
I just hope the Clawhammer is out on time in September, because if it's not AMD are in trouble, even if Prescott is late, as rumoured.

AFAIK... amd is making their upcoming 3200+ a 400mhz fsb component... and that will continue onto their hammer platforms...

they had planned on 333 mhz fsb for the hammers but changing to 400 is harder than it seems because of the on-die memory controller :)

the on-die memory controller will certainly help cut out and eliminate many of today's bottlenecks in applications :) hence resulting in faster overall performance...

Nemesis
02-13-03, 08:56 AM
Originally posted by Sazar
AFAIK... amd is making their upcoming 3200+ a 400mhz fsb component... and that will continue onto their hammer platforms...

they had planned on 333 mhz fsb for the hammers but changing to 400 is harder than it seems because of the on-die memory controller :)

the on-die memory controller will certainly help cut out and eliminate many of today's bottlenecks in applications :) hence resulting in faster overall performance...

I reckon they'll be using 400Mhz DDR memory come September...

The on-die memory controller is a double edged sword, I think. On one hand, it gives the benefits you mention, but on the other, keeping pace with memory technology is more expensive, as the core itself needs to be modified.

Sazar
02-13-03, 09:04 AM
Originally posted by Nemesis
I reckon they'll be using 400Mhz DDR memory come September...

The on-die memory controller is a double edged sword, I think. On one hand, it gives the benefits you mention, but on the other, keeping pace with memory technology is more expensive, as the core itself needs to be modified.

that is indeed a dilemma for amd... perhaps they will come up with an efficient new method to modify the memory controller ? (just flip it out and switch in an updated one in the fab ? )

who knows... but jedec did only recently approve of the ddr 3200 standard so perhaps amd was not looking that far ahead? for whatever reason the on die controller is both good and bad @ the same time :)

hopefully in my case I hope the cpu last me a good coupla years :)

StealthHawk
02-13-03, 04:34 PM
Originally posted by Nemesis
Upping the FSB to 333Mhz is a stop gap as well, the Athlon core has run out of steam, and AMD are desparately wringing every last ounce out of it by any means they can. They should have released Barton with a 400Mhz FSB, and stuffed the P4. But of course, a 400Mhz FSB Barton makes the Clawhammer look bad, IMHO AMD are cooking the books with the PR ratings, holding them back.
I just hope the Clawhammer is out on time in September, because if it's not AMD are in trouble, even if Prescott is late, as rumoured.

i don't really agree with that. increasing the FSB is paramount to maintaining and increasing performance. the higher the multiplier goes on the same FSB, the less power per MHz you're getting. increasing the FSB doesn't cost AMD much of anything, unlike cache.

also we have seen chipsets supporting DDR333 as well as 333MHz FSBs for a long time while AMD was still denying that they would increase the FSB on the Athlons.

Nemesis
02-14-03, 08:22 AM
Originally posted by StealthHawk
i don't really agree with that. increasing the FSB is paramount to maintaining and increasing performance. the higher the multiplier goes on the same FSB, the less power per MHz you're getting. increasing the FSB doesn't cost AMD much of anything, unlike cache.

also we have seen chipsets supporting DDR333 as well as 333MHz FSBs for a long time while AMD was still denying that they would increase the FSB on the Athlons.

I completely agree on this - the higher the FSB, the better, no doubt at all. My point was - the Athlon core has pretty much hit it's ceiling regarding clock speed, so AMD are upping the FSB and cache to maximise the core, and make it last out until Clawhammer. I was just wondering how they will scale the Barton this year, but still keep it inferior to the Clawhammer, which will debut with a PR rating of 3400+, according to most sources. I think if they pushed the FSB of Barton up to 400Mhz, and upped the core to it's max, the Barton may well have the potential to scale to 3400+ as well, which could harm Clawhammer sales.

All IMHO of course!

StealthHawk
02-14-03, 06:35 PM
oh i see what you're saying. i agree. AMD couldn't have gone much higher without increasing the FSB from 266.

however, i still wouldn't exactly call it a stopgap. the optimal solution would be having as high an FSB as possible and as low a multiplier as possible. that's how it was in the old days :) and when you upgraded your system it showed, big time.

Nemesis
02-17-03, 03:46 AM
Originally posted by StealthHawk
the optimal olution would be having as high an FSB as possible and as low a multiplier as possible.

When it comes to FSB overclocking, is it the CPU that limits the frequency or the motherboard? Or both? Just something that's bugged me for a while...

Smokey
02-17-03, 06:20 AM
Originally posted by Nemesis
When it comes to FSB overclocking, is it the CPU that limits the frequency or the motherboard? Or both? Just something that's bugged me for a while...

Dont quote me on this, but I think its a bit of both. Some motherboards will OC higher than others, as will cpus.I think if you dont lower the mutliplyer, you will reach the cpu limit, if you lower the mutiplyer, you will reach the motherboard limit?

Nemesis
02-18-03, 05:35 AM
This is getting a bit off the "Barton" topic, but I was wondering which would be the most fruitful FSB overclock - adding 10Mhz on to an Athlon's DDR FSB, which would add 20Mhz DDR, or 10Mhz on to a P4's QDR FSB, adding 40Mhz. It seems like I'm stating the obvious here, but is this why P4's overclock well, because a relatively small addition to the "base" frequency gives greater actual Mhz?

StealthHawk
02-18-03, 05:54 AM
Originally posted by Nemesis
This is getting a bit off the "Barton" topic, but I was wondering which would be the most fruitful FSB overclock - adding 10Mhz on to an Athlon's DDR FSB, which would add 20Mhz DDR, or 10Mhz on to a P4's QDR FSB, adding 40Mhz.

that's hard to tell, due to architectural differences. the P4 absolutely requires high FSB to get good performance.

It seems like I'm stating the obvious here, but is this why P4's overclock well, because a relatively small addition to the "base" frequency gives greater actual Mhz?

that would be correct.

Shinri Hikari
02-18-03, 04:33 PM
/off topic: If the other parts of the mobo's architecture were given as much attention as the processor, then we would see a performance boost unlike anything we have seen in a long time.:cool: :D :eek:
/on topic: It all depends on amd for their needs and what they think they can get away with. Their image as making new technology is a factor...:rolleyes:

StealthHawk
02-18-03, 04:51 PM
Originally posted by Shinri Hikari
/off topic: If the other parts of the mobo's architecture were given as much attention as the processor, then we would see a performance boost unlike anything we have seen in a long time.:cool: :D :eek:


what specific parts are you thinking off?

borntosoul
02-19-03, 06:01 PM
the athlon 64 might not be the super fast cpu everyone is expecting .well not the first batch anyway .i think it will be sold as a cpu of future potential more than anything else so competing with barton in 32 bit is not really that much of an issue here .similar to how the p4 was against the p3 . thats my 2c