AMD 2600+ REVIEW Still not #1!

[SSXeon5]

Originally posted by: 7757524
The conclusion from ace's sums it up fairly well:

Four out of seven gaming benchmarks proved to be faster on the Athlon XP 2600+ than on the 2.53 GHz Pentium 4. So, for gamers, the Athlon XP 2600+ lives up to its QS rating and will be a very attractive alternative considering its price.

However, the 2.53 Pentium 4 outperforms the Athlon XP 2600+ by a significant margin in typical workstation creative work. For those kinds of applications, AMD's platform will not outperform Intel's before the Hammer family arrives. We strongly suspect that the Athlon has enough firepower on board to perform well in CAD and 3D-modeling workloads, but that the AGP port and memory bandwidth of the current AMD platform is simply not up to par with Intel's. On the flipside, the Athlon XP 2600+ is clearly the fastest processor in the scientific workloads.

The 2600+ is about $300 Retail give or take, so ~$280 OEM, being that it is a small amount faster then a 2.53Ghz being priced at ~$240 Retail in 5 days. Another thing i want to bring up not bashing amd but the reviews. How they use a VIA chipset and then blame the performance on it when the 2600+ doesnt beat the 2.53GHz. Like here:

We are not absolutely sure what happened here. More testing is necessary, and our latest tests indicate that with some different BIOS settings we might get 5% better performance out of the Athlon XP. Nevertheless, it is clear that the Athlon XP 2600+ is not even close to the Pentium 4 in this kind of workload. We suspect that VIA's AGP driver and chipset implementation might not be so effective and rather poorly optimized for professional OpenGL applications. Typically these kind of applications move around huge amounts of geometry data, and therefore memory bandwidth and AGP drivers can make a big difference. Both systems were running at AGP 4x, though. In any event, the 2.53 GHz Pentium 4 is the clear leader in this benchmark.

Well Saint, you said I shouldnt blame the performance of a cpu on its chipset/memory, so van shouldnt either . Why did they use a KT333 and then blamed the performance on it, just boggles my mind :Q . Then ill be like van and blame the performance of the 2.53GHz PC1066 because Granite Bay isnt out and RDRAM sux j/k lol. And the ocing is a bit better but I still see close scores between the 2.4GHz XP and 2.53GHz P4. I still cant wait for the i845EP and 2.8GHz in 5 days

SSXeon

 

[paralazarguer]

Well Saint, you said I shouldnt blame the performance of a cpu on its chipset/memory, so van shouldnt either

What van? That's from Ace's hardware. They had to use that chipset because it's the fastest. If you're not going to factor in chipsets then how are you going to benchmark? You take the fastest chipset for each platform to determine which platform is "best."

 

[SSXeon5]

Originally posted by: 7757524

Well Saint, you said I shouldnt blame the performance of a cpu on its chipset/memory, so van shouldnt either

What van? That's from Ace's hardware. They had to use that chipset because it's the fastest. If you're not going to factor in chipsets then how are you going to benchmark? You take the fastest chipset for each platform to determine which platform is "best."

My mistake it was Acehardware Well they sounded like if the XP lost it was the chipsets falt. If the KT333 is the best then they shouldnt blame the AGP driver for its performance drop, just MHO

SSXeon

 

[CrazySaint]

Originally posted by: SSXeon5

Originally posted by: 7757524
The 2600+ is about $300 Retail give or take, so ~$280 OEM, being that it is a small amount faster then a 2.53Ghz being priced at ~$240 Retail in 5 days. Another thing i want to bring up not bashing amd but the reviews. How they use a VIA chipset and then blame the performance on it when the 2600+ doesnt beat the 2.53GHz. Like here:

We are not absolutely sure what happened here. More testing is necessary, and our latest tests indicate that with some different BIOS settings we might get 5% better performance out of the Athlon XP. Nevertheless, it is clear that the Athlon XP 2600+ is not even close to the Pentium 4 in this kind of workload. We suspect that VIA's AGP driver and chipset implementation might not be so effective and rather poorly optimized for professional OpenGL applications. Typically these kind of applications move around huge amounts of geometry data, and therefore memory bandwidth and AGP drivers can make a big difference. Both systems were running at AGP 4x, though. In any event, the 2.53 GHz Pentium 4 is the clear leader in this benchmark.

Well Saint, you said I shouldnt blame the performance of a cpu on its chipset/memory, so van shouldnt either . Why did they use a KT333 and then blamed the performance on it, just boggles my mind :Q . Then ill be like van and blame the performance of the 2.53GHz PC1066 because Granite Bay isnt out and RDRAM sux j/k lol. And the ocing is a bit better but I still see close scores between the 2.4GHz XP and 2.53GHz P4. I still cant wait for the i845EP and 2.8GHz in 5 days

SSXeon

Well, actually what I said is you shouldn't blame the performance of the P4 on the use of PC800 instead of PC2700 (which is about the same speed). If they had been using SDRAM or some crap, then I'd agree that they're morons Btw, I think Ace's actually used DDR on their P4 benchies and I actually agree with you that blaming the benchmark results on the KT333 chipset is stupid (particularly when the KT333 is the fastest chipset for the XP, at the moment). I think at this point its getting to be pretty obvious that the AMD needs more memory bandwidth. The P4 nearly always beats the XP on memory-intensive benchmarks.

STILL waiting for a review from AT, I'm surprised they don't have one up, yet :frown:

 

[SSXeon5]

Originally posted by: CrazySaint

Originally posted by: SSXeon5

Originally posted by: 7757524
The 2600+ is about $300 Retail give or take, so ~$280 OEM, being that it is a small amount faster then a 2.53Ghz being priced at ~$240 Retail in 5 days. Another thing i want to bring up not bashing amd but the reviews. How they use a VIA chipset and then blame the performance on it when the 2600+ doesnt beat the 2.53GHz. Like here:

We are not absolutely sure what happened here. More testing is necessary, and our latest tests indicate that with some different BIOS settings we might get 5% better performance out of the Athlon XP. Nevertheless, it is clear that the Athlon XP 2600+ is not even close to the Pentium 4 in this kind of workload. We suspect that VIA's AGP driver and chipset implementation might not be so effective and rather poorly optimized for professional OpenGL applications. Typically these kind of applications move around huge amounts of geometry data, and therefore memory bandwidth and AGP drivers can make a big difference. Both systems were running at AGP 4x, though. In any event, the 2.53 GHz Pentium 4 is the clear leader in this benchmark.

Well Saint, you said I shouldnt blame the performance of a cpu on its chipset/memory, so van shouldnt either . Why did they use a KT333 and then blamed the performance on it, just boggles my mind :Q . Then ill be like van and blame the performance of the 2.53GHz PC1066 because Granite Bay isnt out and RDRAM sux j/k lol. And the ocing is a bit better but I still see close scores between the 2.4GHz XP and 2.53GHz P4. I still cant wait for the i845EP and 2.8GHz in 5 days

SSXeon

Well, actually what I said is you shouldn't blame the performance of the P4 on the use of PC800 instead of PC2700 (which is about the same speed). If they had been using SDRAM or some crap, then I'd agree that they're morons Btw, I think Ace's actually used DDR on their P4 benchies and I actually agree with you that blaming the benchmark results on the KT333 chipset is stupid (particularly when the KT333 is the fastest chipset for the XP, at the moment). I think at this point its getting to be pretty obvious that the AMD needs more memory bandwidth. The P4 nearly always beats the XP on memory-intensive benchmarks.

STILL waiting for a review from AT, I'm surprised they don't have one up, yet :frown:

Well HardOCP did the 2.48GHz XP with DDR333 and 333Mhz fsb, thats 2.7GB/s bandwidth and still stayed close to the 2.53GHz P4. And I know, where the hell is ATs REVIEW!!!!!!!!!!!!!

SSXeon

 

[paralazarguer]

I don't understand why you're saying that you don't understand why they're knocking the chipset. The fact is that the KT333/KT400 is the fastest chipset for the XP. If that chipset bottlenecks it then it's the same as the processor being that slow since you cannot buy a faster chipset for it. In this case, the chipset is a limitation of the XP platform.

 

[paralazarguer]

With the Thoroughbred Revision B core AMD has given new life to the Athlon XP, and it couldn't have come at a better time

That's a quote from anand's review. couldn't have come at a better time? How about a couple of months ago? He seems to believe that it's a paper launch akin to the P3 1ghz and that implies that we may not see them for a while.

 

[PG]

Hey guys, AMD is NOT using SOI for these chips...they are not the Barton core. Check out the reviews.

 

[paralazarguer]

AMD has a new technique, most likely involving SOI, and clockspeeds are going to ramp quickly. For AMD, a new product cycle is just beginning.

This guy obviously didn't take the time to read a review or didn't understand it. There is no SOI. It's only an extra layer. The benefits of this are explained in every review. This also leads to higher manufacturing costs therefore higher priced chips. SOI will not give AMD an edge. Strained silicon that Intel will use in Precott is superior to SOI. SOI would not benefit prescott at all although it may be implemented on top of strained silicon in a later processor.

 

[RideFree]

OK!
Everyone start your engines!
The race is on (after the yellow flag for the AMD interlude)!

Hmmm!
(Now just how fast can I load that two page document in Word?)
__________________________
Hey, just pulling everybody's chain. Haven't you heard? Speed kills!

It's killing my pocketbook! I need two 1.8a Northwoods & a hundred Phenobarbital

It's quite simple...we are all hopeless addicts

 

[Jace]

Anand wrote a pretty good summary:

With the Thoroughbred Revision B core AMD has given new life to the Athlon XP, and it couldn't have come at a better time. The Athlon XP 2600+, for the most part, offers performance competitive with the Pentium 4 2.53GHz; the same can be said about the Athlon XP 2400+ and the Pentium 4 2.4B. There's no clear performance advantage in either case for the vast majority of applications, but where AMD does hold the advantage is in price. The Athlon XP 2600+ and 2400+, once available, will retail for significantly less than their equivalently performing Intel counterparts.

Strange, I didn't know that 3com was making AMD's chips now... (look at the manufacturer Anand chose in the article title ;-)

 

[KhoiFather]

AMD ROCKS BABY!!! Intel can't compare to their price and performance!!!

 

[Jeff7181]

I would say AMD is ahead of Intel. Maybe not in sheer speed, but when you consider the cost of Intel's extra 5 frames per second, I don't think it's worth it.

 

[dew042]

I would not trust tom's hardware for any hw based review. they are heavily intel slanted.

dew.

 

[BowlingNut]

what i'd like to see if some max-oc'ed 2600/2400 against 2.53/2.4b's. just for fun and all.

again, i'd like to reiterate what i've been saying for a while now. p4's are faster in some places....xps are faster in others - just pick one and stop arguing.

 

[paralazarguer]

I would not trust tom's hardware for any hw based review. they are heavily intel slanted.

dew.

Okay, this is a flast out dumb statement and Tom himself has addressed it countless times. Tomshardware is not Pro Intel. They are pro whatever is best. When AMD was doing well they were most definitely favoring AMD. When the P3 1.13 came out they were the first to pursue it and prove it unstable. When the Thunderbird was king and the XP was new, toms was accused of being pro AMD. They're not pro anyone, they're unbiased.

 

[Vic]

Originally posted by: 7757524

I would not trust tom's hardware for any hw based review. they are heavily intel slanted.

dew.

Okay, this is a flast out dumb statement and Tom himself has addressed it countless times. Tomshardware is not Pro Intel. They are pro whatever is best. When AMD was doing well they were most definitely favoring AMD. When the P3 1.13 came out they were the first to pursue it and prove it unstable. When the Thunderbird was king and the XP was new, toms was accused of being pro AMD. They're not pro anyone, they're unbiased.

Actually Kyle Bennett at HardOCP was the first to discover that. Tom's stealing that claim from Kyle (and a very unfortunate remark by Tom on 9/11/01) is probably why Kyle (and most other Americans) hate Tom so much. The anti-AMD stand comes from Tom's videos of Athlons burning a year or so ago, right as AMD was beginning to implement thermal protection, that caused AMD stock price to drop. No one loves Tom. He made his bed and now he has to lie in it.

edit: oh yeah, and my comment you bashed was from Tom's review which, at the time (and it was one of the very few out at that time), didn't mention the new layer but a new "interconnect technique." Obviously, when you read crap, you get crap conclusions, eh?

 

[apoppin]

Originally posted by: dew042
I would not trust tom's hardware for any hw based review. they are heavily intel slanted.

dew.

Then read ANAND'S REVIEW

While we don't approve of AMD pushing for essentially a paper-launch of these faster Thoroughbred parts, the CPUs are a welcome addition to the Athlon XP line. With the Pentium 4 still on track to hit 3GHz by the end of this year, and with a Q1-03 introduction of Hammer being very likely, it may be up to the Athlon XP to close the year out for AMD.

You can still get a 1.6A/1.8A @ 2.4Ghz for far cheaper than the "paper launched" XP 2400. However, it looks like the modified XP 2400 will ALSO overclock nicely. Perhaps Intel still holds the lead in O/C'ability (perhaps not). But it is really nice to see AMD actually competing with Chipzilla again. Intel's price drop is less than 2 weeks away and I'd guess it is at least partly due to the pressure from AMD.

 

[SupermanCK]

i don't think the amd would be significantly cheaper than intel's once intel does drop it's prices on sept 1st as stated...$240 for a 2.53GHz...
unless AMD is going to sell their flag chip at $200 or less...just my 2 cents

 

[dexvx]

Yea, I dont know whats up with all this criticism of Tom. I think everyone hates him because he is a radical. He was the first to denounce 3dfx (god forbid) during the voodoo 2 (thats right TWO) era. He is just a little wacko and favors mostly the underdog.

For the most part, they (2600+ and p4-2.53) are even, give or take. Hothardware's pricing is obviously flawed because by the time the 2600+ comes in volume, 2.53-p4's will be LOWER in price in 1000 unit quantities. Either way, zealots for any company makes that company look bad.

As it is, facts are facts. AMD has a new toy and it hangs with the best Intel has to offer right now. I love it, because it only means that cpu prices are going down again (which is why I root AMD on as I VERY clearly remember Intel dominance and $900 cpus ). And the battle between the two companies is so much fun to watch.... the drama, I love it.

Thats odd, I remember AMD charging $1299 for a 1Ghz Athlon. I dunno about you, but thats the highest priced desktop chip I've ever seen.

As for SOI, that is overrated hype for now. Go read some actual technical references first.

 

[alexruiz]

Guys, I am not going to quote anyone and I will try to make it short. I wanna present COLD minded reasons, so fanboys (either side) please go out and take a freshening breath.

After reading all the reviews, it seems it is a tie. Please notice also that the results are way too different between all the reviewers.... (divx 5.02 is the biggest example, a site got over 50 fps for the P4 2.53 while other got a more realistic low 30......)

I have talked a lot about benchmarks in the past. Let's be critical and try to discriminate how valid a benchmark is. It doesn't matter if your CPU sweeps the competition if that is a category no one uses. Also, if a benchmark favor one approach over the other the result is not valid. Finally, if a benchmark is not considered as "stressing" but it is very popular, then it can considered as valid (unreal, half life)

My clasification of fairness would be:
AMD biased: optimized for 3d now! only (not pro)
intel biased: optimized for SSE (some cases), SSE2
fair: optimized for SSE (some cases), not optimized.

In some sites, the 2600+ won the majority of tests, while in others the P4 was the clear victor..... Why?? Test beds, tweakings and methodology, including benchmarks.

I was honestly surprised to see such a variation is results.

So I will evaluate the becnhmarks, but please feel free to disagree.

1) Sysmark: Do you believe it???
2) PCmark: Do you believe it?
3) Quack, I mean Q3: It has been told a lot it depends heavily on "memory bandwidth", but when a P3 edged a Tbird easily on the same RAM, and the same tbird with DDR barely beated the same P3 with SDR.... where is the "heavily" factor?? (Please check older reviews) Carmack himself acknowledged the use of "some" SIMD optimizations.
4) Content creation winstone 2002: It is the same that CC 2001, with the exception thath WME is included in the 2002 version.... if you have the 2001 version, run some benches. The AXP will win with no problem... in the 2002 version the P4 wins...
5) 3dmark: Not my favorite, but fair enough. P4 wins.
6) Comanche 4: new, fair enough. P4 wins.
7) Serious Sam: no optimizations, maybe unfair because of that. XP wins.
.
.
.
.

The list could keep going on, but the point is there.

The Winner is:

The CPU that suits your needs the best.

Use a lot of WME??? P4
Play RTS games?? XP
Play simulators? XP
Do a lot of video using optimized software?? P4
Do a lot af CAD? XP
Play older games? XP
Pkay newer optimized games (comanche, aquanox)?? P4
Play new games no optimized?? XP

The situation is the same than in Nov 2000, the P4 wins some, the Athlon wins others. How was a winner decided??? Based in the number of bechmarks they won, and how close they stayed in the benhmarks that lost. Let's be critical.

I hope this stays as a civilized discussion.

SSXeon5, I saw you are upgrading your system. I hope you remember our challenge, your granite bay + prescott against the hammer of my choice.... or are you going to upgrade again in a few months????

 

[paralazarguer]

Yeah, I'm not sure what's up with these anti-toms guys. Regardless of who encountered the 1.13ghz Intel problem first, the fact remains that toms harped on it and bitched and moaned about it. Then they called the willametter a POS and pumped the athlon. If you can still call them biased against AMD after all that then you're quite biased yourself. I have no idea what comment you're talking about on 9/11 nor do I care. Any comment that you found offensive has nothing to do with the fact that toms gives good reviews to intel and very, very bad reviews to intel. Same to amd.

 

[apoppin]

I find the reviews at TomsHG and AnandTech to be very complementary. Tom goes where Anand doesn't (O/C'ing and more history details).

I don't find Tom to be anymore "biased" than Anand (who is decidedly non-biased). Tom is "opinionated" (which is a big difference and he states it as so).

Did you anti-Tom guys even read the article. His conclusion is pro-AMD:

Compared to its predecessor, the Thoroughbred "A," this one shows above all that a significantly higher clock rate (this top model now runs at 2133 MHz) automatically means greater speed. In the benchmark tests, the Athlon XP 2600+ manages to surpass the Intel Pentium 4/2533 once more, but not in all disciplines.

This brings AMD back in the running. These excellent results were achieved by modifying the CPU core. . . .

Intel P4's seemingly permanent performance lead has been broken. AMD's CPU has succeeded in becoming a powerful rival to Intel's top P4 model. At the same time, the power consumption has been lowered, allowing more room for higher clock speeds. Above all, Intel should brace itself for the near future, because successor of the Thoroughbred "B," the Barton core, will work with a double L2-Cache. To top it off, the identical interconnect technique will be used as it is with the latest T-bred.

And we see from BOTH reviews that the XP2600 is worthy of its PR rating.

 

[Acanthus]

Originally posted by: PSYWVic

Acanthus, get a clue. The Hammer was never due before '03. So how it can be "10 months late already" is WAY beyond me. And it won't even have an FSB, so I guess that takes care of that problem, don't it? Even though that FSB problem has absolutely nothing to do with their chipsets...

FSB has a lot to do with AMD decisions about chipsets, my GUESS (yes GUESS) would be that if AMD continued to design their own chipsets like they used to, It would be much easier to adopt FSB increases on the CPUs, because this isnt the case. AMD has to send out new athlon specs to VIA and SiS and ALi and hope one of them comes up with a working chipset in a few months.

Let me stress that I own 2 AMD rigs, im far from an Intel fanboy and i love seeing AMD competitive. But if i hear that stupid clock for clock thing one more time im gonna freak, OF COURSE ITS FASTER CLOCK FOR CLOCK, Intels pipeline is twice as long!!!!

Anands review sheds a little more light on how an average AMD rig would go against and average Intel rig. And its close, real close.
Only thing that i wonder is why they used PC800 instead of 1066, no biggie though, PC800 is comparable to PC2700 in performance and is much lower in cost, so it makes the systems more comparable in my view.

As for whoever said "the price to performance ratio is the best" the AMD Athlon XP 2600+ will be more expensive than the Northwood 2.53Ghz. As for overall performance, dont forget that Intel is launching the 2.5, 2.6, 2.66 (533), and 2.80 (533) cpus.

And im searching for linkys to a much earlier than today expected release date of hammer, ill be back with those.

 

[lookin4dlz]

First of all, in counting up the wins by non-overclocked chips the P4 2.53MHz chip wins about 61% of the comparisons with the AMD 2600+.

Additionally, the Tom's review shows a maxed-out overclocked AMD chip versus an Intel chip that's not maxed out.

Finally, if you've been reading Tom's for very long (and I've been reading that site since he started it) you'd know that he is VERY PRO AMD! I even did a word analysis of several articles to show the heavy use of negative connotated words used for Intel versus those used for AMD. If you look at Tom's bnechmarks, they usually pit an AMD chip that is running at a high fsb or with faster ram or overclocked to the max versus an Intel chip at stock speed.