Sign in with
Sign up | Sign in
Your question

AMD Unviels World's Highest-Performing Processor

Last response: in CPUs
Share
August 21, 2002 6:15:15 AM

<A HREF="http://biz.yahoo.com/bw/020821/202512_1.html" target="_new">http://biz.yahoo.com/bw/020821/202512_1.html&lt;/A>
for you amd fanboys out there. enjoy it. good for amd.

<A HREF="http://biz.yahoo.com/rf/020820/tech_intel_pentium4_1.ht..." target="_new">http://biz.yahoo.com/rf/020820/tech_intel_pentium4_1.ht...;/A>for you intel fanboys out there. enjoy it. good for intel.

pricewatch doesn't even have xp 2400+ or 2600+ yet, but the p4 2.8 is there. anyways, good to finally see progress.

<font color=green> there's more to life than increasing its speed -Ghandi</font color=green>
August 21, 2002 6:51:20 AM

Vindication.

:wink: The Cash Left In My Pocket,The BEST Benchmark :wink:
August 21, 2002 1:46:45 PM

Um, looking through tom's benchmarks, it looks to me that the 2.53 was still leading in more benchmarks. But anyways, its nice to see AMD finally do something about the performance gap.

<font color=blue>Unofficial Forum Cop</font color=blue>
Related resources
August 21, 2002 1:53:20 PM

And plus, the 2.53B will be $243, therefore less than the ~$300 the XP2600+ will be.

<i>Past mistakes may make you look stupid, but avoiding future ones will make you look smart!</i>
August 21, 2002 2:00:08 PM

Quote:

And plus, the 2.53B will be $243, therefore less than the ~$300 the XP2600+ will be.

AMD will probably lower the price accordingly. Since I have RDRAM now, my next processor will probably be a 3GHz+ P4 with Hyperthreading or a Prescott.

Intelligence is not merely the wealth of knowledge but the sum of perception, wisdom, and knowledge.
August 21, 2002 4:46:32 PM

AMD is lying, I smell a class action suit.

"<b>AMD/VIA!</b>...you are <i>still</i> the weakest link, good bye!"
August 21, 2002 5:04:24 PM

What I find interesting is the Tom talks like the 2600+ completely destroys the P4 2.53... But from looking at those benchmarks it doesn't really seem that great.

That and yes chuck is right, the 2600+ is more expensive.

If you look at the article tom wrote, during their tests they O/Ced it to 2400mhz (or 3000+) but that was using the "best water cooling system available" which most people probably can't afford anywys so I doubt most normal end uesrs will be able to reproduce such results. Maybe they should have tried O/Cing it using more conventional means for more regular joe blow users.

So I don't see how the 2600+ which is more expensive, slower and being pushed closer to it's limit is better than a p4 2.53...

Cooj
August 21, 2002 5:07:06 PM

Yeah, they just said they have the best processor.

<font color=blue>Unofficial Forum Cop</font color=blue>
August 21, 2002 5:12:29 PM

Hey go over to the "Comments on XP2600+" or whatever thread. I commented on that... :smile:

<i>Past mistakes may make you look stupid, but avoiding future ones will make you look smart!</i>
August 21, 2002 5:19:31 PM

That's what we call bias. When the 2.53 p4 came out and was undoubtedly the fastest chip, Tom's article was that basically it barely beat the Athlon, even though it led in every test. Now that the 2.53 (or at the least the 2.8) p4 is neck and neck with the 2600, he treats it as Intel is dead, long live AMD...I don't know why i even bother reading these articles anymore. I should just go to www.amd.com if i want this type of "news".

Athlons and Pentiums are just melted rock. Who’s rock is better? Who cares, let’s play some games
August 21, 2002 5:34:24 PM

Quote:
What I find interesting is the Tom talks like the 2600+ completely destroys the P4 2.53... But from looking at those benchmarks it doesn't really seem that great.

THG never said AXP 2600+ compltely destroys P4 2.53. They said-

<b> In the benchmark tests, the Athlon XP 2600+ manages to surpass the Intel Pentium 4/2533 once more, but not in all disciplines. </b>
August 21, 2002 6:17:27 PM

But you see, it only beats it in a few occasions, the P4 2.53 still leads in most apps, yet they say the XP2600+ manages to surpass the 2.53, as though it leads most of the time.

<i>Past mistakes may make you look stupid, but avoiding future ones will make you look smart!</i>
August 21, 2002 7:29:34 PM

you know, the AXP 2600+ prolly would pass the 2.533Mhz performance and leave it in the dust <b>IF</b> it had a "real" chipset backing it up!</font color=red>

<b><font color=orange>sing to prolong HDD life; spin right round like a record baby Right round round round
August 21, 2002 8:55:55 PM

how come everyone is talking about 2600+ vs 2.533 when there's 2.800 is just around the corner.



<font color=green> there's more to life than increasing its speed -Ghandi</font color=green>
August 21, 2002 9:00:55 PM

There's a 2.8 around the corner, but then there's also faster and more powerful xp's based on the barton core "Around the corner" as well. While I said in a different thread, this athlon is no P4 killer, for it running 400+MHz slower, half the FSB speed, half the cache and a generally sh!t chipset it's pretty damn good.

"Opinions are like assholes, everyone has one"
a b à CPUs
August 21, 2002 11:27:10 PM

If AMD leads by 1% in 7 out of 10 test, but looses by 10% in 3 out of 10 test, that still leaves them trailing even if they won more test.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 21, 2002 11:40:55 PM

Maybe, maybe not that would depend on the tests themselves and how much they translate to real world usage. Head over to anandtech or Ace's and you will see it all comes down to what test you choose to use. Its not a P4 killer, but the current p4 can't claim undisputed rights by any means. Give it a better chipset (aka nfore2), and things could really start to get interesting again. As for those p4 users hoping to upgrade to higher speed processors themselves, it looks like a new motherboard is in there future as well. <A HREF="http://www.digitimes.com/NewsShow/Article.asp?datePubli..." target="_new">http://www.digitimes.com/NewsShow/Article.asp?datePubli...;/A>

It's not what they tell you, its what they don't tell you!
August 22, 2002 2:19:00 AM

um... what to say?...

for the same reason you wouldn't race a sentra with a McLarenF1

its kinda nice to see which CPU does what better at the same "speed rate". there is a point in racing cars of the same class, but it would be pointless to race a sentra with a McLarenF1 you see... i mean you would know the outcome!

<b><font color=orange>sing to prolong HDD life; spin right round like a record baby Right round round round
August 22, 2002 2:40:15 AM

"Around the corner" for Intel is about a week from now......you can already order the 2.8ghz. Wont ship till months end. Be a good while before you can get your hands on a XP2600, much less a Barton.....

This sig runs too hot.
August 22, 2002 2:46:30 AM

You are right and once again THG's comments just don't match the evidence.

Crash, you have been saying all along how DDR (even fast DDR) kills the P4's performance. Now I was just looking at <A HREF="http://www.aceshardware.com/read.jsp?id=50000284" target="_new">Aceshardware.com's review of the XP 2600+</A> where it fares far better (vs P4 2.53) but in this case both platforms use DDR. When compared to THG's review this seems to prove your claim, at least I think it does. Do you concur that the differing results of the two reviews can be explained by P4's memory bandwidth differences (RDRAM in one case, DDR in the other)?

<b>I have so many cookies I now have a FAT problem!</b>
August 22, 2002 3:57:59 AM

As Crash would say: We need Dual Channel DDR to even match and outperform RDRAM based P4 systems.

--
Is the opportunity to earn money by working, free?
August 22, 2002 3:59:22 AM

Quote:
Vindication.

Why do I always crack and ROFL when I read that word from your posts?!

--
Is the opportunity to earn money by working, free?
August 22, 2002 4:00:49 AM

The XP2600 WILL cost less than the 2.53GHZ. As I stated before, retailers always sell AMDs below MSRPs, while they sell overprice Intels.
And that is why my claim, that AMD's price/performance at stock, will NEVER be beaten

--
Is the opportunity to earn money by working, free?
August 22, 2002 4:06:39 AM

Yeah, I remember that. Wasn't SIS working on a chipset with dual channel DDR support for the P4? Is DDR-II dual channel?

<b>I have so many cookies I now have a FAT problem!</b>
August 22, 2002 4:08:41 AM

First of all, the Athlon's current performance limitation seems to be its L2 cache. As the CPU scales, its L2 cache, with is relatively high latency (pipelined may I add), would not scale as fast as the CPU would. I think we're seeing it reach beyond the "sweet spot" where the L2 cache with 8 cycle latency is still useful, but not a bottleneck. But yes, we're also getting into memory limitations, although I doubt dual channel DDR is really needed. I'd say something along the lines of a bus increase to 166-200MHz DDR.
As for the price/performance thing. If you're talking about vendors, they actually charge a lot less for the CPU they sell in their machines than you'd find retail. Also, the 2.53's price has dropped to around $260 last I checked on pricewatch. Whether not it'll be a better deal than the $250ish (which is usually the price for the newest AMD chip upon introduction) is debatable.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
August 22, 2002 4:52:03 AM

"As Crash would say: We need Dual Channel DDR to even match and outperform RDRAM based P4 systems."

It wouldn't help it any. The AXP's bus is simply too limited to take advantage of that kind of bandwidth, so it would be pretty much useless.

Knowledge is the key to understanding
a b à CPUs
August 22, 2002 5:14:47 AM

I was just there, noticing how the 2.53 mananged to outmanuver the 2600+ on around half the test, even when confined to an i845 chipset DDR board. It seems that sites that LOVE AMD will go to whatever measures are necessary to make the Intel system look worse, going so far as to use an inferior (by Intel standards) board and memory configuration and hoping noone notices they've been tricked!

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 22, 2002 5:32:22 AM

yes! how observant you are, crash!

<font color=green> there's more to life than increasing its speed -Ghandi</font color=green>
August 22, 2002 5:32:22 AM

Really? What chipset should they have used for the P4? That new SiS one?

Knowledge is the key to understanding
a b à CPUs
August 22, 2002 6:18:10 AM

How about the chipset this CPU was DESIGNED for, the i850E? A P4T533 would have laid waste to those scores!

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 22, 2002 2:58:48 PM

Agreed about the chipset choice. At least they didn't use SDRAM.

Half the benchmarks? Well that depends on whether you count the seven components of the SPEC ViewPerf suite separately or one benchmark suite. Likewise for the two Science Mark benchmarks. However, I'll concede your point. The SPEC ViewPerf results are very signifcant and XP 2600+ performs poorly.

Did you notice the following comment in the conclusions?

"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"

Do you think Aceshardware found something else that is weak in VIA chipsets, namely the AGP port? or are they just fishing for excuses. They are obviously way off the mark with the memory bandwidth comment since they paired the P4 with DDR memory erasing the P4's memory bandwidth advantage.


<b>I have so many cookies I now have a FAT problem!</b>
August 22, 2002 6:08:51 PM

Goddamn it, each time I write that statement, everyone thinks I speak of the AXP!
I was referring to what Crash and I both agree on, Dual Channel DDR FOR P4 would be what it needs to compete and outrun RDRAM. DDR 333 is now standard, dual channel that and you get 5.4GB/sec, and although P4 won't use that much, the latency is still lower than RDRAM and therefore SHOULD beat PC1066. DDR still costs much less than RDRAM as well.

--
Is the opportunity to earn money by working, free?
a b à CPUs
August 22, 2002 8:16:31 PM

It could be a legit gripe on the VIA chipset for the AMD, but since they also hadnicapped the P4, and since they had access to other boards, that point is mute IMO.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 22, 2002 8:52:28 PM

Quote:
that point is mute IMO.

LOL! "Mute" as going unheard [by you] or "moot" as in immaterial?


<b>I have so many cookies I now have a FAT problem!</b>
August 22, 2002 9:02:12 PM

Quote:


LOL! "Mute" as going unheard [by you] or "moot" as in immaterial?


Well, I've seen people write moot as mute so many times that it's not funny.

Intelligence is not merely the wealth of knowledge but the sum of perception, wisdom, and knowledge.
August 22, 2002 9:14:59 PM

This is to all you speed freaks out there. What do you need all of this speed for??? What can we honestly use "software wise" that we need this much speed. Isn't this war getting a little redundant. That is of course if your not making the next FF movie in your living room. This has been a huge hobby for myself for the past 8 years and I love it just as much as the next person. I currently have a P3 1Ghz and everything runs great but I still have the urge to upgrade. Maybe thats what it is an addiction kind of. Although I know I don't need the extra speed for anything I use I still want to upgrade. How fast do we have to go and will we ever come to a plateua? Please don't take this the wrong way, I just want to get everyone's opinion on this.

:tongue: <font color=red>Cut the mullet!!</font color=red> :tongue:
a b à CPUs
August 22, 2002 10:26:59 PM

"Moot", but still not worth hearing about (lol, might as well be mute).

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 23, 2002 12:41:17 AM

i thought rambus was dead? wassup with PC1066? intel 850E?

i guess they'll have cute little mini fans on the rimms soon. heheh

a p4 2.8GHz with PC1066 would be so nice though.

<font color=green> there's more to life than increasing its speed -Ghandi</font color=green>
a b à CPUs
August 23, 2002 2:21:13 AM

PC1066 has been THE performance standard for Intel P4's with the "533" bus ever since the "533" bus was introduced. At the time of i850E release, Intel had not yet verified PC1066 RDRAM, so Intel brand boards are the only i850E boards that don't support it.

<font color=blue>You're posting in a forum with class. It may be third class, but it's still class!</font color=blue>
August 23, 2002 7:36:44 AM

my original thought in bringing up the two articles on the latest and greatest intel and amd offerings is that progress is being made. i sure like to think that we, the community of tomshardware, had something to do with it. but without complaint, there cannot be no progress (at least not immediate). so my suggestion to all the amd fanboys out there is try complaining about amd more.. make them stop using PR+ system. people rightly complained sooo much when intel introduced the p4, they had no choice but to accelerate its performance.

Quote:
Although I know I don't need the extra speed for anything I use I still want to upgrade. How fast do we have to go and will we ever come to a plateua?

i love technology because it is the fastest industry for progress so far. it's human nature to want to do more. wouldn't you like to be able to do twice as much in half the time?

(i think the main reason why humans are so eager to ramp tech up so fast is so that we can get more time in the day to sleep) nite nite.

<font color=green> there's more to life than increasing its speed -Ghandi</font color=green>
August 24, 2002 12:13:44 AM

Rambus ain't dead yet. That PC1066 RAM not only has a faster clock speed than it's predecessors, but it also has a 32 bit interface - twice as much as anything that came before it - which combines to have a memory bandwidth of 4.2 GBps.

Knowledge is the key to understanding
August 24, 2002 12:19:22 AM

Just to nitpick, "PC1066" is the 533MHz, 16-bit version. PC4200 is the 533MHz, 32-bit version.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
August 24, 2002 2:30:55 AM

What about it? The 32-bit version is clearly labeled "RIMM4200". The 16-bit version is used in parity which offers comparable performance to the single-channel 32-bit version. But "PC1066" is the 16-bit 533MHz RDRAM module.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
August 24, 2002 2:34:54 AM

Which is DDRed to get PC1066.

--
And now, an advice from your friendly Nike shoes slogan: JUST DO HER!
August 24, 2002 2:48:05 AM

The frequency is 1066 MHz, but the name is completely different. 133MHz DDR memory is called "PC2100" as opposed to SDR. They both run at the same frequency though.

"We are Microsoft, resistance is futile." - Bill Gates, 2015.
August 24, 2002 2:56:18 AM

You love to be picky and try to correct the already correct stuff, don't you?!

--
And now, an advice from your friendly Nike shoes slogan: JUST DO HER!
August 24, 2002 3:03:30 AM

And - once again - RIMM4200 is not truely 32-bit RDRAM. The devices are the same PC1066 16-bit RDRAM devices found on PC1066 RIMMs. RIMM4200 is basically two PC1066 (RIMM2100) single-channel, single-sided RIMMs combined into a dual-channel, dual-sided RIMM. Nothing has changed in the technology - it is simply re-packaged.

If the thought I thought I thought had been the thought I thought, I wouldn't have thought so much.
August 24, 2002 3:09:44 AM

Wow you finally came back man!
What's up with you being so absent?


--
And now, an advice from your friendly Nike shoes slogan: JUST DO HER!
August 24, 2002 3:26:53 AM

LOL. i knew people would eventually be confused by using bandwidth and frequency.

"ddr sdram" uses both!
if a it operates at say, 200mhz, which is doubled to get 400MT/s (using my trusty multiply by 8 to get bandwidth), you get 400*8=3200.... hence PC3200!
oh yeah, i mentioned that ddr uses both, so they also go by the name "DDR400"

so, PC3200=DDR400="ddr sdram with 200mhz"

now, rambus is slightly different because it doesn't double anything nor does it use bandwidth... rambus uses the convention that ddr400 is using.. so when u see rambus 1066MHz, it's really 533MHz, AND it's really 1066MT/s. (i'm going to shove this down people throat to educate someone to use MT/s..LOL)

anyways, you have rambus 1066="rambus at 533MHz"

now some rambus idiot decided since everything else is PCxxxx, let's just use that tooo!

luckily the lowest ddr is ddr200 (which is "ddr at 100MHz").. which is (do it with me now, 200*8=1600).... yes..PC1600

imagine if there was a ddr133 (hmm maybe there was, i dunno) now ddr133 (which is "ddr at 67MHz") would be? yup, ding ding ding!!!!PC1066!!!!!

<font color=green> there's more to life than increasing its speed -Ghandi</font color=green>
!