Athlon 64 anyone?

fmah · Post by **fmah** » Mon Sep 22, 2003 9:20 pm

Okay, who's first? I see they are listed as in stock in some places.

Semm · Post by **Semm** » Mon Sep 22, 2003 9:42 pm

Oh, right. It's the 23rd. I completely forgot! I'm waiting, myself. No money to buy one, at the moment, and I'm curious what Intel's response will be. Competition is fun

Mark

fmah · Post by **fmah** » Mon Sep 22, 2003 9:49 pm

Intel's response, well, one thing they are doing is promoting the 3GHz parts to people. They have a special sale to retail store workers. This gives them temporary new "sales people."

Semm · Post by **Semm** » Mon Sep 22, 2003 9:55 pm

*laughs* Well, I'm not going to get into an intel/amd argument. I use both, and am happy with both. But regardless of what you think of them as a company, they are capable of some impressive engineering. As I said, it'll be interesting.

Mark

Kostik · Post by **Kostik** » Tue Sep 23, 2003 3:22 am

French website x86-secret.com just posted Athlon 64 and Athlon 64 fx benchmarks (they also include Opteron, Athlon XP, Duron, P4, P4 EE, and Xeon).

Performance (Sandra, SuperPi, PCMark)
Performance (3D Studio, PovRay...)
Performance (3DMark 2001, 3DMark 2003, UT2003...)
Overclocking page 1, overclocking page 2 (Benchmarks).

Very interesting, but they don't say anything about cooling.

Post by **Ralf Hutter** » Tue Sep 23, 2003 3:28 am

Apparently the NDA expires today at 10 AM PST. I'm sure we'll see plenty of info after that.

Mwave lists them for sale.

CoolColJ · Post by **CoolColJ** » Tue Sep 23, 2003 4:31 am

2 hot, too expensive and not fast enough is my impression

dago · Post by **dago** » Tue Sep 23, 2003 5:09 am

CoolColJ wrote:2 hot, too expensive and not fast enough is my impression

-1;Troll

Or you have to better define your criterias :

For example, if you compare an Athlon64 3200+ to a P4 3.2
- Hot : no numbers yet, but at least 20-30W less (?)
- Expensive : 200$ less (about 30%)
- Speed : equivalent

-> you can say that it's cooler, cheaper while having the same speed.

Of course, if you compare with a VIA C3 1GHz, then it's
- Hot : something like 10x
- Expensive : idem
- Speed : blast the C3 off

-> so, it's much faster, but at an awful financial and thermal cost

Tobias · Post by **Tobias** » Tue Sep 23, 2003 5:27 am

*grin* If we are to enter that debate, even though it is very off-topic...

[no pun intended]

dago: Since Coolcolj didn´t make a relativistic statement (but rather a normative one:) it can not be demned faulty on the grounds you supplement, even though they in relative terms are correct.

And having said that, I have to side with Coolcolj, it IS to hot, to expensive and to slow:) Isn´t it the ideal of every puter-geek out there (here?) to be able to run a blasting fast chip without need for any fans that costs us nothing?

Until that comes true I guess Dago has nailed it;)

CoolColJ · Post by **CoolColJ** » Tue Sep 23, 2003 5:27 am

Well the prices I've seen talked about seem pretty expensive for an AMD.

It's much slower for some of the apps I would use it for, like 3d rendering, except with Povray.
It does seem better suited for in some games etc.
Look at the benchmarks

http://translate.google.com/translate?s ... %3Fid%3D91

all i can say is to amd buyers who only go for em cause their cheap they can forget cheap high performance amd's from now on, amd have adopted intel pricing scheme and i'm wondering how this is going to affect the pricing of the p4ee

Gone are the days of nice $200 chips beating $1000 intel chips...sorry guys, u have to actually *pay* for your performance now

As anand says, let the wars begin....personally I think the AMD chip will be forgotten about once we see some prescot numbers

wumpus · Post by **wumpus** » Tue Sep 23, 2003 5:30 am

Yes the A64 runs cooler than the P4 of similar speed (in applications). No question about that.. has been confirmed multiple places. Probably because of the higher performance per clock, eg, lower clock rate.

So if you want high performance, low noise, the A64 might be of interest.

CoolColJ · Post by **CoolColJ** » Tue Sep 23, 2003 5:45 am

Originally posted by chainbolt
The French site " x86-secret" has a full A64 review and compars it with the 3.2 EE. They have tons of tests, very big read. And the results are VERY interesting. To make it short:

(a) In terms of performance it seems that neither the A64 3200+ nor the A64 FX51 doesn't come through against the 3200 EE and often not even against the 3200.
(b) Overclocking was a little limited: They managed 11 x 218 MHz at the best. That's a 9% overclock.

The tests were done with the XP 64-bit Edition. 2 interesting examples. There are others tests which have the A 64 in front, but that's a clear minority.

Here is a translation of the results:

" After several hundreds of benchmarks carried out, we would like first of all to thank all the people who trusted us and allowed in this article to be born. Prepared for many months and several times pushed back, it was a race against the clock, but the result is there and that is most important: ..... Athlon 64 3200+ dominates the Pentium 4 3.2 Ghz in UT2003. Unfortunately, in all our benchs, it is almost the only test the Athlon 64 3200+ finishes in front of Pentium 4 3.2 Ghz with a clear lead. Offering identical results under 3DMark 2001/2003, it's on par with Pentium 4 in Quake 3, Wolfenstein 3d or Comanche 4 (carried out but not published, to see low). Overall the INTEL processor is slightly ahead. But it has to be noted that AMD this time at least selected a correct P-Rating of 3200+ which was not the case with the Athlon XP 3200+. .... Suffering from a lack from band-width memory and especially from a low core frequency, Athlon XP 3200+ is beaten in 3D related bechmarks like 3DStudio max or Kribi. That's logic you might say: This software is band-width related. However in a few heavy calculation oriented benchmarks the A64 is ahead. In the whole of the benchmarks, Pentium 4 3.2 Ghz thus remains overall in front of Athlon 64 3200+ .....Does it mean the Athlon 64 is a bad processor? Not, undoubtedly, but in spite of the efforts makes compared to Athlon XP, Athlon 64 would have had still better performances with the controller report of Athlon 64 FX. At present, the choice to adopt a completely new platform including a new motherboard to achieve not more than 32 bit performance equal to that of a Pentium 4 remains difficult. As for 64 bits, it still remains to see the real profit.

Athlon 64 would have been the king of this comparative without the Pentium 4 ' Extreme Edition " The Gangster. One did not believe any more INTEL capable of a sudden start of last minute of the kind. However, the mammoth of Santa Clara, obviously aggravated by the spine in the foot baptized "Athlon 64 FX" woke up. In less than two months it was done: One takes the currently available most powerful core at INTEL. ..... This said, INTEL answered the evil by the evil and Pentium 4 ' EE' is a success in term of performances. P4 EE is catching up with Athlon 64 FX in almost all tests (except for UT2003 in BotMatch) and exceeding it, sometimes largely, in others.

sgtpokey · Post by **sgtpokey** » Tue Sep 23, 2003 5:48 am

Sorry, performance-wise / price-wise / and heat-wise... I'm with Overclockers.com (see link below). If it's performance & relative quietness , I really appreciate my overclocked 1700+. Overclocked with a 1200rpm fan, it gets performance in the range of the Athlon 64 benchmarks for much less cost.

http://www.overclockers.com/articles837/

So bottom line, neither the 64 or the P4-EE can justify their performance/cost if you factor in overclocking (but still maintaining reasonable quietness). If you are not an overclocker, than you might not be too interested in raw performance anyway (And a desktop Pentium M ought to become the cpu of choice for those tasks!!)

I'll skip on 64 and P4-EE... They ARE too hot, too expensive, not enough performance difference...

dago · Post by **dago** » Tue Sep 23, 2003 6:14 am

At least my (flamebait ?) comment started a little bit more of discussion

As said in another topic, each of us has different weight for all the factors.
Even if you take (heat, price, performance), I have very different weight for a server at work than for my personnal use at home.

For my workstation, I'll be very happy with a dual low-voltage opteron, while for my main home pc (game/media), I prefer going the Athlon XP way.

In any case, AMD is much cheaper than Intel (and it's one of my main factors). Currently, dissipated heat for desktop CPU is also on AMD's side.
Well - I don't 3GHz performances, so here I'm very happy with my 1700.

And in any case, for the price, better wait a little bit.

Zhentar · Post by **Zhentar** » Tue Sep 23, 2003 6:32 am

It looks like AMD plans to shrink the price difference with the hammer. I"m dissapointed by that, but AMD does have to start turning a profit or it won't matter.

And the Athlon 64 should be a good heat/performance upgrade, especially when the 90 nanometer die shrink comes.

Tobias · Post by **Tobias** » Tue Sep 23, 2003 6:39 am

CoolColJ: I get the impression that those benchmarks are run in an 64-bit environment, but the benchmarkcode is still run at 32-bit. (as there are no benchmarks for 64-bit yet, save a few server-benches...)
Anyway, If tested on a 64-bit platform, the A64 will be in 64bit-mode and emulating 32-bit. Make no mistake, when we are talking pure hardware, the A64 is NOT a 32-bit AND 64-bit CPU, it is either or. In 64-bit mode it emulates 32-bit code, and should in thise case realy be compared to an Ithanium.
Had the testing been done on a 32-bit platform, the A64 would be running in legacy-mode and would be a true 32-bit CPU with built in memorycontroller and SSE2 instructions.
Judgind from the brief test you have provided, I have to say that the A64 fairs well considered that it is emulating and therefore not utlizies its full capacity to tun the benches.

Then again, this is the performance one can expect when running 32-bit games and using a 64-bit OS, a situation I guess will prevail for many users for a long time. It will be very interesting to see what UT2004 will be like on a 64bit platform. It will be the first game to run the A64 in true 64bit mode....

On a sidenote: Assuming a 5x multiplyer, 200 FSB and undervolted, I guess there be no comparason which CPU will be most efficient in computaional power/heat.

Kostik · Post by **Kostik** » Tue Sep 23, 2003 8:01 am

Athlon 64 datasheets released!

Am I reading right ? 89w max power ?

whitter · Post by **whitter** » Tue Sep 23, 2003 8:08 am

saw this on 'The Register' today. The inference is that the A64 might well be based on the notebook chip.

Is this heading off Pentium-M-in-desktops at the pass (as it were)?

http://www.theregister.co.uk/content/3/32982.html

Kostik · Post by **Kostik** » Tue Sep 23, 2003 8:19 am

Ok, so as I understand it they can change their clock and vcore while running. Which must be the reason why AMD gives various power consumptions depending on clock and vcore :

Athlon 64 3200+ :
max 89w @2000Mhz (vcore = 1.5)
max 66w @1800Mhz (vcore = 1.4)
max 35w @800Mhz (vcore = 1.3)

Tobias · Post by **Tobias** » Tue Sep 23, 2003 8:40 am

I guess you are reading it right, on the other hand, they are also listing a minimum value, and as the article from "the Register" implied, it seems that the CPU only use as much juice as is needed. This means that while doing other things than gaming and other processorhungry things the CPU will slow down on its own, which reduces the need for cooling. SO, I would guess that typical power usage will be a bit lower than what we are used to from yesterdays chips.
On the other hand, the minimum power required is higher than today, so the idea of running it passivly goes down the drain.

On another sidenote, the 3000+ mobile CPU:s, if identical, is very much better priced than the 3200+, which would make them a very much more attractive deal.

AndrewC · Post by **AndrewC** » Tue Sep 23, 2003 2:05 pm

What suprised me looking at the thermal design guide was the 60-90lbs clip force and the use of thermal grease instead of a phase-change pad(it said don't use these).

Zhentar · Post by **Zhentar** » Tue Sep 23, 2003 2:44 pm

90 lbs force wouldn't surprise me with the IHS, and I guess its the same reason for the pad.

Tobias · Post by **Tobias** » Wed Sep 24, 2003 2:48 am

90 lbs, thats somewhere between 40-50kg, right?

dago · Post by **dago** » Wed Sep 24, 2003 3:01 am

Google : 90 pound = 40.8233133 kilogram

and kilograms are used for mass, not forces. So it should be about 400.34 N in average (read the != here).

Post by **Rusty075** » Wed Sep 24, 2003 5:04 am

90lbs isn't a big change. Current XP's have a max clip force of 100lbs.

And its nice to see that they finally reversed themselves on the thermal pad vs grease thing. Their reasoning for not using pads now is it may cause adhesion between the HS and the CPU.

fmah · Post by **fmah** » Wed Sep 24, 2003 7:09 am

I found this bit interesting. Anandtech ran benchmarks in Linux 64 bit. They recomplied the Lame MP3 encoder making encoding times go from 3.07 to 2 minutes.

http://anandtech.com/cpu/showdoc.html?i=1884&p=17

Pads are easier to install for mass production, but are normally not going to perform as well as properly applied grease.

fmah · Post by **fmah** » Wed Sep 24, 2003 7:11 am

I found this bit interesting. Anandtech ran benchmarks in Linux 64 bit. They recomplied the Lame MP3 encoder making encoding times go from 3.07 to 2 minutes.

http://anandtech.com/cpu/showdoc.html?i=1884&p=17

Pads are easier to install for mass production, but are normally not going to perform as well as properly applied grease.