Athlon64 Preview: nForce3 at 2.0GHz @ anandtech

[UlricT]

Im pretty sure they are using X2 - The Threat. Xmen 2: Wolwerines Revenge was a badly done Ps2 port....

 

[UlricT]

Originally posted by: BFG10K

Those are some of the most inefficiently coded games on the market. Their system resource to eye candy ratio is so low its embarassing.

For Unreal II you may have some degree of correctness but you're dead wrong about UT2003. The UT2003 engine has one of the highest eye candy/performance ratios around, right up there with Quake III and the Serious engine.

I think both UT2003 and Unreal II use the exact same engine...

 

[0roo0roo]

Originally posted by: UlricT

Originally posted by: BFG10K

Those are some of the most inefficiently coded games on the market. Their system resource to eye candy ratio is so low its embarassing.

For Unreal II you may have some degree of correctness but you're dead wrong about UT2003. The UT2003 engine has one of the highest eye candy/performance ratios around, right up there with Quake III and the Serious engine.

I think both UT2003 and Unreal II use the exact same engine...

really? but ur2 is newer. i thought all unreal games sorta used the same engine, but the engine keeps getting updated constantly so its not "really" the same engine.

 

[ViRGE]

0roo0roo, that's correct. Both games are based off what's called the "second generation Unreal engine." The exact engine is different, in that U2's was customized for the needs of the game, and UT2003's also customized accordingly. It's like how games ranging from Elite Force to Jedi Academy are all based on the Quake3 engine, but still seperate engines because of unique features.

 

[Duvie]

A couple of things....

1) It looked impressive in gaming which I can give a Shite about....I do multimedia stuff, cadd rendering, etc and right now it doesn't look much better then the best intel has...makes me think the 3.4 prescott will pull away further in those categories....

2) Anandtech assummed too much that the single channel memory controller will perform the same as the opterons dual channel...where is there proof that the single and dual will perform the same. will those avg 10% gains drop to 5% or less??? I don't know, but either do they it appears...I little too much Assumming....


So what do we have here---- a chip rated 3200+ that may actually deserve the pr rating versus the mockery that the barton 3200+ was---a chip that will be expensive initailly----a chip that will need to be ran with ecc memory???

 

[ViRGE]

Duvie, the Athlon64 FX will only require registered, not ECC, and the 3200+, since it's significantly different, will probably work with unregistered(although no one is sure about this until the final specs are released).

 

[Pariah]

You seem to assume that when we refer to ?Athlon 64? that we are referring to the rumored Socket 754 single channel DDR CPU. And yet again, we aren't allowed to directly (or even indirectly) confirm any specs or names, and even in this instance you?re assuming that we?re referring to something when we?ve already said we?re under NDA. If you are still quibbling over the fact that we didn't use "FX" or some other rumored term in the title based on what I've already said about NDA, what else can I tell you, you won't ever understand where I?m coming from, or admit that it?s just a plain waste of time to dissect exactly how misleading a title might be at this point in time.

You knew this thread was coming back. OK, all the NDA's are up. Would you like to explain to us now, why the benchmarks are so different between the preview and any of the A64 iterations? After seeing the benchmarks of the final product it looks like your preview is more misleading now with all the information than it was before when those of us lowly peons didn't have all the top secret insider info you had. Or is the CPU in your preview still under NDA?

 

[Accord99]

Especially in Gun Metal, which was used numerous times by AMD fans to suggest that the A64 would be 40-50% faster in games, when the actual difference was more in the single digits.

 

[UNCjigga]

I don't like integrated memory controllers...WTF, early adopters are stuck with either single-channel DDR400 or buying expensive, registered DDR400 dimms?? That's it...waiting game over. My next PC will have an 865/875 chipset and an HT Pentium 3.0C or above...

 

[LanEvoVI]

Answer to Pariah's question?

 

[sandorski]

Originally posted by: Accord99
Especially in Gun Metal, which was used numerous times by AMD fans to suggest that the A64 would be 40-50% faster in games, when the actual difference was more in the single digits.

Gunmetal wasn't used in the Review.

 

[sandorski]

Originally posted by: Pariah

You seem to assume that when we refer to ?Athlon 64? that we are referring to the rumored Socket 754 single channel DDR CPU. And yet again, we aren't allowed to directly (or even indirectly) confirm any specs or names, and even in this instance you?re assuming that we?re referring to something when we?ve already said we?re under NDA. If you are still quibbling over the fact that we didn't use "FX" or some other rumored term in the title based on what I've already said about NDA, what else can I tell you, you won't ever understand where I?m coming from, or admit that it?s just a plain waste of time to dissect exactly how misleading a title might be at this point in time.

You knew this thread was coming back. OK, all the NDA's are up. Would you like to explain to us now, why the benchmarks are so different between the preview and any of the A64 iterations? After seeing the benchmarks of the final product it looks like your preview is more misleading now with all the information than it was before when those of us lowly peons didn't have all the top secret insider info you had. Or is the CPU in your preview still under NDA?

In Q3 the Preview used "High Quality", Review used "demo four": possibly different benches, but I dunno.

Don't forget that the cpus used in the Review and the cpu used in the Preview are different, so a difference in Bench scores only seems logical. I didn't bother to check if all drivers are the same version, but that too could be a difference. Even the P4 3.2c scores are different(often higher)between the two, so it's likely there is a difference in setup. At anyrate, though the 64/FX scores are generally lower in the Review than in the Preview, they still place in approx the same position compared to the P4 3.2c.

 

[Accord99]

Originally posted by: sandorski

Originally posted by: Accord99
Especially in Gun Metal, which was used numerous times by AMD fans to suggest that the A64 would be 40-50% faster in games, when the actual difference was more in the single digits.

Gunmetal wasn't used in the Review.

But other reviews did.

 

[ST4RCUTTER]

Originally posted by: Pariah

You seem to assume that when we refer to ?Athlon 64? that we are referring to the rumored Socket 754 single channel DDR CPU. And yet again, we aren't allowed to directly (or even indirectly) confirm any specs or names, and even in this instance you?re assuming that we?re referring to something when we?ve already said we?re under NDA. If you are still quibbling over the fact that we didn't use "FX" or some other rumored term in the title based on what I've already said about NDA, what else can I tell you, you won't ever understand where I?m coming from, or admit that it?s just a plain waste of time to dissect exactly how misleading a title might be at this point in time.

You knew this thread was coming back. OK, all the NDA's are up. Would you like to explain to us now, why the benchmarks are so different between the preview and any of the A64 iterations? After seeing the benchmarks of the final product it looks like your preview is more misleading now with all the information than it was before when those of us lowly peons didn't have all the top secret insider info you had. Or is the CPU in your preview still under NDA?

Not to be argumentative, but I have to disagree with your assesment of the A64 Preview article vs. today's article. It's pretty obvious that Wesley's piece was a preview of the Athlon FX and not the A64 as mentioned previously in this thread. The scores between an overclocked 1.8Ghz Opteron (2.0Ghz) and the Athlon FX (2.2Ghz) are extremely close. This is not really surprising considering the CPU's are virtually identical without respect to clockspeed. Here are a few similar benchmarks performed in both reviews that should be of note:

Quake III Arena
Athlon FX 413
Opteron 423

Business Winstone 2002
Athlon FX 43.5
Opteron 38.2

Unreal Tournament 2003 Flyby
Athlon FX 252.9
Opteron 268.1

In each of these the difference is less than 1%. How is that misleading?

edit: formatting

 

[Pariah]

edit: formatting

You might want to edit your post for math as well. The differences between those are roughly 2.5%, 6%, and 12%.

 

[ST4RCUTTER]

Pariah,

You are correct! Sorry about that, it's late over here...anyway. I think my point is still valid considering the various hardware that can come into play when throwing together a system. Anything from the RAM timings used to the driver version could explain the variances we're seeing here. I still feel that Wesley's article did a good job of attempting to show the performance of a system that was previously under NDA's.

 

[First]

Originally posted by: Pariah

You seem to assume that when we refer to ?Athlon 64? that we are referring to the rumored Socket 754 single channel DDR CPU. And yet again, we aren't allowed to directly (or even indirectly) confirm any specs or names, and even in this instance you?re assuming that we?re referring to something when we?ve already said we?re under NDA. If you are still quibbling over the fact that we didn't use "FX" or some other rumored term in the title based on what I've already said about NDA, what else can I tell you, you won't ever understand where I?m coming from, or admit that it?s just a plain waste of time to dissect exactly how misleading a title might be at this point in time.

You knew this thread was coming back. OK, all the NDA's are up. Would you like to explain to us now, why the benchmarks are so different between the preview and any of the A64 iterations?

Uh, I wouldn't like to explain that because that's a ludicrous statement. Do you have any data backing up this claim, because looking at our own benchmarks it's clear Wesley was dead on with his A64 preview.

After seeing the benchmarks of the final product it looks like your preview is more misleading now with all the information than it was before when those of us lowly peons didn't have all the top secret insider info you had.

See above.

Or is the CPU in your preview still under NDA?

Subtle insults will get you no where, especially since your blanket statements make basically no sense when comparing Wesley's preview results to his final review results. And no, comparing them to Anand's results doesn't make any sense considering the different BIOS settings (memory timings, AGP settings, etc.) , different hardware (HDDs, etc.) and drivers (IDE, AGP, video, etc.).

 

[sandorski]

Originally posted by: Accord99

Originally posted by: sandorski

Originally posted by: Accord99
Especially in Gun Metal, which was used numerous times by AMD fans to suggest that the A64 would be 40-50% faster in games, when the actual difference was more in the single digits.

Gunmetal wasn't used in the Review.

But other reviews did.

Could you point the sites out that did, I don't wanna go on a wild goose chase.

 

[Pariah]

And no, comparing them to Anand's results doesn't make any sense considering the different BIOS settings (memory timings, AGP settings, etc.) , different hardware (HDDs, etc.) and drivers (IDE, AGP, video, etc.).

I guess this as close as we are going to get to admittance by you that nothing in the preview was really comparable to the actual release hardware, so any similarity in performance, of which there weren't many, were more coincidental than actually forward looking knowledge.

 

[First]

Originally posted by: Pariah

And no, comparing them to Anand's results doesn't make any sense considering the different BIOS settings (memory timings, AGP settings, etc.) , different hardware (HDDs, etc.) and drivers (IDE, AGP, video, etc.).

I guess this as close as we are going to get to admittance by you that nothing in the preview was really comparable to the actual release hardware, so any similarity in performance, of which there weren't many, were more coincidental than actually forward looking knowledge.

Admit what? There's nothing to admit, the setups between Wesley's review and Anand's were different, and Wesley used an identical setup for his preview and his actual review where the results were (surprise surprise) almost identical.

Ignore those two vital facts if you wish.

 

[Pariah]

Uhh, if your original article was a preview of the FX, which you seem to confirm and deny depending on how it benefits your arguement, you previewed a clock speed that doesn't even exist. Shouldn't your insider info at least predicted the right clock speed? Unless I'm missing something there is no 2.0GHz FX, only the FX-51 @ 2.2GHz (AMD Processor Pricing. If this was a preview of a standard A64, then the numbers aren't all that close and a pretty poor estimation of performance. Regardless, this isn't going anywhere, you seem to think this was a relevant and accurate preview of product which you don't to want to specify, and I don't think it was for either an Athlon 64 3200+ or FX-51. I don't think either side is going to agree on anything, so we'll just agree to disagree.

 

[First]

Originally posted by: Pariah
Uhh, if your original article was a preview of the FX, which you seem to confirm and deny depending on how it benefits your arguement, you previewed a clock speed that doesn't even exist. Shouldn't your insider info at least predicted the right clock speed? Unless I'm missing something there is no 2.0GHz FX, only the FX-51 @ 2.2GHz (AMD Processor Pricing. If this was a preview of a standard A64, then the numbers aren't all that close and a pretty poor estimation of performance. Regardless, this isn't going anywhere, you seem to think this was a relevant and accurate preview of product which you don't to want to specify, and I don't think it was for either an Athlon 64 3200+ or FX-51. I don't think either side is going to agree on anything, so we'll just agree to disagree.

Guess what, the Preview predicted the A64 and A64 FX's performance. I don't know what's so hard to understand about this; the FX and the plain A64 using identical chipsets only differ in socket layout, registered memory, and single versus dual channel support. Single/dual channel capability is the only difference that would affect performance, and due to the fact that the A64's memory controller is on-die, dual channel support means almost nothing. Gee, do you think we knew all of this information and were under NDA when the Preview was released earlier this month or do you think I'm just trying to give you a hard time?

But it's OK, believe whatever you want to believe...