More confusion from AMD, can you handle it?

[jiffylube1024]

Originally posted by: stevejst
We shall shortly see what these new AMD processors are capable of. Unfortunately AMD has been engaged in paper releases of processors for more than a year already, and their fan sites like AMDzone are just another twist in the sorry story that AMD has become.
The performance is not on their side for some time already and business has never been.

Yeah, you're right, the P4 3.2 Ghz really crushes the Athlon XP 3200+ doesn't it?

Originally posted by: DivideBYZero

Originally posted by: stevejst
Yes, I have to say I am getting lost with all this but I know I have no intention paying $350 for a single channel one. And then waht, when the 940 becomes more affordable to throw 754 chip and buy both mobo and a new chip? Nice idea.

P1 60 > 66 Anyone?

or Slot 1 to socket anyone?

or coppermine to tualatin anyone?

or the P4? How many "platforms" has that had?

Move away from the greenhouse before releasing missiles....

The P4 has had two separate pinouts (423 and 478), and quite a few "platforms". Anyone who bought a socket 423 P4 needs to have a cat scan however, especially an avid Anand reader. Socket 423 was a runt from birth, with S478 looming overhead always. Early adopters of new CPU technology usually get screwed, especially when it's really major and they have to work the kinds out through a few "platforms".

 

[stevejst]

the P4 3.2 Ghz really crushes the Athlon XP 3200+ doesn't it?

Exactly, as a matter of fact Barton "3200+" measures up to perhaps Pentium 2.8 GHz B, and trails badly nearly all of the line of C processors except the slowest one, 2.4C. Benchmarks are widely available.
Is that a successfull processor? Apparently not for AMD suffered 8th consecutive financial loss in Q2 2003.
There are all chances the release on Sept 23 this year will be another paper release of a processor they will not be able to deliver in volume before 2004.

No wonder because all of the AMD crowd support is mostly verbal since 99% of them won't purchase any AMD offer for more than $100. That is unfortunate truth AMD executives know well. All the hype and no substance, just like the Hammer myth.

"Huge" success of Opteron has yet to wrestle out 3% of server market despite of "purchases of thousands" as reported on AMD fan web sites.

 

[beatle]

Originally posted by: stevejst
No wonder because all of the AMD crowd support is mostly verbal since 99% of them won't purchase any AMD offer for more than $100.

What's wrong with being frugal? What's wrong with competition? From your posts, it looks like you're viewing AMD users as "bugs" that are all but exterminated. You get bit by a bug or something?

 

[sellmen]

Originally posted by: OddTSi

Originally posted by: sellmen
Opteron Desktop Benchmarks

A 2GHZ opteron outperformed a P4 3.2ghz (11/17 tests won) on desktop benchmarks, while using registered CAS2.5 RAM. Switching to regular CAS2 DDR should improve performance significantly.

The setups on the two are not equal. The Opteron has 4x the amount of RAM and it also uses 4 slots vs. 2 slots on the P4. If you read the recent memory article on the AnandTech homepage you'll see that using 4 slots is the ideal setup. The fact that a few of those benchmarks were memory bandwidth benchmarks, and that the memory setup was biased towards the Opteron makes those benchmarks useless. Not to mention the fact that AMDZone themselves labels some of the non memory bandwidth benchmarks (memory intensive) makes me eager to see a true, unbiased, and fair comparison between the two CPUs, because this one you linked to is ANYTHING but that.

And on a side note, why did you call them "Desktop Benchmarks"? Half of those are professional/science benchmarks. That has nothing to do with their results, just wondering why you decided to call them Desktop Benchmarks.

There were a fair amount of desktop benchmarks in that article: Sandra, 3dmark2001se, 3dmark2003, Codecreatures, Commanche 4, Unreal tournament 2003, Quake 3 arena. That is more then enough benchmarks to make an accurate desktop comparison. As for using 4x the memory, which of these benchmarks benefit from more than 512MB of RAM? The highest 3dmark scores are normally achieved with 512mb.

The Opteron has a significant performance penalty by using registered CAS2.5 RAM. Given the extremely low latencies of the IMC, each wasted cycle results in a much greater percentage penalty. I don't see how these tests are biased against the P4, when the Opteron is crippled by slow RAM.

 

[uunitdwestnd]

Did everyone read the article from xbits about the testing that took place with yes an ATHLON 64 running rated at 2800+@1,600 mhz. The Current 2800+@2,083 mhz outperformed it in just about every application except memory intensive applications, because of the new built in memory controller. But all things considered I think that AMD is using very interesting strategy, giving out 2800+ athlon 64's operating at 1,600 mhz for testing reviews. That is like showcasing Barry bonds at 50-75 percent of his strength, as to make Intel think they are going to use real low clock speeds and then bam they are gonna hit P4 speeds and just annihilate the P4. What a ploy the old cry wolf trick, but only between the two major powerhouses in the cpu manufacturing market. I've heard of read a couple of places, maybe just a rumor but the athlon 64 is supposedly gonna have 3 ghz range processing speeds. Whoa AMD! I think it is a beautiful strategy, even if it only buys them 3-6 months on top of the cpu manufacturing world. I love it. Think, do you REALLY think AMD is gonna build this new chip and market it to the public the way they are and have it to only be a second rate solution and operating at 1.6 ghz LET ME THINK???? I TRULY DOUBT IT!!!! Props to AMD for using a great strategy and Hopefully this will shut all the critics PIE HOLES!!!!!!!!! LONG LIVE AMD!!!!!!!!!!!!

 

[stevejst]

Strategy?

Intel reportedly upped its expected 3rd quarter earnings per share to 56c today quoting larger than expected processor sales.
You can almost bet your house that AMD will end the quarter with another loss, 9th in a row.

That tells you everything you need to know, what AMD has to offer - benchmarks of nonexistent processors on nonexistent motherboards on AMD fan web sites. In reality they have inferior processors to offer and people are not buying except for the discounted ones.

That is about strategy.

 

[Jeff7181]

I haven't seen any Athlon-64 tests... I've seen Opteron tests... what tests are you talking about, uunitdwestnd?

 

[stevejst]

You are new to the game then, I have seen many Hammer fictional "tests" and "benches" before, it started with Hammer at 800MHz compared to Intel 1.6A time ago. It was always "better" than Intel at the hand, yet nobody could verify that. Only on the last Compudex there were 4 or 5 different rigs with all sorts of numbers nobody could verify. Always "confidential" and always propagated all around AMD fanatics base.
And that "magnificent" Opteron that showed up this year has been sold at 4-5,000 units per months as compared to 150,000 Xeons. Apparently a huge success.
Remember, according to AMD, that was because of Sars epidemic.

 

[Jeff7181]

Of course there are Opteron tests and crap like that which are supposed to be indicative of Athlon-64 performance. I'm saying I haven't seen a real Athlon-64 test anywhere. Nobody has, because none of the engineering samples have been leaked that I know of. If you know of some, please post links.

 

[stevejst]

Athlon 64 is a Hammer core, AMD has no money and resources to develop anything radically new than what they have been delaying already for more than a year. Expect Opteron type processor in different clothes that might perform similarly.

 

[sellmen]

Originally posted by: stevejst
You are new to the game then, I have seen many Hammer fictional "tests" and "benches" before, it started with Hammer at 800MHz compared to Intel 1.6A time ago. It was always "better" than Intel at the hand, yet nobody could verify that. Only on the last Compudex there were 4 or 5 different rigs with all sorts of numbers nobody could verify. Always "confidential" and always propagated all around AMD fanatics base.
And that "magnificent" Opteron that showed up this year has been sold at 4-5,000 units per months as compared to 150,000 Xeons. Apparently a huge success.
Remember, according to AMD, that was because of Sars epidemic.

Opteron is much more geared towards the Itanium than the Xeon, why don't you compare sales of those two chips.

You have a tendency to berate AMD for every little thing. If AMD performs better, you attack its price, if AMD is cheaper, you attack its performance, if AMD changes motherboards, you attack it to no end, despite Intels history of constantly changing sockets. If Intel has problems, i.e. Prescott's power consumption, you defend it as the "natural progression" of chips.

If Athlon 64 comes out in one month and outperforms the P4, will you stop whining?

 

[stevejst]

If Athlon 64 comes out in one month and outperforms the P4, will you stop whining?

And if it doesn't, will you stop cheerleading it?

"If" is all there is. AMD has no performance, what is offered as a desktop chip is miserable when compared to HT Pentium, you all know that and drool for the real thing. I can see the benchmarks of AMD PCs posted here.

You can read thread upon thread about people changing boards, processors, cooling and what not to get that misery of Barton 2500 up to overclocked 2.3 GHz where it can somehow, with a stretch, be compared to Pentium 2.8C. Humorous.

 

[Jeff7181]

You and your benchmarks...

That should be in your sig... "I love benchmarks... I make nasty monkey love to them."

 

[stevejst]

That should be in your sig

Like the ones in your?

 

[sellmen]

Originally posted by: stevejst

If Athlon 64 comes out in one month and outperforms the P4, will you stop whining?

And if it doesn't, will you stop cheerleading it?

"If" is all there is. AMD has no performance, what is offered as a desktop chip is miserable when compared to HT Pentium, you all know that and drool for the real thing. I can see the benchmarks of AMD PCs posted here.

I don't think I'm cheerleading anything, simply explaining simple ideas to you. You seem absolutely stunned when a $650 AMD CPU comes out, despite the P4 3.2ghz being $640. You seem amazed when AMD saves money by fabbing only Bartons, then disabling half the L2 cache to sell as Tbreds. Either you are really dull, or you like starting threads that will let you flame AMD.

If AMD comes out with a high performing CPU next month, I'm sure you'll bash them for some reason.

 

[stevejst]

You seem absolutely stunned when a $650 AMD CPU comes out, despite the P4 3.2ghz being $640. You seem amazed when AMD saves money by fabbing only Bartons, then disabling half the L2 cache to sell as Tbreds.

3.2C is not the only HT processor out there that can offer performance. Athlon XP-51 is to be expected a single one on offer on Sept 23, and some $350 crippled version of the similar. Thats it. ON LIMITED OFFER."If" or "whether" it would be performance comparable to HT 3.2, we'll see. When it would be truly available.
As far as crippled Barton are concerned, yes, I am amazed that anybody would think that is a serious business venture but I am not amazed that AMD folks would. For the rationale has long gone from that company.
I see a great "success" with Barton sales spurred up with crippled Barton.

 

[OddTSi]

Originally posted by: sellmen
As for using 4x the memory, which of these benchmarks benefit from more than 512MB of RAM? The highest 3dmark scores are normally achieved with 512mb.

The Opteron has a significant performance penalty by using registered CAS2.5 RAM. Given the extremely low latencies of the IMC, each wasted cycle results in a much greater percentage penalty. I don't see how these tests are biased against the P4, when the Opteron is crippled by slow RAM.

That is all part of my point. The setups were not equal. The type of RAM was different, the amount of ram was different, etc.

When doing a benchmark comparison the ideal setup is to have as few changes between setups as possible, this being only changing the motherboard and processor. But they didn't do that, they changed the RAM setup too. So now we're left with useless benchmarks because it's nothing more than speculation trying to figure out which processor is faster. The whole point of a benchmark is to eliminate all the speculation, to have hard proof showing THIS is the how they stack up against each other. With the AMDZone article we don't have hard proof. The AMDZone article was made useless the second they changed the RAM setup upon doing the test.

And I'm not taking any sides, not for the P4 nor the Opteron, I just want to see a FAIR benchmark so I know where I should stand on all this. And I'm sorry to say, this article did NOT do a fair benchmark.

 

[sellmen]

Originally posted by: stevejst

You seem absolutely stunned when a $650 AMD CPU comes out, despite the P4 3.2ghz being $640. You seem amazed when AMD saves money by fabbing only Bartons, then disabling half the L2 cache to sell as Tbreds.

3.2C is not the only HT processor out there that can offer performance. Athlon XP-51 is to be expected a single one on offer on Sept 23, and some $350 crippled version of the similar. Thats it. ON LIMITED OFFER."If" or "whether" it would be performance comparable to HT 3.2, we'll see. When it would be truly available.
As far as crippled Barton are concerned, yes, I am amazed that anybody would think that is a serious business venture but I am not amazed that AMD folks would. For the rationale has long gone from that company.
I see a great "success" with Barton sales spurred up with crippled Barton.

Why don't you replace the word "crippled Barton" with "Athlon XP Thoroughbred". Does it shock you that AMD still wants to sell Tbred chips?

You seem awfully concerned about AMD's financial status, yet you flame them when the make any move to lower production costs.

I don't think anyone knows the final launch speeds of the Athlon 64, and there are no benchmarks on the 754 version. Called it "crippled" without any evidence to back that up is sightly rediculous.

 

[sellmen]

Originally posted by: OddTSi

Originally posted by: sellmen
As for using 4x the memory, which of these benchmarks benefit from more than 512MB of RAM? The highest 3dmark scores are normally achieved with 512mb.

The Opteron has a significant performance penalty by using registered CAS2.5 RAM. Given the extremely low latencies of the IMC, each wasted cycle results in a much greater percentage penalty. I don't see how these tests are biased against the P4, when the Opteron is crippled by slow RAM.

That is all part of my point. The setups were not equal. The type of RAM was different, the amount of ram was different, etc.

When doing a benchmark comparison the ideal setup is to have as few changes between setups as possible, this being only changing the motherboard and processor. But they didn't do that, they changed the RAM setup too. So now we're left with useless benchmarks because it's nothing more than speculation trying to figure out which processor is faster. The whole point of a benchmark is to eliminate all the speculation, to have hard proof showing THIS is the how they stack up against each other. With the AMDZone article we don't have hard proof. The AMDZone article was made useless the second they changed the RAM setup upon doing the test.

And I'm not taking any sides, not for the P4 nor the Opteron, I just want to see a FAIR benchmark so I know where I should stand on all this. And I'm sorry to say, this article did NOT do a fair benchmark.

I agree, it wasn't a fair benchmark, but there was no way to do a fair benchmark. Consider it a performance preview of Athlon 64 performance @ 2ghz; it should give a general idea of what to expect.

 

[stevejst]

I don't think anyone knows the final launch speeds of the Athlon 64, and there are no benchmarks on the 754 version. Called it "crippled" without any evidence to back that up is sightly rediculous.

Yup, single channel Athlon FX or whatever else it is going to be called is to be considered a crippled dual channel Athlon FX. Or a discount one, if you like that better ($350??). I am guessing the performance is going to be ... lower?

T'bred? T'bred line is on a deathbed, in a month from now I don't see myself buying any more T'breds. That is like buying Palomino now. If they introduced a new Palomino three months ago I would not buy it. I don't know who would?

 

[sellmen]

Originally posted by: stevejst

I don't think anyone knows the final launch speeds of the Athlon 64, and there are no benchmarks on the 754 version. Called it "crippled" without any evidence to back that up is sightly rediculous.

Yup, single channel Athlon FX or whatever else it is going to be called is to be considered a crippled dual channel Athlon FX. Or a discount one, if you like that better ($350??). I am guessing the performance is going to be ... lower?

T'bred? T'bred line is on a deathbed, in a month from now I don't see myself buying any more T'breds. That is like buying Palomino now. If they introduced a new Palomino three months ago I would not buy it. I don't know who would?

Performance will be lower, but you have no idea how much, seeing as how they are no benchmarks. If performance is comparible to a $640 3.2ghz P4, they will be a bargain at $350.

The Tbred line is still viable, the performance of a Barton is only 2-5% more per clock, depending on the application. If you are overclocking, a $65 1700+ @ 2.3/2.4ghz is one of the better options you have.

 

[stevejst]

Performance will be lower, but you have no idea how much, seeing as how they are no benchmarks. If performance is comparible to a $640 3.2ghz P4, they will be a bargain at $350.

It won't be comparable to 3.2C, I can take a bet on that one with you. I suspect that the full fledged FX-51 won't be able to beat HT 3.2.

Performance will be lower, but you have no idea how much, seeing as how they are no benchmarks. If performance is comparible to a $640 3.2ghz P4, they will be a bargain at $350.

The Tbred line is still viable, the performance of a Barton is only 2-5% more per clock, depending on the application. If you are overclocking, a $65 1700+ @ 2.3/2.4ghz is one of the better options you have.

Overclocking 1700 to 2.3-2.4 is a gamble not reality, what exactly 1 out of 3? Beside to do that you need $30 heatsink+fan+shipping, at least. Stability is another issue as well.
Even if you can do that 2.4 GHz XP 1700 is inferior to 3.0 GHz Pentiu, shortly $170 2.4C Pentium. And everyone will do that without additional cooling.

 

[Jeff7181]

Beside to do that you need $30 heatsink+fan+shipping, at least

You sound less and less knowledgeable about AMD processors the more you say things like this.

 

[stevejst]

Look at your sig. What could you do? Barton overclocked to 2.2 GHz with benchmarks below 2.6GHz Pentium C.
That is a proof enough to me.

$30 = $20 SK-7 + $5 high speed fan (3000+ RPM) +$5-6 shipping.

 

[Jeff7181]

What the hell are you talking about? What could I do?!?! I could sit here and watch you make an ass out of yourself by spouting your "anti-AMD Intel fanboyism." I could sit here while you masturbate while looking at your memory bandwidth in Sandra. I could wonder what you have against AMD... maybe you bought a $400 AthlonXP and cracked the core while putting the heatsink on and now you're pissed at AMD for not putting a massive shield on the core to protect it from morons.