Xbitlabs Xclusive preview on Athlon64.

[Idontcare]

Originally posted by: Lord Evermore
If a CPU combines two cores in one, then the yield isn't going to be lower based on that. It would be lower based on the simple existence of twice as many transistors and circuits (probably more than twice due to interconnecting). A single core with 500 million transistors would have pretty much the same yield as a 2-core CPU with 250 million transistors per core. Obviously if you have a 2-core die with 500M transistors, it's going to have lower yields than a single core die with only 250M transistors. The Athlon64 doesn't contain two cores though, just a LOT of transistors, obviously making the yield lower just like Intel raising the transistor count makes their yield lower for awhile.

The processor is NOT a 64-bit processor. It's not doing the same thing that the Itanium does, where it has to process 32-bit code differently from 64-bit code. The CPU is still a 32-bit processor, it just has 64-bit extensions that allow it to address higher amounts of memory, and give it more registers and some other things. It's still just a 32-bit processor using the 32-bit x86 ISA. So the resources in the chip aren't "dedicated" to any particular type of processing.

It uses all the resources it can presumably, whether in 32 or 64 bit mode. The OS won't be able to make use of some things I presume if it's only 32-bit, because it can't address 64-bit memory space, but the CPU should be able to make use of the new registers just like a Pentium3 or Athlon makes use of more than the 8 registers of the basic x86 ISA by register renaming.

The onboard memory controller would have been awesome if they'd gotten the part out on time originally. Now it's just falling behind the times compared to everything else. However I was under the impression that it may be possible, either immediately or with the next version, for a chipset maker to integrate a memory controller that takes over and shuts down the A64's controller, so you get the better memory types and speeds, but at the loss of the low-latency from the integrated controller.

Incidentally, whether the A64 needs dual-channel memory or not is entirely dependent on how the internal memory to core bus is designed. I don't know yet whether it is capable of throughput between the memory controller and the core that's essentially limited only by the core speed, or if there's a "bus" that will limit it.

Of course, to AMD they're not "limiting" it at all. They're making the Athlon64 with a single channel because they are focusing it on one market segment, and have another product line that they've put dual-channel into.

So is the Athlon XP not a 32bit processor since it is basically a 16bit chip with 32bit extensions?

 

[Sohcan]

Originally posted by: classy
Despite what some guy posted earlier, this will be the first high performance DESKTOP cpu with an integrated memory controller and 64 bit capable.

No it is not; I was using a 64-bit desktop computer with an integrated memory controller 5 years ago (the UltraSPARC IIi). I have little doubt that the Athlon 64 will be a fine microprocessor and a great performer, but the revolution you are expecting is merely evolutionary.

* not speaking for Intel Corp. *

 

[DOOPYLOOPY]

by reading Rand's posts my expectation for this CPU has been set....

thanks for the clear thinking.

 

[SiliconGuy1001]

Does anybody know if AMD has to pay intel royalties for x86-64 ISA? As far as x86-32 ISA goes, I am pretty sure that they do pay royalties.

 

[majewski9]

Yeah I was dissapppointed with those benches for the most part. I was expecting a bit more but like everyone smart here is saying this a beta chip and not a full out retail Athlon 64. I question some of X bits benchmarks and whether how much reliance I can put on them. First off I'd like to say that all these benches are done using the Via K8 chipset. Do I need to say more about the chipset ? I would have loved to seen a Nvidia chipset or the very least the AMD K8 solution. I serioulsy believe that the PC mark suite is suspect first off and Sys mark is total crap. In fact dont they usually use the 2002 suite which has long been thought and proven pro Intel? Windows media encoder resaults can easily be explained by the lack of software optimizations including ones for SSE 2 with the Athlon 64. In fact the previous versions of WM encoder had problems with SSE and Athlon XP. All the other benchmarks are due to the fact of a much lower clock. I think we can expect to see once AMD ramps the clock benches are going to look more like Unreal 2003 benches. The Athlon 64 smoked on Unreal! 1 meg L2 none the less is probaly why it absolutley destroyed. None the less Athlon 64 showed that it can do a lot at a nominal speed. I dont get why people dont think this thing is going to ramp? 130 nanometer and going soon to 90 nanometer as well as SOI with a much better cooling solution. Heat spreader for all those people who break their cores ( I installed 100 's of Tbirds, Durons, and Xp's and not one cracked core ). Athlon 64 simply needs to ramp in clock speed.

I do have some qualls with Athlon 64. First they should have released a better FPU units for the Athlon 64 and not just used the Athlon XP units. Second they should have forgone a single channel DDR and went straight to Dual channel just like Opteron. Third they should ramp in clock speed more aggressively and which they are proably already doing.

Any how I would love to see an Athlon 64 with an AMD 8000 chipset and @ working frequency of 2 ghz or specifically at 2.25 ghz. Lets see Xbit bench that!

In closing I like to say that AMD is trying to do something that no one in the X86 world has ever done. Extend the register set of the X86 instructions to 64 bits. Memory on die I believe is new to the x86 world but has been in a lot of chips in the 64 bit world ie the Ultra Sparc. If you ever seen an Ultra Sparc in action then you knoe that on die memory controller is a very nice feature. They are trying to break once and for all away from Intel and its aging x86. No wonder they have had problems! Maybe they were a little full hearty but I think Barton has provided some breathing room till when they decide to release Athlon 64. Hopefully before september and they better release it before 04 comes around.

 

[majewski9]

Originally posted by: SiliconGuy1001
Does anybody know if AMD has to pay intel royalties for x86-64 ISA? As far as x86-32 ISA goes, I am pretty sure that they do pay royalties.

No I dont think they do for IA 32 or if they do it isnt much. As far as x86-64 Intel would have to probaly pay AMD royalities or at least buy a lsicense like transmeta.

 

[SiliconGuy1001]

Originally posted by: majewski9

Originally posted by: SiliconGuy1001
Does anybody know if AMD has to pay intel royalties for x86-64 ISA? As far as x86-32 ISA goes, I am pretty sure that they do pay royalties.

No I dont think they do for IA 32 or if they do it isnt much. As far as x86-64 Intel would have to probaly pay AMD royalities or at least buy a lsicense like transmeta.

Intel owns the x86 ISA. If anything, it would be AMD paying intel. I read somewhere, probably xbitlabs, that AMD has to pay 5% of their sale price to intel. And as for the chipset licensees, they have to pay around $5 per chipset sold.

Concerning x86 itself, it is a dated ISA and AMD is only killing it further by extending the instruction set. The architecture is over 20 years old and has to be replaced. AMD is betting the farm that corporate america will not move on to EPIC or other high end RISC CPUs.

 

[classy]

Here is an interesting take on the Xbitlabs article. Read Here.

 

[Duvie]

To majewski...That is why when I listed roundup of benchmark percentages I left out those sysmarks and all the other mark crap!!! They are so synthetic it is tough to gather any info from it in comparisons across platforms....

I think to classy......It was abundantly clear to me this was a preview of a limited chip and a new platform. That being said for many to wipe these intial scores under the rug and say they don't count is a cop out. If these numbers showed the 1.6ghz 2800+ athlon 64 beating the 2.8 p4c in benchmarks the numberts would surprisingly have validity. I know ppl would say that it still had headroom....Lets see this for what it is...It is a progress report on the state of AMD's hammer...What can we garner from this PREVIEW??? The fact is they better makes some improvements in all areas and get those multimedia scores up as that is what I run and they were sad.

Also...Are all you guys selective readers??? Did you read the part about the test setup don't even running some of the multimedia software they wanted to run??? There are a number of problems that is for certain. Good news!!! They have a while to get it resolved before September...My gut tells me they will delay it again. Bad news!!! The opteron is supposed to be out in channels now so hopefully the optimizations are done for that....

 

[Mrburns2007]

The Athlon 64 will probably be a lot better when everything is finalized and clock is faster.


BTW: The Athlon 64 hit 1.8 Ghz a month or two back so 2.0 Ghz isn't out of the question.

 

[Lord Evermore]

Not out of the question, but it is worrisome to those who've been eagerly awaiting this CPU that after such a long delay, they MAY get to 2GHz at the highest, with the expected relatively low yields for the fastest-clocked cores. Intel's Prescott at 3.4GHz and higher will be out by then, with Prescott having a higher performance than even the 800MHz FSB P4's have now, and even a 2GHz Athlon64 shouldn't be getting a 3400+ rating unless the performance is WAY better than these scores.

They're going to need to modify their ratings eventually, they can't just keep using the same "Thunderbird" comparison without people getting even more tired of it than when the PR started in the first place. Only the official line says that the number is meant to apply to comparisons to a Thunderbird core, everybody knows it's meant to give them a number to compete with Intel's highest speeds. But Intel has made their performance levels go up with the faster bus speeds and increased cache, enough that AMD already needs to start adding more MHz in order to get the same PR number increase. (I.e., to actually be comparable to a P4, a move from x PR number to 100 higher used to only take 66MHz, but due to the increased P4 performance, 66MHz should only mean a 66 PR increase. If they continued using 66MHz per 100 points, then the real performance gets further and further from the "rating".) If they're giving this sort of performance a 2800+ rating, then they're just going to destroy any credibility with the rating system (as if the ratings on the Opteron line aren't bad enough).

To make that a little clearer, Intel is not only increasing frequency, they're increasing the amount of performance they get out of each cycle with increased cache and bus speed. AMD is starting out with the Athlon64 at a certain performance level, and is only going to be able to increase the frequency at a certain rate, and isn't likely to start doing things like increasing bus speeds and cache size anytime soon. So their performance will go up linearly, while Intel's goes up geometrically. So when AMD increases frequency and gives it a 100 point PR number increase, it's only going to apply relative to the previous speed of the A64, it'll no longer be comparable in real life to a P4's frequency. And no matter what AMD tries to do in order to keep from looking like they're not even trying, by claiming the PR number has nothing to do with Intel's speeds, the PR rating is not going to save them if it isn't comparable.

 

[Mrburns2007]

Originally posted by: Lord Evermore
Not out of the question, but it is worrisome to those who've been eagerly awaiting this CPU that after such a long delay, they MAY get to 2GHz at the highest, with the expected relatively low yields for the fastest-clocked cores. Intel's Prescott at 3.4GHz and higher will be out by then, with Prescott having a higher performance than even the 800MHz FSB P4's have now, and even a 2GHz Athlon64 shouldn't be getting a 3400+ rating unless the performance is WAY better than these scores.

They're going to need to modify their ratings eventually, they can't just keep using the same "Thunderbird" comparison without people getting even more tired of it than when the PR started in the first place. Only the official line says that the number is meant to apply to comparisons to a Thunderbird core, everybody knows it's meant to give them a number to compete with Intel's highest speeds. But Intel has made their performance levels go up with the faster bus speeds and increased cache, enough that AMD already needs to start adding more MHz in order to get the same PR number increase. (I.e., to actually be comparable to a P4, a move from x PR number to 100 higher used to only take 66MHz, but due to the increased P4 performance, 66MHz should only mean a 66 PR increase. If they continued using 66MHz per 100 points, then the real performance gets further and further from the "rating".) If they're giving this sort of performance a 2800+ rating, then they're just going to destroy any credibility with the rating system (as if the ratings on the Opteron line aren't bad enough).

To make that a little clearer, Intel is not only increasing frequency, they're increasing the amount of performance they get out of each cycle with increased cache and bus speed. AMD is starting out with the Athlon64 at a certain performance level, and is only going to be able to increase the frequency at a certain rate, and isn't likely to start doing things like increasing bus speeds and cache size anytime soon. So their performance will go up linearly, while Intel's goes up geometrically. So when AMD increases frequency and gives it a 100 point PR number increase, it's only going to apply relative to the previous speed of the A64, it'll no longer be comparable in real life to a P4's frequency. And no matter what AMD tries to do in order to keep from looking like they're not even trying, by claiming the PR number has nothing to do with Intel's speeds, the PR rating is not going to save them if it isn't comparable.

Yeah but the Athlon 64 has super low latency compare to the P4 and memory bandwidth is more hype then actually needed for most users. Also the Opteron is not intended for Joe Six Pack it's for Pro's who should know exactly what there getting and not get fooled into buying the hype.


PS: It makes me wonder if the real Athlon 64 will be a lot faster and AMD is just releasing these early ones to fool the competition.

 

[Duvie]

Originally posted by: Mrburns2007
The Athlon 64 will probably be a lot better when everything is finalized and clock is faster.


BTW: The Athlon 64 hit 1.8 Ghz a month or two back so 2.0 Ghz isn't out of the question.

Yeah but the whole thing was supposed to be released some time ago and supposed to have been 2ghz then....We saw hammer mobos back along time ago. It is AMD delaying this whole thing...

I like the rose colored glasses you look through. For AMD's sake I hope you are right. I think the hammer and all the processes involved in the making of this new arhictecture and chip was quite an investment in a rather sparse growth revenue period. They can't take a lackluster release already outdone by its current Barton chip, and a chip that will not ramp well and give them the type of return they need in this chip with longevity.

 

[classy]

Originally posted by: Duvie

Originally posted by: Mrburns2007
The Athlon 64 will probably be a lot better when everything is finalized and clock is faster.


BTW: The Athlon 64 hit 1.8 Ghz a month or two back so 2.0 Ghz isn't out of the question.

Yeah but the whole thing was supposed to be released some time ago and supposed to have been 2ghz then....We saw hammer mobos back along time ago. It is AMD delaying this whole thing...

I like the rose colored glasses you look through. For AMD's sake I hope you are right. I think the hammer and all the processes involved in the making of this new arhictecture and chip was quite an investment in a rather sparse growth revenue period. They can't take a lackluster release already outdone by its current Barton chip, and a chip that will not ramp well and give them the type of return they need in this chip with longevity.

Rose colored glasses? Wow what color are your glasses? I keep hearing the "Intelites", well it was suppose to be released a long time ago. So what. You know when Prescott was introduced? Last freaking March. And only time will tell what the final bottomline in performance will be.

 

[Wingznut]

Originally posted by: classy
I keep hearing the "Intelites", well it was suppose to be released a long time ago. So what. You know when Prescott was introduced? Last freaking March. And only time will tell what the final bottomline in performance will be.

I presume you do realize that Prescott never had a projected release of "Last freaking March".

 

[SexyK]

Originally posted by: Wingznut

Originally posted by: classy
I keep hearing the "Intelites", well it was suppose to be released a long time ago. So what. You know when Prescott was introduced? Last freaking March. And only time will tell what the final bottomline in performance will be.

I presume you do realize that Prescott never had a projected release of "Last freaking March".

Wingz, don't bother with this guy, if you don't believe the Athlon 64 "is technically superior to anything before it and anything that will released in the forseeable future" then you're obviously an "Intelite"

 

[Wingznut]

Originally posted by: SexyK

Originally posted by: Wingznut

Originally posted by: classy
I keep hearing the "Intelites", well it was suppose to be released a long time ago. So what. You know when Prescott was introduced? Last freaking March. And only time will tell what the final bottomline in performance will be.

I presume you do realize that Prescott never had a projected release of "Last freaking March".

Wingz, don't bother with this guy, if you don't believe the Athlon 64 "is technically superior to anything before it and anything that will released in the forseeable future" then you're obviously an "Intelite"

Oh, I'm sure when he was throwing out the accusation, that he didn't intend it for me! I'm no "Intelite"!!!

 

[dexvx]

Originally posted by: classy

Originally posted by: Duvie

Originally posted by: Mrburns2007
The Athlon 64 will probably be a lot better when everything is finalized and clock is faster.


BTW: The Athlon 64 hit 1.8 Ghz a month or two back so 2.0 Ghz isn't out of the question.

Yeah but the whole thing was supposed to be released some time ago and supposed to have been 2ghz then....We saw hammer mobos back along time ago. It is AMD delaying this whole thing...

I like the rose colored glasses you look through. For AMD's sake I hope you are right. I think the hammer and all the processes involved in the making of this new arhictecture and chip was quite an investment in a rather sparse growth revenue period. They can't take a lackluster release already outdone by its current Barton chip, and a chip that will not ramp well and give them the type of return they need in this chip with longevity.

Rose colored glasses? Wow what color are your glasses? I keep hearing the "Intelites", well it was suppose to be released a long time ago. So what. You know when Prescott was introduced? Last freaking March. And only time will tell what the final bottomline in performance will be.

It was slated Q2 2003 for the last one and a half years.

Seems to me you're the only fanboi around here.

 

[classy]

Originally posted by: Wingznut

Originally posted by: classy
I keep hearing the "Intelites", well it was suppose to be released a long time ago. So what. You know when Prescott was introduced? Last freaking March. And only time will tell what the final bottomline in performance will be.

I presume you do realize that Prescott never had a projected release of "Last freaking March".

So AMD sets a release date for their next big thing and comes up short. Intel sets no release date so they don't look like they are short? Its absolutely amazing some hardware freaks. I swear some of you guys actually sleep with your hardware. And let me address this one more time for the guys wearing their Intel rubbers. Look both companies have made some great releases. But this chip is a major step forward. The chip is bringing to mainstream users a couple of things like never before. So how much superior it is no one really knows yet. But it is superior in theory and we'll just have to wait to see if it translates into actuality. What is such a joke is harping on when it was supposed to be released. Who gives a flying bleep. If performs like it suppose to.................news flash, it doesn't matter when the release date was suppose to be. Whats even sadder is the absolute finality that I see from the Intel fan base here of the failure of the chip. Fact, this review was based on a pre production chip on a pre production board. Those are the facts. So what does this review really bring to light? In reality, nothing, but a lot more of what we already had, speculation.

 

[astroview]

Wow, I've seen a ton of flaming in this thread but also a ton of information. Lots of interesting stuff, most of which is new to me and a little bit confusing.

I think we can all agree at least the future will be interesting with the new P4s coming out and the *non-preview* version of the Athlon64. This article was linked at /. as well, and in that thread an anonymous coward who claimed to work for AMD said this was a beta CPU and not indicative of final results. Let's hope thats the reality for competition's sake at least.


Also, what bothers me is how did Xbitlabs not violate an NDA, or at least not balance it's ability to get future AMD materials by printing these rough numbers?

 

[Wingznut]

Originally posted by: astroview
Also, what bothers me is how did Xbitlabs not violate an NDA, or at least not balance it's ability to get future AMD materials by printing these rough numbers?

They don't always sign NDA's, and don't quite have very high standards for accuracy. That's what differentiates sites like xbit and theinquirer from AnandTech, ArsTechnica, etc...

Originally posted by: dexvx
It was slated Q2 2003 for the last one and a half years.

Not to be picky, but I thought H2 is what Intel's been saying.

Originally posted by: classy
And let me address this one more time for the guys wearing their Intel rubbers.

Maybe I should change my sig to "Prophylactic Technician"?

 

[First]

Originally posted by: classy
And let me address this one more time for the guys wearing their Intel rubbers.

Maybe I should change my sig to "Prophylactic Technician"? [/quote]

ROFL!

 

[dexvx]

Originally posted by: Wingznut

Originally posted by: astroview

Originally posted by: dexvx
It was slated Q2 2003 for the last one and a half years.

Not to be picky, but I thought H2 is what Intel's been saying.

Yea, sorry, it is H2 2003, I got those 2 mixed up.

 

[Nemesis77]

The chip they tested is an early B0 revision. One should NOT base ones opinions regarding Athlon 64 on it. And we don't know about the chipset it used. It was Via chipset (that alone should lower the scores ), but what revision?

As to the 64bit desktops. I'm planning on using 64bit OS. I would love to get those extra GP-registers. Since I use Linux, getting a 64bit OS is not a problem .

 

[Lord Evermore]

B0? Early? In an age when A0 chipsets are being shipped on retail products?

We expect by the time they've reached the B stage, the product is nearly completed. Intel ships B0 products and everyone knows that Intel won't let something out until they're nearly sure it's perfect (and only minor glitches are usually found after that, not crippling performance problems).