Athlon 64 roadmap

[Spitfire_x86]

<A HREF="http://images.tweaktown.com/news/news_athlon64-roadmap.jpg" target="_new">http://images.tweaktown.com/news/news_athlon64-roadmap.jpg</A>

All the information used in this roadmap was supplied by Virtual Zone.

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new"> My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new"> My Rig</A></b>

 

[Spitfire_x86]

Assuming the roadmap is correct:-

# How AMD can call 2 GHz value A64 "3400+" since the performance "3400+" has equal clock speed and 1 MB L2 cache? Is AMD PR of value A64s based on Celeron?

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new"> My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new"> My Rig</A></b>

 

[eden]

This is very confusing, and when you consider that a jump to 4000+ is 100MHZ for 300pts, something is wrong.

Then you get the 3700+ which is 300MHZ more for 300pts while the 3400+ is 300 pts up BUT BY 200MHZ from the 1.8GHZ 3100+!

Anomalies.

--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue:

 

[spud]

Remeber guys this is AMD we are talking about monkeys run that show (monkeys are stupid ass munching investors that want their divident chques to be big and fat.)

-Jeremy

<A HREF="http://service.futuremark.com/compare?2k1=5341387" target="_new">Busting Sh@t Up!!!</A>
<A HREF="http://service.futuremark.com/compare?pcm=1060900" target="_new">Busting More Sh@t Up!!!</A>

 

[Spitfire_x86]

There's a major difference between 3700+ and 4000+, according to this roadmap.

4000+ will have Dual DDR400, and 3700+ will have Single DDR400 mem. controller. This may be the reason for 100 MHz = 300 PR.

I wonder, what they are going to call the value version of A64. Duron 64?

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new"> My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new"> My Rig</A></b><P ID="edit"><FONT SIZE=-1><EM>Edited by Spitfire_x86 on 06/20/03 03:14 AM.</EM></FONT></P>

 

[flashpan]

This is where a faithful AMD basher/realist comes along and rains on your little parade.

HELLO!!!??? 3400+, 3700+, 4000+, these are all numbers that are going to change. Not because the procs are going to be any better. But because AMD will need to make them look better to the idiots of the earth and bump the PR up to make it look like their doing just fine competing with Intel.

The thing that is pissing you off right now isn't that I'm being the jerk of the world, its that I right, and you know its going to happen.

 

[ChipDeath]

<b>its that I right,

No! I Is much more righter!




---
:tongue:

 

[eden]

Yeah but Opteron has Dual Channel, that makes it too much similar in performance to A64 in the end and their sales could drop if they make the A64 much less costly.
In either case I don't see Dual Channel doing wonders.
Hope nVidia tries first, with their excellent optimized DC mode. (no need to use the extra bandwidth)

--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue:

 

[flashpan]

So your best response is “hey you have a type-O!!!”. To witch my only excuse is, “sorry, but it was 2 AM”

 

[Spitfire_x86]

I guess, AMD may increase the ammount of L2 cache of Opteron, to make it better than 2nd gen Athlon 64. Or may introduce DDR-II memory controller in Opteron.

I don't think that nVidia mem controller will be better than the on-die memory controller. Plus it will make nVidia chipset based mobos much expensive copared to other mobos without memory controller.

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new"> My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new"> My Rig</A></b>

 

[reever2]

roadmaps are subject to change....

 

[eden]

I was basing this off the fact that the speed boost taken from DC DDR on nForce 2, did not come by the extra bandwidth but a special aggressive timing on the banks, and not on the Northbridge (formerly). I was assuming this, otherwise yeah there would be little advantage for nVidia on the K8.

--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue:

 

[endyen]

I wouldn't want to try to peg nvidia's chips to any limit. They have shown that when they set thier minds to a task, they can make good things happen. (and that from an Ati fan) As far as the opteron goes, it's its multiple hyper transport that gives it an advantage, and a few other things, not just it's dual channel ram and 1 meg cache. It seems a bit strange that the 64 will have that much cache, do you thing that is part of a thermal solution?

 

[kojax_neonix]

I think that if AMD doesn't want to be following in intel's footsteps, they should drop the PR rating system and focus on the fact that they have 64-bit processing. people don't really care anymore whether they own a 2.4 or a 2.6 ghz machine, but being able to say that a 64-bit athlon 64 can process twice as many instructions in the same time is something no intel chip can claim, even with hyperthreading

 

[Mephistopheles]

twice as many instructions in the same time

I don't think AMD processors using 64 bit can claim that... And I'm guessing that, by time, you mean clock...

 

[eden]

Hyper Transport helped whenever you had multiple CPUs unfortunately. So, a real comparison would be a single Opteron vs single Athlon 64, to which they come very close.

--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue:

 

[Spitfire_x86]

So you are telling "A64 vs. Opteron" is going to be "P4 vs. Xeon" or "AXP vs. Athlon MP" ?

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new"> My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new"> My Rig</A></b>

 

[reever2]

Its not only for multiple cpu's, its possible to hook up more than HT link to something if it is not being used such as an opteron 2XX running in a 1 processor board

 

[eden]

If you're trying to counter something I said, think again, because I am not saying anything here or stating what something should be.

I only indicated that to know if Athlon 64 can really surpass per clock its bigger brother, it should be compared one-on-one, period.
And who said that Athlon MP isn't pretty much the same per clock as its AthlonXP brother?

EDITOk my first paragraph didn't really make sense as it is contradictory :redface: )

--
If I could see the Matrix, I'd tell you I am only seeing 0s inside your head! :tongue: <P ID="edit"><FONT SIZE=-1><EM>Edited by Eden on 06/22/03 10:54 AM.</EM></FONT></P>

 

[slvr_phoenix]

This is where a faithful AMD basher/realist comes along and rains on your little parade.

HELLO!!!??? 3400+, 3700+, 4000+, these are all numbers that are going to change. Not because the procs are going to be any better. But because AMD will need to make them look better to the idiots of the earth and bump the PR up to make it look like their doing just fine competing with Intel.

The thing that is pissing you off right now isn't that I'm being the jerk of the world, its that I right, and you know its going to happen.

Actually, I disagree. I would wager that those ratings are going to stay exactly the same.

Why?

Because the ratings are based on a system that has <i>already</i> been extended into a realm of meaningless numbers that are over-inflated when compared to Intel's MHz. Basically, the system is <i>already</i> skewed so there is no reason to further change those numbers. It has all been handled in advance and I doubt that any of those processors will have near the performance that their PR would suggest when compared to Intel.

"<i>Yeah, if you treat them like equals, it'll only encourage them to think they <b>ARE</b> your equals.</i>" - Thief from <A HREF="http://www.nuklearpower.com/daily.php?date=030603" target="_new">8-Bit Theater</A>

 

[slvr_phoenix]

I only indicated that to know if Athlon 64 can really surpass per clock its bigger brother, it should be compared one-on-one, period.
And who said that Athlon MP isn't pretty much the same per clock as its AthlonXP brother?

Exactly.

On a chip-for-chip basis the A64 will very likely end up with performance that is almost identical to the Opteron. The real difference is that Opterons can be used in motherboards with multiple HT busses. A64s can't. (And Opterons will probably have a lower thermal output.) So the Opterons will always have their place in the market even should the A64 match or exceed their performance in a single-CPU setup.

And so long as the A64 is given that stupid bloody rating and the Opteron is given just a part number, no one will even care.

"<i>Yeah, if you treat them like equals, it'll only encourage them to think they <b>ARE</b> your equals.</i>" - Thief from <A HREF="http://www.nuklearpower.com/daily.php?date=030603" target="_new">8-Bit Theater</A>

 

[dwellman]

Since some of you have mentioned PR numbers...

No one can deny that it is politically smart for AMD to have the general buying public believe that they have a processor capable of similar performance to their competitor's top model, whether or not they <i>actually</i> do. Still, I wonder if it may work in AMD's favor to use more realistic PR numbers (lower the PR), such that a XP 2500+ or 2800+ actually mirrors or surpases a P4 2.4 or 2.8 GHz based system.

Its an interesting dicotomy that borders on hipocrisy. On on hand, AMD is pushing the idea that clock speed doesn't matter (is it not time for "data flow model"/clockless computing yet?), but then, persits in irrelevant PR rating shenanigans.

My plea to AMD: Please, no more PR numbers for Athlon 64. It makes you look silly.

Regards,

Dave

Intel giveth and Microsoft taketh away.