dual-core, dual-socket PM@2.3Ghz - validation NOW

[Mephistopheles]

....If <A HREF="http://www.x86-secret.com/pics/news/q404/sossaman.gif" target="_new">this slide</A> is to be believed in. Using a 667Mhz FSB and starting at around ~2.3Ghz. TDP: 31W standard SKUs, 16W LV SKUs....

<A HREF="http://babelfish.altavista.com/babelfish/trurl_pagecontent?lp=fr_en&trurl=http%3a%2f%2fwww.x86-secret.com%2f%3foption%3dnewsd%26nid%3d809" target="_new">Full news story here.</A>

 

[HardWareBoss]

Thanks for the link. Looks good. Plus thats the time frame 06 for my next upgrade. Can't wait to see what's coming out with dual core and 64 bit.

 

[Mephistopheles]

I think it's intended as a workstation part, not a desktop one, hardwareboss... but I'd love to have one of those in my bedroom as well.

Maybe I'll actually manage that, because I'll be doing some heavy mathematical stuff which might really use 4 logical processors.

Can you actually imagine how that would be like? 4 logical processors within maybe 15-18 months from now?... wow.... Up to now, quad-processor rigs were all very, very expensive... It's either Xeon MP, or Opteron 8xx series...

 

[Spitfire_x86]

403 Forbidden

------------
<font color=orange><b><A HREF="http://www.mozilla.org/products/firefox" target="_new">Rediscover the web</A></b></font color=orange>

 

[Ned_Flanders]

Dead link


_______________________
<A HREF="http://www.moviewavs.com/MP3S/TV_Shows/Simpsons/flanderssong.mp3" target="_new">Audio Sig</A>

 

[addiarmadar]

Cant wait to see those on upcomming laptops and how it will effect battery life.

<i><font color=red>Only an overclocker can make a computer into a convectional oven.</i></font color=red>

 

[trooper11]

well um most of this we all thought was true, dont see what is being validated. maybe the launch date of Q1 06 lol, but thats about it.

now when they say ~2.3ghz in speed, I can only assume they mena in total, split across the two cores. i mean this doesnt make me anymore excited aobut it, but its nice to see some possible official info on the core lol.

 

[coylter]

Link only work in explorer...

Signature (up to 200 characters). You may use <font color=blue><b>Markup</b></font color=blue> in your signature

 

[Mephistopheles]

I can only assume they mena in total, split across the two cores.

Why?

My reasoning tells me they'd never dream of doing that! By that time, they'll have 2.26Ghz single-core dothans out! Why launch a dual 1.1Ghz or whatver?

 

[trooper11]

then why dont they say that. you cant base your assumption on some feeling or what you want to see, im going by what they SAID. and they did not say anything about 2.3ghz per core, but instead 2.3ghz for the dual core chip, now how do you infer that means per core?

if you know of other evidence pointing to 2.3ghz per core, then by all means, share it, im just going by what ive seen so far and this new info.

 

[Vapor]

It's actually been semi-common knowledge since the news of the future dual-core processors that the designated speed of the processor would be the speed of a single core (and of course there would be two cores at that speed in the processor).

Both AMD and Intel do not plan on releasing dual core chips at the same clocks as their top single core chip, but both promise (and to a small degree have shown) that the performance of them will be exactly equal to or upto 90+% better than a single core chip at the same speed. If you compare a <i>Digital</i> X40 (Intel's 3.2GHz dual core chip based off two Prescott cores) to a P4 3.2E and NEVER get worse performance and often get 40+% boost, how do you figure that there will be two 1.6GHz cores in there?!

In addition, there's simply no way that these would sell with 3/5ths of the performance at the same clocks (yet a higher price). Don't even let me get into the whole issue of a shared memory bus (lack of bandwidth, the latency, etc.) and what that would mean to performance of two slow cores compared to one core at twice the clocks....

Maxtor disgraces the six letters that make Matrox.

 

[dobster99]

i've never seen anyone (official reps from intel or AMD) refer to speed of a dualcore as the sum of both individual cores. referring to a dualcore set up as 2.3Ghz where each individual core is 1.15Ghz would be just as ridiculous as referring to Xeon dp system as 7.2 GHz because each processor run at 3.6Ghz.

 

[endyen]

By 06, 2.3 may not be fast enough. I would expect opterons to be at ,or at least close to 2.6. The odmc may leave the dothan in it's dust.

 

[Vapor]

I don't see how they could only plan on hitting 2.3GHz, either. Current Dothans hit 2.6GHz with relative ease, and putting two next to each other will not change their ability to hit those clocks (Dothan is far from thermally limited from what I can tell). Maybe Intel is afraid of not fulfilling their roadmaps? (not that it's stopped them recently or anything--they've taken to cancellations pretty darn well, IMO)

As for a dcOpteron, I'm going to guess they'll have a dual 2.8GHz version by '06 (def a 2.6GHz version based on how kickass their 90nm chips are), and Opteron is slightly faster clock for clock in everything except gaming, not to mention it has the odmc (and doesn't even use all the bandwidth it has! [unlike Dothan, which is bandwidth <b>starved</b>]).

Whoa, lightbulb(!): since Dothan is so incredibly bandwidth limited (as evidenced by its excellent OC scaling [15% boost in FSB yields 15% performance boost in most apps--simply incredible]), giving it a dual channel DDR2-667 bus and a 133MHz FSB will eliminate the bottleneck and probably provide enough for the second core as well, allowing Dothan to possibly show it's true colors.

Yet as promising as this may be, you're right in saying that 2.3GHz isn't that extraordinary (consolation may be that they OC well?). We want more, Intel!

Maxtor disgraces the six letters that make Matrox.

 

[Mephistopheles]

This is a much more interesting point if you consider that Intel could try using dual-core dothans in LGA775 with 800Mhz or 1066Mhz FSB... these cores would NOT be starved and would make for one hell of a processor.

 

[trooper11]

well of course ive heard all the talk and rumors that each core would be at the speed they classify, but i didnt hear that exactly confirmed by either amd or intel, thats what i was talking about. did amd or itnel state that clearly or have all sites pretty much just assumed that?

hey i agree single cores that equal the 2.3 owuld be ludacris, but again, there was no clarification.

 

[trooper11]

hey im glad your so excited by dothan, but im still holding my breath. intel hasnt convinced me yet that they have learned thier lesson lol.

but ill say i really hope it can prove to be a competitor, but again, these few overclcoking tests didnt prove enough to me to make me overjoyed lol. there were pieces of info missing and so many unanswered questions.

for instance, i think many are making this sound like its obvious that dothan can just be dropped into the shoes of prescott and suddenly its the new flagship. im taking a much more conservative approach here. i think this is going to take alot more time and effort then some of you may hope.

we already know dual core dothans wont come till Q1 06, so id bet it wil be at least till the end of 2006, probably into 2007, before we see intel make a shift to putting dothan on the desktop completely. now this oculd lal change if intel tears up thier roadmap again lol, and try to get something out faster.

 

[dobster99]

i bet the reason we will be seeing only a small step in speed with the dothan is that alviso/sonoma chipset, will offer a boost in performance with its faster FSB and memory along with the dualcore that will provide additional performance with optimized software. GHz steps will probably come later to give the dothan a longer market life. I mean there is no need to ramp the dothan to super incredible performance when its already the leader in the small/light notebook market. There still the issue of heat and power requirements. Ramping GHz, new faster chipset and dualcore is alot to give the dothan, while at the same time maintaining power and heat requirements.

 

[Vapor]

Need I remind you (and everyone else) who Dothan (and Yonah/Jonah) will be competing against in the coming years: 90nm dual Opterons (aka K9). By the time these chips roll around, K9 will likely be at 2.6 or even 2.8GHz. I don't think a Dothan at 2.3GHz on Alviso can compete with an FX-55, let alone an FX-57. Now double the cores and remember that Opteron uses about 50-60% of its current bandwidth (meaning both cores can get almost full bandwidth) while the cores of Yonah will be fighting for their bandwidth (and still won't have an ODMC)...Intel definitely has an uphill battle right now and from the looks of it, AMD won't be backing down a bit.

Maxtor disgraces the six letters that make Matrox.

 

[dobster99]

Yonah/Jonah is targeted at the notebook segment. There is no direct competition between the upcoming Yonah/Jonah and the Opteron/FX-55/FX-57.

On the notebook front, the dothan doesnt need a giant step in performance. A gradual ramp of performance is all that is needed until AMD offers a equivalent highperformance low power chip.

Porting the dothan to the desktop seems to unlikely since I doubt Intel is willing to admit that the prescott failed and/or to introduce a desk top cpu that lacks x86-64 during or after the introduction XP64. A x86-64 architecture based on the dothan core might replace the prescott, but thats way down the line.

You seriously think that by Q1 2006 AMD will offer a 2.8 dualcore opteron. The dualcore opteron will be a beast, but it won't operate at 2.8GHz by early 2006. AMD already claims that their dualcore will operate 3 to 5 speed steps slower then their single core equivalents. K9s are just the dualcore equvialents of K8s. Any speed you can coax out of individual cores of a K9 can be coax out of a K8 at a factor greater than one. Expecting 2.8 Ghz dualcore opteron at the beginning of 2006 is like expecting a single core opteron running at 3.4Ghz to 3.8Ghz at the beginning of 2006. Since it took the FX, close to 8 months to move from 2.4Ghz to 2.6Ghz. I think its highly unlikely of AMD to boost speeds by 1GHz in a year and half.

Some may say that the current situation with Intel hitting a speed bump as a reason we haven't seen a large speed increase in the K8 architecture during 2004 with AMD not needing to do much to keep a performance lead in the server and desktop market. My response to them is that probably won't change within the next year and a half. Intel is dualcoring the Prescott/Xeon architecture which probably means AMD won't need a 2.8 dualcore K9 for the first half of 2006. What Intel needs is a change of architecture anything less will produce the same issues that revolve around the prescott architecture. Until Intel's issues are resolved AMD will be on cruise control.

 

[trooper11]

"You seriously think that by Q1 2006 AMD will offer a 2.8 dualcore opteron. The dualcore opteron will be a beast, but it won't operate at 2.8GHz by early 2006. AMD already claims that their dualcore will operate 3 to 5 speed steps slower then their single core equivalents. K9s are just the dualcore equvialents of K8s. Any speed you can coax out of individual cores of a K9 can be coax out of a K8 at a factor greater than one. Expecting 2.8 Ghz dualcore opteron at the beginning of 2006 is like expecting a single core opteron running at 3.4Ghz to 3.8Ghz at the beginning of 2006. Since it took the FX, close to 8 months to move from 2.4Ghz to 2.6Ghz. I think its highly unlikely of AMD to boost speeds by 1GHz in a year and half."

now remeber, that statement was aobut the chips to be released in 2H 05, that doesnt mean it has to apply to Q1 06. Just as 90nm parts started at the low end and will eventually become the high end, so will dual core at some point. dual core will start off at the lower end and then once the process is ready, they put it at the high end. Now I would agree that Q1 06 is very tight for getting that process far enough along to put dual core at the high end, but its not impossible, seeing as amd already had a lead development wise to begin with. its very possible they could reach 2.8 for dual core while single cores have only reached 3ghz or so.

 

[dobster99]

"its very possible they could reach 2.8 for dual core while single cores have only reached 3ghz or so."

Aren't K9s just dualcored K8s? Explain to me how you can up the Ghz of individual cores of a dualcore. Yet be GHz restrictive when implementing the same core as a single core cpu especially when the two use the same chipset and are bound by the same power/heat restrictions. A single core will always have a luxury of higher heat output and being more power hungry than individual cores of a dually.

The situation you are talking about can only happen if AMD continue developments on dualcores and stop it on K8. Problem with that is single core aren't going anywhere any time soon. Im sure singlecore K8 will exist in the low end market. It will simply be the Sempron of the server world.

 

[phial]

referring to a dualcore set up as 2.3Ghz where each individual core is 1.15Ghz would be just as ridiculous as referring to Xeon dp system as 7.2 GHz because each processor run at 3.6Ghz.

AMD refers to their full duplex 800mhz HT bus as 1600mhz,

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
Brand name whores are stupid!

 

[dobster99]

Let AMD/Intel do that with their dualcores and see what kind uproar it generates. You know both of us would be typing all types of !#@#^%$& ^%*%!@$#@$% #@%$@%$#@#$.

Imagine a dualcored A64 with its individual cores clocked at 1.7GHz called a A64 5000+, but plays HL2 like a xp3200.
<P ID="edit"><FONT SIZE=-1><EM>Edited by dobster99 on 11/19/04 06:17 PM.</EM></FONT></P>

 

[endyen]

I may have read/interpretted the article wrong, but it seems to me that Sossaman will be a sever chip, not a mobile chip. The concept of putting 2 dual core chips in a laptop just seems strange.
As far as the speed of opterons at that time, a lot depends on the interpretation of "3 to 5 speed steps" If they mean 300 to 500 pr mhz, that would be a lot different from say 300 to 500 real mhz.
Speculation is that amd could already be pushing 3ghz chips out the door, though not in very large quantities. It does not seem unlikely that in a little over 16 months, they would be capable of hitting a few hundred mhz more.
As far as competing with dothans, lets agree to dissagree. Amd is capable of putting out low power opterons that come very close to PM numbers. Could they cut a little on error protection to gain the difference? Would cool and quiet make a big difference? I believe so, but what company would develop a platform to compete with the dothans?