Dual Cores

[Paul5186]

AMD Dual cores will be released in Q2 2005!!

Can you imagine if you got the dosh what kind of system you can build!!!

e.g

Dual Athlon64 cores
Dual nVidia SLI Graphics cards

Now that would be a monster machine!

Personaly I would rather buy and overclock, but hey that just me

 

[gpol2162]

That would be a sweet machine if sotware could utilize dual core machines.

 

[phial]

whered you hear this now?

-------
<A HREF="http://www.albinoblacksheep.com/flash/you.html" target="_new">please dont click here! </A>
<A HREF="http://www.subhi.com/keyboard.jpg" target="_new">This is you, interweb junky</A>

 

[Thobphon]

Sound good

But can't think how much is it?
Dose Dual Core mean twice as much price?

P4 2.8E
GA IK-1100
Gigabyte R9700 Pro
512 DDR 400
2 x 80 GB Maxtor ATA 133
Enermax 360
and 8 Fans

 

[trooper11]

initially i wouldnt be suprised if the first dual core opterons were in the $1000+ price range

 

[Spitfire_x86]

<A HREF="http://www.theinquirer.net/?article=17262" target="_new">Inquirer Link</A>

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

<A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig</A> & <A HREF="http://geocities.com/spitfire_x86/benchmark.html" target="_new">3DMark score</A>

 

[Paul5186]

Yes, they would be expensive at first, but if I decide to buy, I would buy the lowest clocked dual cores.

"I wonder what the overclocking headroom would be like?"
The motherboard manufacturers beter provide the neccesary BIOS options!

I just wish that AMD would release something similar to hyperthreading for its Athlon64's, Then AMD would be in the driving seat.

 

[Spitfire_x86]

I think many dual core CPUs won't overclock so well, because both cores have to be equally good overclocker.

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

<A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig</A> & <A HREF="http://geocities.com/spitfire_x86/benchmark.html" target="_new">3DMark score</A>

 

[Spitfire_x86]

With 2 cores, there's no need for HyperTheading. The system will work like SMP system.

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

<A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig</A> & <A HREF="http://geocities.com/spitfire_x86/benchmark.html" target="_new">3DMark score</A>

 

[Ned_Flanders]

OR, if they did implement an AMD varient of hyperthreading...BANG looks to windows like 4CPUS's!

<font color=blue>My dick is so big, that my dick has a dick. And my dicks' dick is bigger than yours.</font color=blue>

 

[Xeon]

AMD Dual cores will be released in Q2 2005!!

I got a rumor for you too, Intel intends for dual Prescott’s in Q3 2004.

Dual nVidia SLI Graphics cards

You’ll need the 925 or Nforce4 last I checked for that to work.

Inquirer Link

Wow that sure reads to me that hopefully maybe we can show people dual core by Q2 2005. But it’s AMD I would imagine it will show its face around Q4 2005, just like they promised 0.09u Q4 last year.

I think many dual core CPUs won't overclock so well, because both cores have to be equally good overclocker.

I think they will over clock very well, with considerations the data bridge between them will be clock speed, I don’t see timing being an issue. Also the cores should be clocked down 1/2 or more since the CPU should be able to theoretically execute 2x more code, also with cutting the clock speeds they lower the thermal output. If not AMD and Intel will have some seriously hot silicon.

I just wish that AMD would release something similar to hyper threading for its Athlon64's, Then AMD would be in the driving seat.

Last I checked AMD was quoted to have said that they would not go down that avenue, dual cores were the intention.

With 2 cores, there's no need for Hyper Threading. The system will work like SMP system.

Debatable if it will work in that manner. With considerations of limited bandwidth to the memory sub system that exists in dual processor machines today. I could almost say it would be worse on the memory subsystem if 2 processors are on the same bus.

Also there was nothing said on how they are going to execute, are they going to be running 2 independent threads, combined effort on one thread, or 1 thread split up similar to HT?

With considerations that AMD is also not number one is IC logic design such as pre-fetch, memory management, execution resource management. I see the first implementation being quite sloppy. This is where Intel will gain a lot of ground on running more than one thread then spreading the load out. Since Intel’s HT the CPU logic deals with it, telling the OS there are 2 processors but one thread, unless the programmer goes out of their way to code for it. But 2 processors on the same die will be a completely new story. IC logic design will be critical and that’s where Intel will pull ahead technologically.

Xeon

<font color=red>Post created with being a dickhead in mind.</font color=red>
<font color=white>For all emotional and slanderous statements contact THG for all law suits.</font color=white>

 

[trooper11]

i have a fealing that amd dual core in Q2 '05 (opterson of course) is mroe reasonable then dual core prescott in Q3 '04 and more likely.

Wow that sure reads to me that hopefully maybe we can show people dual core by Q2 2005. But it’s AMD I would imagine it will show its face around Q4 2005, just like they promised 0.09u Q4 last year.

i dont rmember amd every saying officialy that it would debut 90nm by Q4 of last year, but if oyu have links to back that up, let me check them out. As far as I am aware. AMD set Q3 '04 as thier target date for volume production. Now AMD did have working 90nm as early as q4 last year (if you check out the roadmap they discussed at a conference earlier this year), but volume production seems ot be on schedual, Q3 isnt over yet. So well will see how that goes.

I think they will over clock very well, with considerations the data bridge between them will be clock speed, I don’t see timing being an issue. Also the cores should be clocked down 1/2 or more since the CPU should be able to theoretically execute 2x more code, also with cutting the clock speeds they lower the thermal output. If not AMD and Intel will have some seriously hot silicon.

well i belive you are correct taht its possible for them to be great overclockers in potential, but its too early to say how far intel or amd will go ot prevent such overclocking. intel especially seems bent on putting a stop to it (although they move in small steps to reach thier goal). perhaps they wil put another block to prevent it and maybe some mobo makers can get around it again, but who knows. overclcoking potential isnt a selling point for me anyway, and it isnt soemthing amd or intel build chips to do, its just a bonus for some users lol. i dont think either side should be given points or points taken away for overclocking.

Debatable if it will work in that manner. With considerations of limited bandwidth to the memory sub system that exists in dual processor machines today. I could almost say it would be worse on the memory subsystem if 2 processors are on the same bus.

and that is exactly why xeon cpus get taken by opterons moreso when they scale to multiple cpus, limited bandwidth on a shared bus. now there has been alot of debate aobut the dsign of amd's dual core, but its possible that you wont see a share nbandwidth problem at all. if a new spec for HT comes along or even the newest 2.0 spec now might be enough to supply both cores if it only has 1 HT link. its still too early to know what design the bus will take.

i also think your point that amd's first dual core will be sloppy while intel's will be much cleaner. i think both have potential to bring out well architectured cpus the first time around. AMD has the background and experience since they desinged the K8 from the ground up for dual core, they have obviously been working the kinks out. Intel alos has long experience at elast on the software end of handling dual core since they have hyperthreading. i think the two have idffernt pieces of the puzzle, amd could have an edge on the hardware implimentation, intel an edge on the coding/software implimentation. im excited to see what both offer.

 

[blackphoenix77]

Why would you type that much?! :frown:

<font color=blue>AthlonXP-M 2500+</font color=blue>
<font color=green>Abit NF7-S</font color=green>
<font color=red>Kingston DDR400 2x256Mb</font color=red>

 

[Xeon]

but if oyu have links to back that up, let me check them out.

I will certainly try and find them since I remember discussing this very thing with a buddy of mine that said that IBM and AMD would come out at the same time Intel did.

overclcoking potential isnt a selling point for me anyway

Wow same with me, I have buddies baffled by my very good water cooling kit that I don’t use to over clock with.

if a new spec for HT comes along or even the newest 2.0 spec now might be enough to supply both cores if it only has 1 HT link

Correct me if I am wrong but I was always under the impression that HT for AMD was link for CPU's and increased bandwidth for PCI AGP ect. I was not aware it was directly accessing the memory subsystem, thought it did its access's through the processor.

Hence why I said the CPU's would probably be data hungry unless they introduced a new socket. Since logically I see them making the processors with exact amount of pins needed (current A64's)

I also only see good usage in software designed around cache locality that memory subsystem locality, due to bandwidth restrictions. But if they have good prefect logic I don’t see it being super critical to its performance.

i also think your point that amd's first dual core will be sloppy while intel's will be much cleaner

Bad wording on that someone is bound to take offence, what I mean though is when Intel sits down and really tries to engineer something. They really can create something great, just as AMD has done.

im excited to see what both offer

So am I.

Xeon

<font color=red>Post created with being a dickhead in mind.</font color=red>
<font color=white>For all emotional and slanderous statements contact THG for all law suits.</font color=white>

 

[trooper11]

yeah i would like to see those links if you can find them, not that i think your not right, i just want to see them.

Correct me if I am wrong but I was always under the impression that HT for AMD was link for CPU's and increased bandwidth for PCI AGP ect. I was not aware it was directly accessing the memory subsystem, thought it did its access's through the processor.

ok here are some links, i cant find the exact one i was thinking of, but these do say similar things:

<A HREF="http://www.opteronics.com/opteron-cpu.htm" target="_new">http://www.opteronics.com/opteron-cpu.htm</A>
look at the HT section where it mentions better memory usage

<A HREF="http://www.opteronics.com/pdf/26635A_HT_System_Design.pdf" target="_new">http://www.opteronics.com/pdf/26635A_HT_System_Design.pdf</A>

thats white paper listed on that same page which goes into detail, it mentions near the beginning about the memory subsystem. when i find the page i was talking about, ill paste it up.

Bad wording on that someone is bound to take offence, what I mean though is when Intel sits down and really tries to engineer something. They really can create something great, just as AMD has done.

ok thats much better wording. I agree they can, they have the resources, but its a question of if they will. its never been a question of resources for intel, thier successes and failures all came when they had the backing to pull it off. it comes down execution, just as with amd.

 

[Mephistopheles]

i have a fealing that amd dual core in Q2 '05 (opterson of course) is mroe reasonable then dual core prescott in Q3 '04 and more likely.

I have a feeling that an Intel Dothan-based dual-core processor is more practical than either of those. Remember, Intel is already developing a quad-core dothan-based Xeon with 64-bit extensions and increased clock in its india research center as we speak. Quad-core Opterons at their highest clock are probably at 200W and beyond power dissipation. Current Dothan cores, however, dissipate only 22~24W of power, which would put a quad-core dothan at its highest current clock (2.0Ghz) at less than 100W power dissipation. And that's over a larger die, which lessens the impact of a high thermal output.

This is exactly why people tend to speculate that the dual-core processors might be dothan-based...

Prescott-based dual-core CPUs are not very smart designs by definition, right? And using Dothan is using one of the best CPU designs out there.

<i><font color=red>You never change the existing reality by fighting it. Instead, create a new model that makes the old one obsolete</font color=red> - Buckminster Fuller </i>

 

[trooper11]

yeah i totally agree about the dothan dual core. my money is on intel bringing out dual core dothans across the board, starting in servers. but down count amd out of quad core just yet. if you look at thier EE line of opterons that are rated at 30W and top out at 2ghz, you can see they are right in line with the dothans. so quad core opterons isnt out of the quest, i have a feeling amd is heading in the multi core direction anyway, and the launch of various low power parts makes that possible. Now considering that those 30w opterons are on 130nm, if amd can somehow pull off 90nm, those numbers could easily drop down to equal a dothan or perhaps even lower, we will have to see.