News Conroe 2 Duo will be beat by AMD this year

[Unrealcpu]

I think AMD is striking back with the release of the Conroe 2 Duo.



I think this is just to hold off the Conroe 2 Duo until AMD release their version of the 65NM chip in 2007 Q1. I also think Intel will release Quad core towards the end of 07. So it will be interesting. intel definitly has the lead and i see intel still coming out on top.

AMD counters Intel's Conroe with dual-socket gaming platform


Wolfgang Gruener

June 1, 2006 17:58



Sunnyvale (CA) - AMD used a meeting with the press and analysts to provide some long-term information about its product strategy. Part of the roadmap is an answer to Intel's flagship processor Core 2 Extreme: AMD will launch a new platform with two sockets for two dual-core processors, handing gamers four physical cores.

In case you haven't read it at TG Daily before: 2006 will be recognized as one of the few years so far where fierce competition between AMD and Intel sparks one of the most interesting years for anyone who uses computers for more than preparing tax reports. Just yesterday, Intel demonstrated its Core 2 Extreme running at 2.93 GHz and lots of headroom for performance increases that are likely to pave the way into a whole new world of gaming and multimedia experience.


As expected, AMD has been working on an answer - and while the company does not have a new architecture, the manufacturer scales the current Athlon64 platform until the arrival of 65 nm production generation in Q4 2006 or Q1 2007. At least on the high end, the company believes that it will retain the performance crown with a system the company calls "4x4."

No, unfortunately this term does not describe four quad-core processor within one system. We are talking more about a "2x2" system - a dual-socket platform that is designed for two dual-core processors. The company did not provide a specific launch date, but mentioned that systems with four physical cores will become available during the second half of this year.

AMD justified the move not only with the increased performance that can be expected from Intel's processors, but also with firm's "roots in the enthusiast market." Product manager Brent Berry told TG Daily that "AMD saw significant success with the FX" line recently and felt that it was important that "the top end" gained performance. He did not provide exact performance or benchmark numbers, but he believes that the "dual-socket platform will be the fastest gaming platform by the end of this year."

It is unclear how capable AMD's 4x4 will be especially in the gaming segment, as increased threading still represents a challenge for many software developers. However, 4x4 is likely to have an immediate impact on computing performance as it will feature two independent DDR2 memory controllers, which, at least in theory, will double the memory bandwidth of the system.

[xombie2000]

FX-62 x 2 = $2000, and a motherboard to support it will probably be $300ish? I'm betting Clovertown and a motherboard will be a little cheaper and out perform it. Honestly, thats just speculation but the 8 way dual socket thing that intel showed at that IDF was really impressive from what little I read about it.

[Unrealcpu]

FX-62 x 2 = $2000, and a motherboard to support it will probably be $300ish? I'm betting Clovertown and a motherboard will be a little cheaper and out perform it. Honestly, thats just speculation but the 8 way dual socket thing that intel showed at that IDF was really impressive from what little I read about it.

I think by the time this hardware comes out say 3-4 months , the FX60 will be 300-400 dollars. AMD will even out the playing field.

[xombie2000]

I think by the time this hardware comes out say 3-4 months , the FX60 will be 300-400 dollars. AMD will even out the playing field.

May want to read this (http://www.digitimes.com/mobos/a20060524A6028.html).

[Unrealcpu]

More news,

Looks like AMD will be back in the game in the next 6 months






AMD's K8L die-shot and main points

AMD's K8L cache design

Four 16-bit or Eight 8-bit HyperTransport Links
K8L details continue to pour in at AMD's Technology Analyst Day

During the AMD Technology Analyst Day, AMD’s CTO Phil Hester rehashed the majority of the K8L information we discussed on DailyTech several days ago, but disclosed further details on specifics. Hester was very specific to refer to these new technologies as simply “new architecture,” and never using the K8L core name.

A major push for AMD’s K8L design is in “modular” component design – meaning everything from L3 cache to memory controllers are developed as individual components and linked together with reusable, robust designs. To some extent, processor design is already modular with libraries and designs that are developed individually. However, Hester insists this new modular approach takes this modular approach even further, claiming that the company is developing “better define the interfaces for each of these building blocks.”

Additionally, Hester revealed some more information about the cache specifics on K8L. Each K8L core will have 64KB of dedicated L1 cache, followed by 512KB of dedicated L2 cache. The base models of K8L will have 2MB of shared L3 cache, but Hester also went on to claim that adding more L3 cache was in the company’s roadmap. One thing AMD representatives have not particuarly touched on is the cache reduction from 64+64KB (data+instruction) to 32+32KB. AMD employees have assured us this move is logical with the addition of L3 cache.

A major feature of K8L is DICE, or Dynamic Independent Core Engagement. Essentially, the ACPI layer will have the ability to dynamically adjust individual cores and crossbars on the processor. Each processor core will have the ability to enter its own power-state, or p-state, allowing a K8L processor the ability to conserve power when the system does not have enough threads to utilize the other processor cores. Intel’s Core processors have the ability to enter c-states on a per-core basis, but the AMD demonstration showed a quad-core K8L processor dip individual cores into full halt.

Opteron servers right now are, for the most part, limited to PCs with eight sockets or less. Part of this is due to the fact that each processor has only three HyperTransport links. Hester announced that the next generation Opteron core will have four 16-bit HyperTransport-3 links running at 2.6GHz each. These four links can reconfigure into eight 8-bit HyperTransport links in a process called “un-ganging,” which is a fundamental feature of HyperTransport-3. Essentially, one could have an eight-socket server with thirty-two fully connected cores. Each processor will be able to take advantage of any of the eight memory banks within one memory hop. The HyperTransport-3 specification claims un-ganging mode can work on the fly, meaning that even a fully connected eight-socket server could dynamically change two 8-bit links into a single 16-bit link during operation to increase I/O at critical moments.

K8L is designed as a native quad-core design, although slides from the Technology Analyst Day also revealed that a dual-core desktop SKU will appear in mid-2007. So far, DailyTech has touched a little bit on the 65nm quad-core code names announced in AMD roadmaps, but to our knowledge the code names for dual-core K8L processors have not been disclosed.




http://images.dailytech.com/nimage/1612_large_slide_50.png

http://images.dailytech.com/nimage/1613_large_slide_51.png

http://images.dailytech.com/nimage/1614_large_slide_59.png

[Blacklash]

After Conroe AMD will have to deal with Kentfield.

Like another said I'd rather a 300+usd Conroe than a 1k chip that still gets its arse beat. When AMD puts out a chip in the 300-500usd price range that beats Conroe stock and when they are both overclocked I'll consider it.

[Haltech]

Wow, that was unexpected.

http://www.amdzone.com/pics/cpus/4x4/4x4.JPG



based off of that, you would still need multiple processors.. A tad bit expensive, no?

[Unrealcpu]

May want to read this (http://www.digitimes.com/mobos/a20060524A6028.html).

So it takes intel to convert their whole lineup to 65nm to beat AMD 90nm ?
Seems to me once AMD converts to 65nm and mobos are like 130 bucks and with a integrated memory controler the performance crown will be back in AMDs favor. :)

For me though i will be going with the 2 core duo until this happens. Then switch back to the 65m integrated memory controller

seems to me in that artical AMD will indeed counter the 2 core duo in the next 4-6 months.

[JumpingJack]

I think by the time this hardware comes out say 3-4 months , the FX60 will be 300-400 dollars. AMD will even out the playing field.

They will be in the red if they go this route.

[Unrealcpu]

After Conroe AMD will have to deal with Kentfield.

Like another said I'd rather a 300+usd Conroe than a 1k chip that still gets its arse beat. When AMD puts out a chip in the 300-500usd price range that beats Conroe stock and when they are both overclocked I'll consider it.


Im sure intel will make a move to quad core and so will AMD. The K8L is designed for quad core but will start out with 2 cores. Intel will definitly take back alittle market share the next 4-6 months but after that you will see AMD striking back with the K8L and taking the performance crown. I just hope the integrated memory controller can with stand -60c on the K8L

[xombie2000]

So it takes intel to convert their whole lineup to 65nm to beat AMD 90nm ?
Seems to me once AMD converts to 65nm and mobos are like 130 bucks and with a integrated memory controler the performance crown will be back in AMDs favor. :)

No because if you'll recall, P4 was 65nm and it was a flop. What it took was a full redesign.


For me though i will be going with the 2 core duo until this happens. Then switch back to the 65m integrated memory controller

seems to me in that artical AMD will indeed counter the 2 core duo in the next 4-6 months.

Not me, I'm an every 18-24 months type of guy for PC upgrades, cheers.

[xombie2000]

Im sure intel will make a move to quad core...

Yes in January, if the roadmap holds true.

[NiCKE^]

Seriously, AMD needs 2 cpu's to compete with Intel and one cpu?
Seems like AMD got not so much to come up with. I would love to see 2xE6800 vs 2xFX-62, and how did AMD Fanboys say earlier, you can't compare Conroe to AM2 and other A64 because it's not next gen cpu. How can we compare 2 cpu's against one now? :)

[xombie2000]

Seriously, AMD needs 2 cpu's to compete with Intel and one cpu?
Seems like AMD got not so much to come up with. I would love to see 2xE6800 vs 2xFX-62, and how did AMD Fanboys say earlier, you can't compare Conroe to AM2 and other A64 because it's not next gen cpu. How can we compare 2 cpu's against one now? :)

This is this same thing you will hear over and over, what is fair is to compare whatever current technology you have to each other because the average consumer will not care if their CPU is 90nm or 65nm. What they will care about is paying $700 more for a processor that performs roughtly the same or perhaps worse.

[kl0012]

I think we will see how 300$ E6600 overclocked to 4GHz will beat super-duper 10 kilobuks AMD system in most benchmarks :D. I think that even 4 AMD cores will not help mutch against 4GHz Conroe.

[Absolute_0]

Dual socket dual core is a desperate measure that won't amount to anything for gaming. Few games use 2 threads, how many games need 4? What a waste, exactly what a gamer needs to spend his money on, ANOTHER CPU. Get a decent processor, an X1800XT/7900GT, 2gb of ram, and enjoy the damn games.

Also considering both parties will be releasing quad cores early on 07, not late in 07, this dual CPU socket seems even more superfluous.

Core 2 Duo will dominate the rest of this year, it's 07 that's the real battle ground.

[IvanAndreevich]

This 4x4 thing is garbage, unless you are looking for a cheap server. 2 x X2 3800+ CPUs gives 4 cores on an ATX (?) board with regular RAM, which is cool.

A die shrink will not save AMD. Conroe has upwards of 30% lead at the same clockspeed. Look at Conroes already doing 3.8 GHz on air. 3.8 GHz * 1.3 = 5 GHz K8 to catch up.

There is no way an AM2 CPU (even 65nm) on any cooling can catch Conroe on water. Anybody care to argue?

[BlackX]

I think we will have K8L ES 3-4 months before its official release date. Then we decide what to do, buy conroe or w8 K8L ;)

[Unrealcpu]

This 4x4 thing is garbage, unless you are looking for a cheap server. 2 x X2 3800+ CPUs gives 4 cores on an ATX (?) board with regular RAM, which is cool.

A die shrink will not save AMD. Conroe has upwards of 30% lead at the same clockspeed. Look at Conroes already doing 3.8 GHz on air. 3.8 GHz * 1.3 = 5 GHz K8 to catch up.

There is no way an AM2 CPU (even 65nm) on any cooling can catch Conroe on water. Anybody care to argue?


You are crazy that you think a die shrink will not save AMD. The K8L will be like a differnet design as well ,like the 2 Core duo. That will be the even match. AMD will take the lead with its On die memory contoller once again with 65nm tech. You are compareing AMDs 90nm to intels redesign 65nm chip.

[Unrealcpu]

I think we will have K8L ES 3-4 months before its official release date. Then we decide what to do, buy conroe or w8 K8L ;)

Yeah then you see will OPPAINTER on FIRE with a grin on his face ,taking back the performance crown on XS with the K8L
:D

[duploxxx]

I think we will see how 300$ E6600 overclocked to 4GHz will beat super-duper 10 kilobuks AMD system in most benchmarks :D. I think that even 4 AMD cores will not help mutch against 4GHz Conroe.

duh i have 285opteron system give me a link to any conroe at the moment that outperforms my cpu score in 3dmark06 = 3560

as far as i have seen here on xs the fastest one was a conroe @3,5ghz 3168

and yes AMD must cut prices, but why cut prices now as long as there is no compatitor in store

if some say AMD will ge down under.... same for intel, you think they will make a lot of proffit with such a cut price and looking at die size of conroe with the cache...

x2 is cheaper to produce on same nm level

[zakelwe]

Looks like K8L has reduced L1 and L2 cache but has L3 cache to compensate. Will this make it faster for desktop ( rather than server where L3 seems to be used quite a lot by Intel in recent years ) ?

Regards

Andy

[Unrealcpu]

Hmm looks like a conroe killer

http://images.dailytech.com/nimage/1612_large_slide_50.png

http://images.dailytech.com/nimage/1613_large_slide_51.png

http://images.dailytech.com/nimage/1614_large_slide_59.png

[kl0012]

duh i have 285opteron system give me a link to any conroe at the moment that outperforms my cpu score in 3dmark06 = 3560

as far as i have seen here on xs the fastest one was a conroe @3,5ghz 3168


Greate. As I said, Conroe needs just 4GHz to beat your setup. :p:

[metro.cl]

i dont like amd's aproach of multi socket for desktop pcs.

how many people will buy a top of the line cpu? few then think about 2x

i think it just takes the fun of overclocking

[kl0012]

Hmm looks like a conroe killer

O yeh? How do you see it from the picture? :nono:

[Unrealcpu]

O yeh? How do you see it from the picture?

:slapass: :slapass: :slapass: :slapass: :slapass: :slapass:

[yotomeczek]

O yeh? How do you see it from the picture? :nono:

propably :D
:wierd:

[BlackX]

O yeh? How do you see it from the picture? :nono:

Its hard to explain.. lets say, need to be educated ;)

[TL1000S]

@unrealcpu..
Powerpoint-slides shown at salez-meeting as "proof of concept" is very very pathetic in this forum..
And maybee you have'nt noticed that you are posting AMD "commercials" in the Intel part of the forum... :D

[Unrealcpu]

@unrealcpu..
Powerpoint-slides shown at salez-meeting as "proof of concept" is very very pathetic in this forum..
And maybee you have'nt noticed that you are posting AMD "commercials" in the Intel part of the forum... "Snorreh".. :D

Ignorence :rolleyes: is bliss. :p:

[mesyn191]

@unrealcpu..
Powerpoint-slides shown at salez-meeting as "proof of concept" is very very pathetic in this forum..
And maybee you have'nt noticed that you are posting AMD "commercials" in the Intel part of the forum... "Snorreh".. :D

Umm, if it was a proof of concept they wouldn't have samples and die shots of it already, it'd still be in design phase. Unless they screw up we'll be seeing chips in 6 months... I mean heck if you don't want to talk about unreleased products we can't talk about Conroe then either right?

I'd hardly call some powerpoint slides a commerical either...

[GoThr3k]

how many people will buy a top of the line cpu? few then think about 2x

the ones that are going for the world records, in i dont know which bench maybe??

[NiCKE^]

You should have posted this in AMD forums, I'm 100% sure that this will take a wrong turn and being AMD Fanboys vs Intel Fanboys. How you buy reading know that AMD will take back the top spot is a mystery, since you got now clue about the performance and if it can match the Conroe. Pure speculations from you side IMO and once again, wrong section.

[grimREEFER]

kentsfield in january
works on 965p cihpset afaik
amd is pretty screwed unless i supppose they can run 8 cpu's on one mobo

[Leeee]

ah all this fighting and noones realised that none of this is actually out in
stores yet.

the next year will be very interesting indeed.

may the best cpu win

[thunderstruck!]

I just look at the thread title and cry. How is this news? Everything here is speculation. Unrealcpu thinks that he has the power to see into the future, and a die shrink is all that it will take for AMD to beat Conroe. I'm no fanboy, but starting a flaming thread like this in the Intel section and repeatedly saying
You are crazy that you think a die shrink will not save AMD. isn't going to make it happen. Sounds like this is wishful thinking. K8L may or may not beat Conroe. How can you be so certain when it's a year away, and you haven't seen it in action. Stop having premonitions unreal, and show some facts.

[Unrealcpu]

I just look at the thread title and cry. How is this news? Everything here is speculation. Unrealcpu thinks that he has the power to see into the future, and a die shrink is all that it will take for AMD to beat Conroe. I'm no fanboy, but starting a flaming thread like this in the Intel section and repeatedly saying isn't going to make it happen. Sounds like this is wishful thinking. K8L may or may not beat Conroe. How can you be so certain when it's a year away, and you haven't seen it in action. Stop having premonitions unreal, and show some facts.


I guess what we can do is bring this thread back from the dead 6 months from now or in 2007 and see who is right :)

I will keep you in mind though;)

[NiCKE^]

Yeah let it die but before change the thread title beucase in 6 months everyone will ask who the f*ck doesn't know that Conroe is Core 2 Duo and not Conroe 2 Duo?

[davefr]

Sound like AMD is pretty desperate by throwing out all this vaporware.

Theoretically a dual FX-62 system will beat a Conroe but then Itanium will beat the FX-62's. What's the point??

[davefr]

No because if you'll recall, P4 was 65nm and it was a flop. What it took was a full redesign.


P4 (ie Northwood and Prescott) were both on 90nm. Northwood was 200mm and Prescott was 300mm.

[iboomalot]

This 4x4 thing is garbage, unless you are looking for a cheap server. 2 x X2 3800+ CPUs gives 4 cores on an ATX (?) board with regular RAM, which is cool.

A die shrink will not save AMD. Conroe has upwards of 30% lead at the same clockspeed. Look at Conroes already doing 3.8 GHz on air. 3.8 GHz * 1.3 = 5 GHz K8 to catch up.

There is no way an AM2 CPU (even 65nm) on any cooling can catch Conroe on water. Anybody care to argue?


its more like 1.15 - 1.20 not 1.3 ;)

[Max Tractor]

It borders on embarrassing, the only plan imo, head down/slash prices until they can be competitive with conroe or whatever they are calling it today

[GoThr3k]

why is it embarrassing, its nice for the people who need to break some insane records
and mayber its a thing to perform better than conroe? i dont know if its desperate, but it sounds really nice
are we complaining about SLI or Crossfire?
didnt think so
and for what reason was SLI introduced? idd, to pawn ATI
so stop complaining about how stupid it is

[iterations]

Hmm looks like a conroe killer
Hmm looks like PowerPoint slides

[Unrealcpu]

Hmm looks like PowerPoint slides


Doesnt INTEL use powerpoint slides? :D ;)

[Max Tractor]

why is it embarrassing, its nice for the people who need to break some insane records


So they intend to sell/procure to those on the paylist how many?



are we complaining about SLI or Crossfire?
didnt think so


Yes, problomatic drivers/set up, buy one now get one later, but the get later one gets discontinued. Sh*t I am not talking about sli/cf on topic pullleesss



so stop complaining about how stupid it is

Who is complaining, not I, I just more or less said it was a sh*t response to getting their a** kicked

[Johnny Bravo]

hmmmmm this is all a bit of a flamebait really, wouldn't this be better suited for the AMD forum? You're pretty much walking into the lions den here and then drop kicking a cub in front of the pride. Perhaps not the wisest of moves mate...

[Fixxxer]

hmmmmm this is all a bit of a flamebait really, wouldn't this be better suited for the AMD forum? You're pretty much walking into the lions den here and then drop kicking a cub in front of the pride. Perhaps not the wisest of moves mate...

QFT

[derektm]

Doesnt INTEL use powerpoint slides? :D ;)

Yes but you speculate on something that is many months away :slapass:

I vote for this thread to be closed before all hell breaks loose.

[GoThr3k]

So they intend to sell/procure to those on the paylist how many?



Yes, problomatic drivers/set up, buy one now get one later, but the get later one gets discontinued. Sh*t I am not talking about sli/cf on topic pullleesss



Who is complaining, not I, I just more or less said it was a sh*t response to getting their a** kicked
you also have people that just buy what is the most expensive...
not only the extreme benchers

problematic drivers? didnt know about that
you can buy 2 now also and have some crazy performance gains over single card

who says its a response to conroe?
what if they had planned this hum, about a year ago?

[Max Tractor]

who says its a response to conroe?
what if they had planned this hum, about a year ago?

This thread, unless I am wrong and should be talking about Blackberry Jam

"AMD counters Intel's Conroe with dual-socket gaming platform"

"Sunnyvale (CA) - AMD used a meeting with the press and analysts to provide some long-term information about its product strategy. Part of the roadmap is an answer to Intel's flagship processor Core 2 Extreme: AMD will launch a new platform with two sockets for two dual-core processors, handing gamers four physical cores."

[GoThr3k]

if you are going that way
i rest my case :)
btw, its a reporter who says that, not AMD ;)
anyhow, i dont care, it thinks its a great idea

[Unrealcpu]

hmmmmm this is all a bit of a flamebait really, wouldn't this be better suited for the AMD forum? You're pretty much walking into the lions den here and then drop kicking a cub in front of the pride. Perhaps not the wisest of moves mate...

Yeah intel was dropkicked , pile drived and body slammed the last 4 years so i can understand your feelings. Long term abuse can be brutal :lol: :rofl:

[b0bd0le]

this thread is boring & pointless

please close it

[Ailleur]

Oh Boy! Cpu die pics on a powerpoint slides! Yes im sure all the 15 years old hardware engineers here can really speculate on how this will rock just by seeing freaking IMAGES!

So let me get this straight, it is NOT okay for anyone to say that conroe currently is the performance winner since its comparing old technology to new technology, but it IS okay to say that this is better than conroe, since we all have nothing but an image!

[viper650]

Oh Boy! Cpu die pics on a powerpoint slides! Yes im sure all the 15 years old hardware engineers here can really speculate on how this will rock just by seeing freaking IMAGES!

So let me get this straight, it is NOT okay for anyone to say that conroe currently is the performance winner since its comparing old technology to new technology, but it IS okay to say that this is better than conroe, since we all have nothing but an image!

take your logic and gtfo!:D

this unreal guy is... unreal. hes either trolling, or just really stupid.

[NiCKE^]

Yeah intel was dropkicked , pile drived and body slammed the last 4 years so i can understand your feelings. Long term abuse can be brutal :lol: :rofl:
So the purpose with this thread for you was to be able to tell that AMD had the topspot for the 4 last years and will get it again? Seriously, get a mod to close this because it has nothing with Intel to do and it will just be AMD vs Intel over and over again...

[lowfat]

take your logic and gtfo!:D

this unreal guy is... unreal. hes either trolling, or just really stupid.


I agree. It looks to me, like he posted this just to start a war.


close the thread.

[Kjaks]

Well, I expect that AMD will make something better than Conroe. And then I expect Intel to make a cpu better AMD, and then AMD and then Intel and then AMD and so on.. That's the way it's got to be, just like Conroe is better performing than A64.

If you look on it in that way isn't it for the best for all of us?

[wildcard]

Either way, it's a good thing for us consumers. I think it's a little too early to tell bit it should be interesting.

[GoThr3k]

Well, I expect that AMD will make something better than Conroe. And then I expect Intel to make a cpu better AMD, and then AMD and then Intel and then AMD and so on.. That's the way it's got to be, just like Conroe is better performing than A64.

If you look on it in that way isn't it for the best for all of us?
QFT

[Ailleur]

Well, I expect that AMD will make something better than Conroe. And then I expect Intel to make a cpu better AMD, and then AMD and then Intel and then AMD and so on.. That's the way it's got to be, just like Conroe is better performing than A64.

If you look on it in that way isn't it for the best for all of us?


Sure, but how about we discuss it WHEN they DO make it.
Im sure VIA could release vhld shots and use design names with a lot of "Xtremes" and "gaming enthousiasts" and "multiquadcpumadness" and all the peons would predict the death of intel and amd.

[CPLB]

http://www.amdzone.com/pics/cpus/4x4/4x4.JPG

Does this mean you can only put two "FX" chips in this 4x4 config? If so that's like $2000+ just for processors alone? Seems like too much to me, expecially when you take into account the two $500+ graphics cards in SLI/ Crossfire that they're advertising the benefits for... Basically it's like $3000+ for just the processors and video cards, not even taking into account the case, PS, RAM, MB...

[charlie]

We will see.


'nuff said.

[Unrealcpu]

take your logic and gtfo!:D

this unreal guy is... unreal. hes either trolling, or just really stupid.


Really stupid? Man where do you get your logic for name calling and accuseing people of trolling. I just put out information that has been posted all over the world and some famous sites like Anadtech, HardOCP and others.
Do you think intel will be king forever and ever just because they have something good now. I think you need to take that atitude some where else. You should be banned

[Unrealcpu]

We will see.


'nuff said.




I like your attitude it reminds me of Yoda in your picture , very wise.

[vintage_guitar]

You mean "See, we will."

[keiths]

All it is, is an AM2 version of dual socket boards instead of 940 opteron. There's only one way this "response" could be meaningful; if a lot of threaded games are about to be released. For AMD, preferably at the same time they release this thing.

[Unrealcpu]

You mean "See, we will."

Wait until 2007
http://www.posterplanet.net/images/jedivideo.jpg

[IluvIntel]

I like the Transmeta CPU - TM5500 Processor, IMO that is most efficient processor in terms of power consumption/performance. As low as 1.5W power consumption.
Better than anything AMD or Intel can boast about, right now ! :banana:

http://www.transmeta.com/crusoe/crusoe_tm5800_tm5500.html

:D ;)

[Unrealcpu]

I like the Transmeta CPU - TM5500 Processor, IMO that is most efficient processor in terms of power consumption/performance. As low as 1.5W power consumption.
Better than anything AMD or Intel can boast about, right now ! :banana:

http://www.transmeta.com/crusoe/crusoe_tm5800_tm5500.html

:D ;)


Thats funny your screen name is ILuvINTEL

[Unrealcpu]

this isnt on powerpoint


http://www.hardocp.com/images/news/1149260964gKMXB1SZgn_1_2_l.jpg

http://www.hardocp.com/images/news/1149260964gKMXB1SZgn_1_3_l.jpg


http://www.hardocp.com/images/news/1149260964gKMXB1SZgn_1_1_l.jpg

[BlackX]

Thats funny your screen name is ILuvINTEL

:D :toast:

[wildcard]

Hey unreal, where were those photos taken?

[crackhead2k]

OOOooOOOoOOoOOOooo
2corex2 hmm but what if conroe does the same?
Anyways i would like to use all for rendering
1 core for rendering ... one for video encode.. other two for gaming adn probably 1 gig mem each so 4 gigs ><

Whats teh realease date?

[Turtle 1]

I am an intel kinda guy. But to be honest I find the 4X4 very interestig. More can go into that Socket than just another CPU.We will have wait and see how this plays out. On this story of AMD buying ATI won't happen Intel would step in and buy ATI out of petty cash.

K8L AMD is talking a H2 07 release thats being very optamistic unless they mean 4th Qt. 07. Also I think Intel really shook their tree. Amd gave away to much info to soon . Intel will have lot of time to make a revision on Penryn 1st qt. 08 . Look for Penryn to up FFU and retire 2 threads in one cycle. Amd really shouldn't have shown this info this soon . Big mistake. They are scrambling.

[terrace215]

this isnt on powerpoint

All they could manage were some crummy 65nm samples running at unknown clockspeed and power?

No 65nm shipments until December?

No K8L until MID 2007 (read: Q3 if they are lucky)

First 65nm shipments top out at 2.4GHz because that's the "volume" portion of the market (per Dirk)? What a joke. Sounds like AMD's 65nm process is still badly broken right now.

[IamAnoobieCheez]

More news,

http://images.dailytech.com/nimage/1612_large_slide_50.png

http://images.dailytech.com/nimage/1613_large_slide_51.png

http://images.dailytech.com/nimage/1614_large_slide_59.png
2MB L3 cache doesn't do all that much. It's the size and latency of L1(more importantly) and L2 cache makes more of difference in performance.

Intel will be continuosly mopping kitchin floors, bathroom floors, livingroom marvel floors, and other tile floors with AMD chips.

[GoThr3k]

OOOooOOOoOOoOOOooo
but what if conroe does the same?

and how would they do that?
AMD has hypertransport but was does intel have? :confused:
its quite easey for AMD do make such things with 70% or maybe 80% more performance on very multithreaded tests
but intel doesnt have HT, so how would they let their cpu's communicate with eachother?

[kl0012]

Its hard to explain.. lets say, need to be educated ;)

So if you think that you're educated enough pls take a look on the following two pictures and then try to explain why K8L will be faster.
http://www.iian.ibeam.com/events/thom001/15535/browser/slides/20060601090732085933/Slide54.JPG
http://www.realworldtech.com/includes/images/articles/merom-2.gif

[BlackX]

doesn't do all that much.., yeah for sure they add it for ZE FUN eh :fact:

nice 1 :rofl:

[BlackX]

So if you think that you're educated enough pls take a look on the following two pictures and then try to explain why K8L will be faster.


U wont understand anyway m8 :D

If seriously there is a whole article about this architecture so please read it by ur self. :slapass:

[GoThr3k]

arent the pics too different to see the real differences?
i dont think the amd pic is detailed enough

a hatrick for BlackX :p:

[kl0012]

If seriously there is a whole article about this architecture so please read it by ur self. :slapass:

link, pls

[gullf1sk]

Ugh this thread is full of unpleasantries and flames :(

[M.Beier]

hehe, indeed it is, but not very clever to put this in Intel section instead of AMD where it belongs, was QUITE predictable that alot of ppl wouldnt like to hear such, weather its true or false...

I belive woodcrest will f**k 2x FX62 pretty bad... - Woodcrest is serversolution, and not 1-way, so, guess multi CPU options's better for Intel as well, since FX62 aint anywhere near conroe in performance...

[Gautam]

Intel begins to be heavily bandwidth starved in multi-way computing. I think most would agree that AMD has the performance edge in the enterprise server market, and Woodcrest will not likely change that. Intel will need to move away from an FSB and shift to CSI before they can truly compete with AMD there.

[uOpt]

AMD is simply playing the multiprocessor card, where they should still be stronger thanks to Hypertransport and intergrated memory controller.

Questions are:

1) how good and SMP-scalable is enthusiast software at that time?

2) do they put a cache-coherent Hypertransport link only into the FX CPUs or into all AM2 CPUs?

If you get two X2 for quad-core in an overclocking board with unregistered RAM, and you have software that makes use of 4 CPUs it would be pretty strong.

[Absolute_0]

Ever consider Intel putting an IMC on a Core 2 based design?

[Fzero]

I'm sorry but i am shocked that no one has brought up the point that Intel already has a working 45nm die......

Also i find it a little unfair to compare a dual core conroe to a dual core twin cpu setup :stick: that is just dumb, it would be like comparing a mustang to a pinto :banana:

[Unrealcpu]

Intel begins to be heavily bandwidth starved in multi-way computing. I think most would agree that AMD has the performance edge in the enterprise server market, and Woodcrest will not likely change that. Intel will need to move away from an FSB and shift to CSI before they can truly compete with AMD there.

long of the opinion that (contrary to the market mind you) microprocessors should come in two basically contrary flavors: ones with very small caches (physically) and LOTS of cores, and ones with LOTS of cache, and only a few small, ultra-fast cores. I don't think there is meaningful ground between (that couldn't be covered by a dual-socket MoBo accepting one chip of each type).

For instance - I'm frequently running an incredibly compute-intensive application that probably doesn't utilize even a small fraction of the cache of the processor. It is totally compute-bound. Being multithreaded, it scales almost exactly with the number of cores (yes, hard to believe, but true). I would love to have this machine with a 16-way or 32-way core, "cache be damned". But it wouldn't be a very good gamers or desktop machine, probably.

Problem is, we can't buy such a beast today. And that's kind of a pain.

If AMD (having the cajunas to do it) were to come out with such a line of chips, they would almost instantly "take over the world" of computing - having a GREAT new idea: buying application-tuned chips.

I mean, really: for the last 20+ years it seems we have been arguing the virtues of this processor (within a family) or that for servers, for workstations, for scientific or medical or network or entertainment computing ... but the actual difference of the whole of the chip line-up is pathetically similar.

Not so would by this contra-contemporary chip. The window of opportunity is just about OPEN: as multicore is causing all software houses to rewrite their code (or at least parts of it) to be multithread aware, it only seems obvious to me that the "new FPU" could very well be adding the second chip to the empty socket - and choosing between another 2-way all-around-great processor, or a 16-way ultra-wicked computing animal.

[CPLB]

http://www.amdzone.com/pics/cpus/4x4/4x4.JPG

Does this mean you can only put two "FX" chips in this 4x4 config? If so that's like $2000+ just for processors alone? Seems like too much to me, expecially when you take into account the two $500+ graphics cards in SLI/ Crossfire that they're advertising the benefits for... Basically it's like $3000+ for just the processors and video cards, not even taking into account the case, PS, RAM, MB...

I swear everyone ignores me... Will this 4x4 setup only work with FX processors like the picture shows? I don't understand.

[Iconyu]

Question, which benchmark exactly would this win? Two FX-62's will still be two FX62's. A E6700 beats one, and most benchmarks seem to top out at 2 threads, wouldn't that kinda make spending U$2400 a bit of a waste?
Gamers have been pretty hard pushed to accept dual core cpus, and now AMD want them to buy two at a time? At least SLI/Crossfire is implimented at driver level, but this is just crazy.

And UnrealCPU, the cpu you're looking for has already been made, it's called the Cell. Professors already want to offer their reproductive organs to it, and I'm sure you two will be very happy together.


I swear everyone ignores me... Will this 4x4 setup only work with FX processors like the picture shows? I don't understand.

Until we see this in silicon, I don't think anyone can answer you.

[uOpt]

Question, which benchmark exactly would this win? Two FX-62's will still be two FX62's. A E6700 beats one, and most benchmarks seem to top out at 2 threads, wouldn't that kinda make spending U$2400 a bit of a waste?

The whole thing stands and falls with software maker's ability to truly multithread their applications.

Most games so far don't seem to really multithread their core code. They come up with some side work for a second core. That doesn't scale to more than 2 CPUs/cores, of course.

But if you look long-term: the software makers will be forced to get their acts together either way, and then the processor maker with the stronger, lower power many-CPU/core system will win and single-core performance won't matter anymore.

Question is: will that take 6 months or 9 years?

[mdzcpa]

Intel begins to be heavily bandwidth starved in multi-way computing. I think most would agree that AMD has the performance edge in the enterprise server market, and Woodcrest will not likely change that. Intel will need to move away from an FSB and shift to CSI before they can truly compete with AMD there.

This is about the most on point comment of this thread yet.

Anyone who truly understands AMDs architecture understands that AMD's upcoming 4x4 will be heavily capitalizing on its strengths. Those who haven't a clue may scoff and laugh but that behavior only demostrates a lack of understanding. Make no mistake, 4 x 4 will be a desktop monster.

In addition, and MUCH more importantly than 4 cores is the open architecture where AMD opens up the Hyper Transport Bus. This is huge. If you don't know why, do some google reading.

CPU to CPU there is no need for arguement as Intel has the better design right now. AMD is countering with platform....the same strength that will continue to cause Intel major fits for quite some time in the enterprise market. By opening up the architecture to many partners, AMD is furthering its platform lead.

CPUs are nice. But they don't do anything for you sitting on your desk. So, whether you like it or not, its going to be a year of CPU vs platform. Performance is performance regardless of how you get it.

BTW, for those that think you'll need quad thread games to take advantage of 4 x4, think again. That's simply not the case.

Last point, 4x4 is already happening. There is nothing vaporware about it. You'll be seeing more of it shortly.

That all said, I'll be nestling down with my new Conroe based Intel system until 4x4 comes out and we see some hard and fast numbers :D Right now, Intel has my attention!

BTW, if I see another flame (especially direct name calling) there will be vacations for everyone involved. I know this thread may not be the best place for AMD news, but its not a license for a flamefest either.

[bigjohns97]

This thread title is misleading

[thunderstruck!]

I just put out information that has been posted all over the world and some famous sites like Anadtech, HardOCP and others.

Please return to those forums then and stop trolling ours.

[Absolute_0]

I think what's more important than raw performance is performance per price. It takes nothing more than a big bank roll to go buy an 8 way Opty server.

What matters as far as the majority of desktop consumers are concerned is what's the best bang for your buck, ~200$ CPU. In the near future anyway, i cannot see this solution as anything other than a single CPU setup.

[bigjohns97]

I really don't see how this affects anyone but servers, in which i think AMD already dominated because of their IMC design.

[IvanAndreevich]

What matters as far as the majority of desktop consumers are concerned is what's the best bang for your buck, ~200$ CPU. In the near future anyway, i cannot see this solution as anything other than a single CPU setup.
Agreed. I have never paid these rediculous prices before getting a dual core Athlon X2 / Opteron 165's :( Back in the day, it was 1700+ @ 2.6 GHz for 70 USD brand new.. when the fastest stock was 2800+ 2.17 GHz. I want those days back :slap:

[railer]

Pathetic thread. Not realy intresting to read anyways, half of it is flames. :fact:

[IvanAndreevich]

You are crazy that you think a die shrink will not save AMD. The K8L will be like a differnet design as well ,like the 2 Core duo. That will be the even match. AMD will take the lead with its On die memory contoller once again with 65nm tech. You are compareing AMDs 90nm to intels redesign 65nm chip.
I don't think you know what a die shrink is. A die shrink is a reduction in process WITHOUT significant architectural change. K8L is not a die shrink :slap: K8L should whoop Conroe in FP performance, but it remains to be seen WHEN it arrives. That's really the big question.

I am comparing what is going to be available on the market at the same time. Intel is generally 1/2 - 2/3 of a tech process ahead of AMD (0.13 -> 0.09 -> 0.065 story repeats).

[Absolute_0]

I think K8L has the capacity to meet, and possibly exceed Conroe at the same clocks. However, i don't think they can catch Conroe's clock speed, even on 65nm.

Think back to the 130nm to 90nm jump, we can compare single cores. The highest clocked 130nm was FX55 Claw @ 2.6 Ghz, and 90nm was FX57 @ 2.8 Ghz and it took them a few months of 90nm before it was released iirc. Only much later did were they able to release a 3 Ghz single core, and it's very rare, expensive, and it's s940 server only.
That's a 400 Mhz increase resulting from 130nm -> 90nm

It's not exact, but apply this to AMD dual cores going from 90nm to 65nm. We have 2.8 GHz 90nm out now, applying what we just had it would be quite a few months of development before we see a 3.2 Ghz dual core. By that time, Conroe 3.33 XE will be out.

And you can't forget, as AMD shifts to 65nm, Intel is prepping 45nm. I'm not saying AMD won't pull out some good stuff, but, I think they'll be fighting an uphill battle against Conroe because they can't match the clock speeds.

[Unrealcpu]

Please return to those forums then and stop trolling ours.



:slapass: I am loyal to XS now look at your join date then look at mine:nono:

[vintage_guitar]

:slapass: I am loyal to XS now look at your join date then look at mine:nono:
Join date means nothing..i have signed up to forums in 2002..looked twice..and went back to post a few times. BUT I SIGNED UP IN 2002 THAT MEANS IM LOYAL! under that logic. Not to mention I've been reading around this one for years..just recently signed up..i guess that means i'm not loyal.

[Gautam]

IMHO, AMD is fundamentally in trouble at this point, as you said, where they have to play catch-up with Intel as far as clocks go (and even IPC at this point). HT does wonders for AMD in multi-way computing, but how many of us even here are likely to buy two FX'es in the near future?

K8L is just a logical extension of this "4x4 design." At a point, they are going to have to stop adding cores and rework their architecture fundamentally in order to beat Intel again on a per-core, per-MHz basis.

This is simply reminscent of Intel dumping heaps of cache with the Prescott. It begins to become impractical at a point, not to mention costly and most likely a power guzzler. It's not an elegant design, but rather a quick, brute force fix. They can't go on like this forever.

[p360stick]

I think K8L has the capacity to meet, and possibly exceed Conroe at the same clocks. However, i don't think they can catch Conroe's clock speed, even on 65nm.

Think back to the 130nm to 90nm jump, we can compare single cores. The highest clocked 130nm was FX55 Claw @ 2.6 Ghz, and 90nm was FX57 @ 2.8 Ghz and it took them a few months of 90nm before it was released iirc. Only much later did were they able to release a 3 Ghz single core, and it's very rare, expensive, and it's s940 server only.
That's a 400 Mhz increase resulting from 130nm -> 90nm

It's not exact, but apply this to AMD dual cores going from 90nm to 65nm. We have 2.8 GHz 90nm out now, applying what we just had it would be quite a few months of development before we see a 3.2 Ghz dual core. By that time, Conroe 3.33 XE will be out.

And you can't forget, as AMD shifts to 65nm, Intel is prepping 45nm. I'm not saying AMD won't pull out some good stuff, but, I think they'll be fighting an uphill battle against Conroe because they can't match the clock speeds.


Yup plus

its looking like intel will always be 1 year ahead nm process like intel is on 65nm and amd is on 90nm. which mean higher clock speeds but even on the same nm process conroe can achieve higher clock speeds than amd.

intel has 2, 4, 8, 16mb cache, I simply think that AMD will always be 1 year behind on the cache.

Intel will also release quad core like almost 1 year early than AMD. 4x4 is just like FX opteron with unlock mutiplier.

[Absolute_0]

I am not sure, but i think both companies will release quad cores at approximately the same time: not long into 07.

[GoThr3k]

btw 45nm is sheduled for Q1 2008 intel
mid 2008 AMD
if you want sources => google, i dont save al my websites i surf to
who cares about a quadcore intel? they are way to bandwidth starved remember 8xx anyone?
just slap 2 dual cores together to keep the crowd happy
kentsfield wil not be a big deal, the true quadcore of intel may get things going
but i doubt i will be superior to the AMD quadcore, and hell, i will not upgrade before quadcore
this sinle core 146@2.8 ghz is nice for all games, games doesnt even improve if i clock higher, i will wait till second half of 2007 for some nice quadcores, doesnt matter if they are from intel or AMD

[ethernal]

This is just a desperation move by AMD, just like Intel did back with their EE's and their emergency dual core.

This is targeted soley to gamers. People who deal with workstations that use multi-sockets are already using Opterons, and that isn't going to change.

And games are not mulithreaded. And to be honest, there isn't going to be a magical day where they suddenly are. Games are inherently linear in their nature. At the very most, you can make maybe two or three "major" threads that require significant computer power. I'm not saying you can't split up a game into dozens of threads, but each of those threads would be so short lived and constantly waiting on other threads that it would be the equivalent to simply using a couple of larger threads.

I don't mean to sound like an Intel fanboy or anything, but this is retarded. The only advantage this will have at this point in time will be on multithreaded driver overhead for the GPU's, but that won't make much of a difference. While you may get a 20% gain for buying an FX-62 over an X2-5000 for $500, you'll get a 5-10% gain buying 2x FX-62's for an extra $1000. People may be rich, and I'm sure some people will do it just for their e-penis, but you'd have to be retarded to do so.

[de redhead]

I am not sure, but i think both companies will release quad cores at approximately the same time: not long into 07.

AMD will be releasing quad core server chips initially, their desktop and mobile chips will be introduced over a half year later. Intel will release their quads similarly to the way they are releasing Woodcrest, Conroe and Merom, within a month or two of each other. His point about 45nm is also very valid. Intel plans to release those mid 2007. This should give them a huge price advantage over AMD.

This solution will probably give AMD an edge at the very high end in certain applications for a short time, but very few will be able to afford it. I certainly won't be able to.

[MicroNecessian]

Agreed Gautam, something will give out or maybe AMD has a fix for it, like brainwashing:rolleyes:

@ Unreal So what you are saying is that Kl8 will beat Conroe just because they say so? If thats the case then maybe you should run for president:

I have opinions of my own, strong opinions, but I don't always agree with them.

[theteamaqua]

This thread title is misleading


yeah no banchmark just some drawings on paper wont help AMD, if what AMD say is true, show some benchmark

besides for$2300 USD = 2x $1000 FX 62 + $300 mobo

who would pay that much money for it?? cant u jsut get a dual socket woodcrest rig??

Edit: in the ase they dont slash the price of FX, is the FX going to be cheaper???

[TL1000S]

Ugh this thread is full of unpleasantries and flames :(

Totally agree... :(

[GoThr3k]

AMD will be releasing quad core server chips initially, their desktop and mobile chips will be introduced over a half year later. Intel will release their quads similarly to the way they are releasing Woodcrest, Conroe and Merom, within a month or two of each other. His point about 45nm is also very valid. Intel plans to release those mid 2007. This should give them a huge price advantage over AMD.

This solution will probably give AMD an edge at the very high end in certain applications for a short time, but very few will be able to afford it. I certainly won't be able to.
45nm of intel is Q1 2008 :slapass:

[Rickster_64]

Yup plus

its looking like intel will always be 1 year ahead nm process like intel is on 65nm and amd is on 90nm. which mean higher clock speeds but even on the same nm process conroe can achieve higher clock speeds than amd.

intel has 2, 4, 8, 16mb cache, I simply think that AMD will always be 1 year behind on the cache.

Intel will also release quad core like almost 1 year early than AMD. 4x4 is just like FX opteron with unlock mutiplier.

Not only is Intela l year ahead, but that year gives them time to hit Rev. 2, etc of a process and have it on the market. Sort of like Rev. D for AMD with 90nm then their 'true' launch of 90nm with Rev. E cores.

[TarTheDark]

With this move AMD are basically saying: "Right now we don't have anything to take on Conroe, so we'll use TWO CPUs."

Points to consider about 4x4:

- poor performance/price ratio

- poor performance/watt ratio

- performancewise it will only deliver if the application runs on all cores - ie most of the time performance will be way below the theoretical limit

[M.Beier]

Agreed Gautam, something will give out or maybe AMD has a fix for it, like brainwashing:rolleyes:

@ Unreal So what you are saying is that Kl8 will beat Conroe just because they say so? If thats the case then maybe you should run for president:

ARGH, lost focus, sweet avatar :D

Anyhow, thread DEFF not suppose too be in Intel section..

[BlackX]

Join date means nothing..i have signed up to forums in 2002..looked twice..and went back to post a few times. BUT I SIGNED UP IN 2002 THAT MEANS IM LOYAL! under that logic. Not to mention I've been reading around this one for years..just recently signed up..i guess that means i'm not loyal.

Its a lovely weather today... :rolleyes:

[duploxxx]

IMHO, AMD is fundamentally in trouble at this point, as you said, where they have to play catch-up with Intel as far as clocks go (and even IPC at this point). HT does wonders for AMD in multi-way computing, but how many of us even here are likely to buy two FX'es in the near future?

K8L is just a logical extension of this "4x4 design." At a point, they are going to have to stop adding cores and rework their architecture fundamentally in order to beat Intel again on a per-core, per-MHz basis.

This is simply reminscent of Intel dumping heaps of cache with the Prescott. It begins to become impractical at a point, not to mention costly and most likely a power guzzler. It's not an elegant design, but rather a quick, brute force fix. They can't go on like this forever.


oh man you don't now :banana::banana::banana::banana: about cpu architecture, the intel architecture is not even a bit futuristic... they stretched the core in any way possible dubbel fsb/large cache 128 instructions etc and they did a nice job. but they cant stretch it forever.you talk like the old days of intel where they will get 10ghz....:stick:

what did amd do on their k8l... the same as intel wider instruction sets... except for a shared cache which is again rather new and intel wil also used it in tulsa.

even intel knows they have to go to multicore, as long as they stay with the fsb bottle they will be in troubles, they will never be the real leader. hyperthreading with mem controller is a much better way to scale your cores. the nb is adding power consumption bottelneck and latency.

[M.Beier]

oh man you don't now :banana::banana::banana::banana: about cpu architecture, the intel architecture is not even a bit futuristic... they stretched the core in any way possible dubbel fsb/large cache 128 instructions etc and they did a nice job. but they cant stretch it forever.you talk like the old days of intel where they will get 10ghz....:stick:

what did amd do on their k8l... the same as intel wider instruction sets... except for a shared cache which is again rather new and intel wil also used it in tulsa.

even intel knows they have to go to multicore, as long as they stay with the fsb bottle they will be in troubles, they will never be the real leader. hyperthreading with mem controller is a much better way to scale your cores. the nb is adding power consumption bottelneck and latency.

Watch your mouth..

And second, Guatam aint a noob, so behave.

[GoThr3k]

no offense but i have to agree with duploxxx

[TL1000S]

This thread seems to be started by a troll as flamebait..
I wonder why so many AMD-"people" feel the need to "trash" Intel/Conroe here in the Intel section..The "sig" of Duploxxx is just pathetic.. :confused: (this also goes for those Intel "boys" that holler "halleluja.. AMD will fall")
Conroe is not officially released... K8L/"core quattro" is even further from relase date..
Why not wait until they have been released, tested.. and tweaked by "mainstream" testers/tweakers before passing judgements...
Architecture may be interesting.. but is not necessarly a "critical success factor".. neither for sales..nor for winning benches (for that matter)..
What is good though is that both companies strive to "beat" each other.. both in raw performance.. and in sales (hint: lower prices).. :D

[Clairvoyant129]

After reading this entire topic (wasting 10mins of my life), I have concluded that AMD fanboys are very insecure.

[FUGGER]

Thread closed.