AMD prepares three-core processors

[adamsleath]

http://blog.wired.com/business/2007/...control-a.html

While it's a relatively easy matter to disable one core, Intel wouldn't be able to effectively produce a tri-core processor at this point because of the way the company currently approaches quad-core -- slapping two dual-core dies into a single-socket CPU. AMD, on the other hand, can do tri-core due to its Direct Connect architecture, which provides a dedicated channel between the CPU cores and from each CPU out to the system memory.

http://www.xbitlabs.com/news/cpu/dis...914212726.html

[CBONE]

Are these K8 or K10 cores? With Quad cores getting down to $200, how cheap would these need to be for anyone to bother? There must be some factor that hasn't been revealed. Super high clocks or something special.

[KoHaN69]

as mentioned, its not practical fo rintel to make tricores yet since it"s quad-cores are just 2 daul cores

but how does this work form a programmer"s standpoint?

[Donnie27]

as mentioned, its not practical fo rintel to make tricores yet since it"s quad-cores are just 2 daul cores

but how does this work form a programmer"s standpoint?

NQT Intel doesn't have to sell Tri-Core that's really a Quad processor that had one of its cores fail. Now if that 3rd core was a dedicated PPU, APU or etc..... fine. Yes, they (Intel) could add a single core or one that had a failed second core but that'd be silly when Two Dual or MCM or whatever its being called will still be faster than 3cores

[Shintai]

http://blog.wired.com/business/2007/...control-a.html

While it's a relatively easy matter to disable one core, Intel wouldn't be able to effectively produce a tri-core processor at this point because of the way the company currently approaches quad-core -- slapping two dual-core dies into a single-socket CPU. AMD, on the other hand, can do tri-core due to its Direct Connect architecture, which provides a dedicated channel between the CPU cores and from each CPU out to the system memory.

http://www.xbitlabs.com/news/cpu/dis...914212726.html

Thats so far out its not even funny.

Intel could make a tricore that was just as effective as its quads now. But it doesnt make any sense. Because Intel doesnt reuse parts with defective cores, only with cache defects, and those are usually still full size products.
But again, makign a tricore from a single and a dualcore in a MCM package makes no sense for Intel. They dont have to do the same damage control as AMD with the monolith die approach.

Also AMDs "native" quadcore is so far away from native as it can be. Its close to be 4 singlecores on a die with a shared L3 slapped on. Dedicated my rear end. Gotta love PR these days. Just like the 64 SPs being 320, or the P4 internet speed booster. Or all the other nonsense.

I might die of a heart attack the day we get it served cold, factual and truthly.

[BeardyMan]

Thats so far out its not even funny.

Intel could make a tricore that was just as effective as its quads now. But it doesnt make any sense. Because Intel doesnt reuse parts with defective cores, only with cache defects, and those are usually still full size products.
But again, makign a tricore from a single and a dualcore in a MCM package makes no sense for Intel. They dont have to do the same damage control as AMD with the monolith die approach.

Also AMDs "native" quadcore is so far away from native as it can be. Its close to be 4 singlecores on a die with a shared L3 slapped on. Dedicated my rear end. Gotta love PR these days. Just like the 64 SPs being 320, or the P4 internet speed booster. Or all the other nonsense.

I might die of a heart attack the day we get it served cold, factual and truthly.

Cut the crap and just say that intel can't cut it
youre finding the grapes sour that's all .

[Shintai]

Cut the crap and just say that intel can't cut it
youre finding the grapes sour that's all .

Seriously, if you are to troll. Do it better. Making a tricore when you got a quad is the easiest task in the world. But why, it makes no sense when you got the MCM. yield are massively higher on dualcores than on quads. And since Intels CPUs consist of much more cache thats more resiliant due to extra cache cells can make up for cache defects. Its just not logical.

Tricores is only a desperate move to salvage parts of the huge mistake with the native quad.

So perhaps you should cut the crap and just realize AMD is up to crap to its neck with this native BS.

[Cooper]

Thats so far out its not even funny.

Intel could make a tricore that was just as effective as its quads now. But it doesnt make any sense. Because Intel doesnt reuse parts with defective cores, only with cache defects, and those are usually still full size products.
But again, makign a tricore from a single and a dualcore in a MCM package makes no sense for Intel. They dont have to do the same damage control as AMD with the monolith die approach.

Also AMDs "native" quadcore is so far away from native as it can be. Its close to be 4 singlecores on a die with a shared L3 slapped on. Dedicated my rear end. Gotta love PR these days. Just like the 64 SPs being 320, or the P4 internet speed booster. Or all the other nonsense.

I might die of a heart attack the day we get it served cold, factual and truthly.

Not too long to wait

[pH(x)]

I have hope for AMD trying to get this in the market. And as well that it will be successful.

[BeardyMan]

Seriously, if you are to troll. Do it better. Making a tricore when you got a quad is the easiest task in the world. But why, it makes no sense when you got the MCM. yield are massively higher on dualcores than on quads. And since Intels CPUs consist of much more cache thats more resiliant due to extra cache cells can make up for cache defects. Its just not logical.

Tricores is only a desperate move to salvage parts of the huge mistake with the native quad.

So perhaps you should cut the crap and just realize AMD is up to crap to its neck with this native BS.

Says the master of it

I know it must come hard for you that amd has found a nice way to sell every possible chip they have and that it's AMD exclusive , something Intel can't fitt in ...

yeah having hard times eh

[savantu]

Says the master of it

I know it must come hard for you that amd has found a nice way to sell every possible chip they have and that it's AMD exclusive , something Intel can't fitt in ...

yeah having hard times eh

Only a monkey can think that.There is absolutely nothing that prevents Intel from shipping a 3 core MCM , in fact their Xeon 72xx series have 2x2 cores with only 1 enabled on each die.

Bragging about 3 cores is like bragging for having only 8 fingers instead of 10 and being proud of it...

AMD(TM) : It's not a bug , it's a FEATURE!

[BeardyMan]

Only a monkey can think that.There is absolutely nothing that prevents Intel from shipping a 3 core MCM , in fact their Xeon 72xx series have 2x2 cores with only 1 enabled on each die.

Bragging about 3 cores is like bragging for having only 8 fingers instead of 10 and being proud of it...

AMD(TM) : It's not a bug , it's a FEATURE!

there is, otherwise they would have a true quad core.

by the way i would suggest you go out and buy some humor, my god

[freeloader]

New board rule....

If you want to see AMD threads without the crap, just add a handful of people to your ignore list. The threads look absolutely amazing without the BS in them.

[Hornet331]

wth... what does tricore have to do with a "true" quadcore...

thers no "false" or "true" quadcore... its just a other way of realizing it... if there are 4 cores on one cpu package you have 4 cores and nothing more or less...

maybe one aproach is better then the other one, but 4 cores are 4 cores, period.

Damn... i hate all this marketing brainwashed guys...

[Shintai]

New board rule....

Shintai and a few others MUST stay out of AMD threads...god you can single handedly kill threads that some AMD users want to discuss without any comparisons to Intel.

Lol, what you wanna do. Sit and pad one another on the back nomatter what? I thought only Apple people did that...on the other hand you are getting the same PR lately.

A tricore is more a step backwards than forward when we got quads.

And if I can singlehandedly kill a thread it doesnt contain much, if any substance in it from the start.

[Dimitriman]

Hey I would easily buy TriCore if price is close enough to Dual Core. Plus AMD might be able to squeeze some extra Mhurts on those parts so I'm all for it.

[Hornet331]

im more interessted in the pricerange of this tricores. its gona be hard for amd to sell tricores higher than for 300$, especial when you have quads for less then 250$...

[KTE]

I really don't think so. Being the clowns at Inq, I aint bothered what they said or other sources following them, let alone the idea seems odd in business dimensions alone. Guessing or shall I say "pre-emptive lybrinth-like entwined babbles" is a regularity of such drama outlets when they "hallucinate" something is not going smoothly.

AMD has enough SKUs, projects and problems to worry about for the next 4 months, let alone more SKUs in CPUs alone, which adds production, management, support and marketing costs... more than that of a dual core, and its also a new concept. Although it can equate on such charters where quad core yields are poor, or many yields contain only one defective/problematic core, or for higher clocked SKUs, to fulfill that market niche if it doubts the ability to sustain effective yields in higher clocked quad core parts uptil its 45nm move.

Already, AMD has to push itself hard to catch up and/or then regain market share and server/performance/enthusiast/mainstream/value leadership in a minimum three playing fields: GPUs, CPUs and chipsets.

-All 55nm cards replacing Pele were planned to be launched in Q4. The last was a stuff up for many reasons, mainly because it was so damn late (although a good performer later on) and AMD knows that it needs to catch up and do one better with the new iteration nV card releasing now too.
-Opteron 2352, 2354, 2356, 2358, 2360 and their 83xx series counterparts (or maybe just some) need to release in these next upcoming months.
-Agena FX needs to release.
-Agena needs to release.
-Kuma needs to release.
-AMD platforms and chipsets need to release 'n' compete.
-Platform issues for all offerings need to be ironed out ASAP.

Each one is like a K10 of its own and a massive task for any firm, let alone AMD in it's current position.

What's more, Penryn CPUs/chipsets, nV GPUs/chipsets and Intel mobile processors are all due out any time now for AMD to compete with, or suffer. These next months are major potential sales for all the firms.

I personally doubt and won't entertain the idea of a desktop or server tri-core existing by late Q1 08. However, if there's a problem with something internally which AMD knows, such things are always a possible scape goat.

[DeathReborn]

Actually, I now realize that its very feasible and now Fuad has a picture of the die layout too.

It's right here: http://av.rds.yahoo.com/_ylt=A9ibyKT...m/triforce.jpg

Perkam

Sounds like a pyramid scheme to me.

[KoHaN69]

Sounds like a pyramid scheme to me.

[hollo]

So perhaps you should cut the crap and just realize AMD is up to crap to its neck with this native BS.

i didn't hear anyone complaining when intel released a native (shared cache) dual core

tricore will have its uses, especially for overclockers, in that if you disable the weakest core you can clock the other 3 higher (+ lower TDP). and they'll be useful for a while yet, just like a higher clocked single-core processor was better than a lower clocked dual-core for games for a long time after dual core was released, and still is for older games.

not that i'd choose AM2+ over LGA775 until AM2+ goes to 45nm (and by then nehalem will be out, or almost out). but i still think it's a good move on AMD's part.

[thenrz]

This thread has become one somewhat 'humorous' retort after another.


Someone has a good head on his shoulders to even consider the idea of taking what was a bad quad and turning it into a 3 core, hopefully somewhat higher clocked, chip! It seems very logical to go this direction. Intel has the quad market, and AMD is getting there (We can all hope...). We NEED this competition in the CPU market. It's an absolute necessity in order to keep both companies on the bleeding edge of R&D.


An AMD tri-core product may very much be their financial saving grace, if this isn't BS at all.

[Shintai]

i didn't hear anyone complaining when intel released a native (shared cache) dual core

tricore will have its uses, especially for overclockers, in that if you disable the weakest core you can clock the other 3 higher (+ lower TDP). and they'll be useful for a while yet, just like a higher clocked single-core processor was better than a lower clocked dual-core for games for a long time after dual core was released, and still is for older games.

not that i'd choose AM2+ over LGA775 until AM2+ goes to 45nm (and by then nehalem will be out, or almost out). but i still think it's a good move on AMD's part.

Dualcores is another issue. They were alot smaller and therefore alot more realistic in mass production.

With a possible tricore, you dont pick the core to disable. Its done at the factory and its fixed. Permant. You will have just as much luck unlocking the rest of the SPs and memory in a 8800GTS.

If they release a tricore its nothing but damage control. And they will need new SKUs, BIOS support, validation and such. Plus it would just add to the confusion for normal people. It would be utterly desperate.

But lets see, they might just end as dualcores if they gonna reuse the faulty ones.

[Lightman]

Well Hans de Vries did a qiuck photoshop job and came with this:


As you can see 2MB L3 cache and L3 TAG fits perfectly in space of one of the cores and makes die size just slightly bigger than 90nm Windsor core.

source

[perkam]

Any tri-core proc from AMD will be a crippled quad, not a brand new die...this isn't Intel X(

Perkam