AMD tri core

[TechnologyCoordinator]

Somebody on another site leaked the possible names from the new tri-core:

Oopseron

Subprimeron

Tritanic

Quasinodo

Quadconomo

Brokealona

StrikeThree

Oddcore

Refurbium

Unobtainium

Faileron

Bankruptelona

Tri-plegic

TriedandFailed

QuadCoNoMo

Subopteron

Rejecteron

Phuckitron

Defecteron

Tri-Hard

Dual Core Plus

Quad Core Minus

3/4 Quad

Tri-De-Fecta

Tri-to-sell

Lemonade

Triple Cripple

Shortbus

 

[Reynod]

I want my free core (X3) with my X2 Phenom !!!

Re: the earlier post there was a comment I found which refers to the 4th core being "fuzed" off. I'll go find it and post here ... too lazy now.

So this is unlikely that you can run some kind of patch and up the juice and activate it .... looks like it's lazered off.

Engineering ppl please convert my dribble above into something technical?

I like Dual Core Plus ... or X2+ ...

If they sell them to compete with Intel's duallies we might have Intel selling theirs with free steak knives ... heh heh.

It's all good for the customer ... remember that while you exchange boring maths and spit at each other like cats on a fence between two cheap rental properties.

Man If I were a bunny suit worker ... I'd like know ... alas I do not.

 

[shargrath]

Somebody on another site leaked the possible names from the new tri-core:

Oopseron

Subprimeron

Tritanic
etc...

LOL win.
i believe there is a thread with links stating that x3 will only have 3 cores no 4th one disabled. something about since they know how to do it now they can cut,swap and change cores in a short timeframe again this is coming directly from AMD. and again i say thats making a pretty big statement about their yields.

 

[Guest]

All I know is that I wish AMD will catch up to Intel, and ATI will catch NVidia. This coming from someone who's almost always ran Intel/Nvidia setups... For our sake do something already AMD!!!

 

[weskurtz81]

I thought this was interesting, they are doing it to get higher clocks out of it. Disable the core that is holding it back, and bump the frequency up. It is actually a pretty logical idea.

http://www.hardocp.com/news.html?news=MjgyMDQsLCxoZW50aHVzaWFzdCwsLDE=

 

[yomamafor1]

I thought this was interesting, they are doing it to get higher clocks out of it. Disable the core that is holding it back, and bump the frequency up. It is actually a pretty logical idea.

http://www.hardocp.com/news.html?news=MjgyMDQsLCxoZW50aHVzaWFzdCwsLDE=

I would actually call a BS on that one. Although the logic might be true, it costs AMD a lot to waste a perfectly good core.

 

[sedaine]

It does cost them more, but I mentioned before I believe on this very thread that 3 cores are easier to push to 3.0Ghz than 4 cores.

The 3 core may actaully be the sweet spot, if they hit the 3Ghz mark with the Tri then expect to see way better performance than some of Intels Quads. If they do that - it will be a marketing coup for their Quads - Tom, Dick and Harry in BestBuy would naturally assume - if Tri is this good - then Quad should be better. But Quads will not OC as well as Tri.

Why? Because their is a higher % chance that one of the Cores will not OC well in Quad vs. Tri vs. Dual vs. Solo. Make sense?

 

[shargrath]

mmmm
lets think about this if an AMD tricore can beat an intel dual core for the same price by a decent margin and say intels counters with its own tricore that is slightly more expensive. Then where does this leave all the dual cores seeing that their all out performed by the AMD/Intel tricores?
now i think thats a nice little slap in the face to intel seeing thats one area of market is pretty much where they've got AMD beat revenue and performance wise that they might lose money until they release their cut down quad/native tricore to compete. and according to faud the A02 silicon clocks for barcelona to past 3ghz+ which might be that one in one hundred correct reporting.

 

[weskurtz81]

yomamafor1,

It costs them the same, they say it's easy to disable on core. It all really depends on the selling price. They apparently think they can hit a spot where they can make money with them.

 

[Hatman]

Honestly you cannot tell its a waste of money because you have absolutly no idea how much ti costs them to make one. End of.

For all you know it could cost them £20 to make one and they sell it for £200 instead of £250.

Big deal.

I quite like teh idea of higher clocks because of 1 bad core, makes sense. It will fill the price gap between the dual core AMD's and the high clocked Intel dual cores/low clocked quad cores.

 

[Harrisson]

I would actually call a BS on that one. Although the logic might be true, it costs AMD a lot to waste a perfectly good core.

It costs up to 25% of die self-cost, and for shipping cpu much less. Now lets imagine if X4 at 2GHz would be sold as X3 at 2,6GHz, which would sell for more and have higher demand? Exactly, X3. AMD by doing this may actualy earn more than selling low clock quad.

 

[sedaine]

It costs up to 25% of die self-cost, and for shipping cpu much less. Now lets imagine if X4 at 2GHz would be sold as X3 at 2,6GHz, which would sell for more and have higher demand? Exactly, X3. AMD by doing this may actualy earn more than selling low clock quad.

exactly

 

[BaronMatrix]

It costs up to 25% of die self-cost, and for shipping cpu much less. Now lets imagine if X4 at 2GHz would be sold as X3 at 2,6GHz, which would sell for more and have higher demand? Exactly, X3. AMD by doing this may actualy earn more than selling low clock quad.

I don't see why people don't understand the brilliance of this. There are two different measures of yield, one has to do with binning and the other has to do with functional parts.

I would say that the functional part yield for Barcelona is close to 60% maybe higher while the BA rev maybe around 40% for > 2.2GHz. From what was said at another site there is supposedly a B3 rev in testing that should definitely be better for higher than 2.2GHz. I think it was a DigiTimes report about AM2+ mobos.

Now suppose 30% of the 60% that won't go above 2.2GHz have a core that won't do it, you can disable the core and that 30% is now worth 20% more than a dual core Kuma and close to 40% more than a K8.

Another thing people fail to realize about the whole K8/K10 die size issue is that Kuma will be mainstream and is at least 26% smaller than Agena (with 13% for each core) and at least 70% of AMDs products are made with 200mm wafers @ 90nm. Going to 300mm means 2X the chips. 65nm means 30%+ more at the same transistor count.

Or something like that.

 

[TechnologyCoordinator]

I don't see why people don't understand the brilliance of this. There are two different measures of yield, one has to do with binning and the other has to do with functional parts.

I would say that the functional part yield for Barcelona is close to 60% maybe higher while the BA rev maybe around 40% for > 2.2GHz. From what was said at another site there is supposedly a B3 rev in testing that should definitely be better for higher than 2.2GHz. I think it was a DigiTimes report about AM2+ mobos.

Now suppose 30% of the 60% that won't go above 2.2GHz have a core that won't do it, you can disable the core and that 30% is now worth 20% more than a dual core Kuma and close to 40% more than a K8.

Another thing people fail to realize about the whole K8/K10 die size issue is that Kuma will be mainstream and is at least 26% smaller than Agena (with 13% for each core) and at least 70% of AMDs products are made with 200mm wafers @ 90nm. Going to 300mm means 2X the chips. 65nm means 30%+ more at the same transistor count.

Or something like that.

I agree that it is smart for AMD to do this. But the real issue is that their manufacturing process sucks incredible butt.

MCM = Advantage Intel
Monolithic = Disadvantage AMD

 

[BaronMatrix]

I agree that it is smart for AMD to do this. But the real issue is that their manufacturing process sucks incredible butt.

MCM = Advantage Intel
Monolithic = Disadvantage AMD


You have no information about AMDs yields, regardless of how many yield calculation algorithms are posted. Just like no one can make guesses as to Intel's 45nm yields..

To bluntly say, the yields are crap has no factual meaning. We could then say that Intel's yield sucked since they couldn't get below 130W for a year for the highest clocked 65nm.

 

[reconviperone1]

I agree that it is smart for AMD to do this. But the real issue is that their manufacturing process sucks incredible butt.

MCM = Advantage Intel
Monolithic = Disadvantage AMD


You have no information about AMDs yields, regardless of how many yield calculation algorithms are posted. Just like no one can make guesses as to Intel's 45nm yields..

To bluntly say, the yields are crap has no factual meaning. We could then say that Intel's yield sucked since they couldn't get below 130W for a year for the highest clocked 65nm.

All that I know is that I've been playing a demo of bioshock, it runs on three cores on a 360, i dont usually like fps, but this one is great, if i can have a faster speed phenom x3 instead of a slower quad of e6600, I'm running with it.

 

[TechnologyCoordinator]

I agree that it is smart for AMD to do this. But the real issue is that their manufacturing process sucks incredible butt.

MCM = Advantage Intel
Monolithic = Disadvantage AMD


You have no information about AMDs yields, regardless of how many yield calculation algorithms are posted. Just like no one can make guesses as to Intel's 45nm yields..

To bluntly say, the yields are crap has no factual meaning. We could then say that Intel's yield sucked since they couldn't get below 130W for a year for the highest clocked 65nm.

Baron, you don't have any information either, but you actually quote numbers. Nice try!

Here's some info about the manufacturing process:

- It's only launching at 1.9ghz in small quantites
- 2.0ghz is currently vaporware
- They're going to make tri-core processors because of manufacturing issues

I'm willing to make a bet you're going to see a very delayed Phenom as well as delayed 2.5ghz opterons.

 

[Granite3]

Yeah, I know that the Xbox360 CPU is not a powerhouse; it just happens to be the only current CPU out that has 3 cores. 😛
No, the Xbox360 "Xenon" CPU has 3 cores, each of which is capable of 2 threads per clock cycle. The problems lie within it's deep pipeline, narrow execution core, lack of out of order execution, weak to non-existent branch prediction, and the fact it's not even an x86 based CPU. :lol:

Maybe 1 year, as I got my e6600 this time last year, 1 day after launch.

Prior to that the Opty's ruled the p4 trash. And I was a p4 3.2 trash owner!


Netburst was a bust, high heat and less performance, etc.

We need a good AMD cycle to keep prices and performances moving ahead.

I moved from the e6600 to the GO stepping e6750 for a 20c reduction in heat, as well as a 200 mhz OC performance.

If a tricore AMD will do the same, I am all over it!

 

[WSRC]

All I can say is if the Price/Performance is there then I will be getting one. If not then my next build will be most likely an Intel system.

 

[BaronMatrix]

Baron, you don't have any information either, but you actually quote numbers. Nice try!

Here's some info about the manufacturing process:

- It's only launching at 1.9ghz in small quantites
- 2.0ghz is currently vaporware
- They're going to make tri-core processors because of manufacturing issues

I'm willing to make a bet you're going to see a very delayed Phenom as well as delayed 2.5ghz opterons.

I wouldn't say that. Newegg had them quick. Directron has every SKU. AMDZone listed several other places that have most of the SKUs including 8350.

Dell is already advertising them and they should start shipping within a few weeks. That's normal as shipments started around the last day of August.

You just sound so like an Intel employee with a chip on his shoulder it's incredible. It's one thing to theorize, but to bet as if it will matter to either of us? OK, I bet they will launch mid-October, and show up at Newegg in the end of November with systems in time for XMas.

 

[bfellow]

IBM has the 2347 but they won't be available till November 11th!

 

[function9]

A couple things I'm curious about.
1. Is there any info regarding AMD using X2s with one defective core to fill any single core line(s) back in the day?

2. From the articles I've read, this tricore stuff is for Phenom only?

Also, lets please stop throwing around words like brilliant, genius, etc. If they really were doing the tricore thing to try and come up with another quality product for their customers, you better believe they'd be offering it to their enterprise customers, which as of now I don't see anything talking about tricore Opterons.

 

[onestar]

yeild numbers...smumbers

What is the price/performance ratio?
What will it actually cost the consumer?
and ... when can I have one?

Those are the numbers I desire to see.

 

[BaronMatrix]

A couple things I'm curious about.
1. Is there any info regarding AMD using X2s with one defective core to fill any single core line(s) back in the day?

2. From the articles I've read, this tricore stuff is for Phenom only?

Also, lets please stop throwing around words like brilliant, genius, etc. If they really were doing the tricore thing to try and come up with another quality product for their customers, you better believe they'd be offering it to their enterprise customers, which as of now I don't see anything talking about tricore Opterons.

That's who this is for. Chances are it will be for saving on multi-core in OEMs. Server is not a large enough market for this. Dell is. Kyle from HardOCP talked to AMD about it and they are telling him that it isn't an enthusiast chip.

I wonder what people would be saying if Intel did this first.

 

[chuckshissle]

It's kind of odd AMD is using odd numbers for their core counts. They should call it triplets or triad.