AMD Triple Core Information

Unlike some people in this thread, I can see these chips at high enough clock speeds being very attractive alternatives to more expensive lower clocked quads. My only concern would be potential problems with multi-threaded software working optimally on a 3-core cpu since I'm sure nobody designed the software with anything but multiples of 2 in mind.
I doubt that would be a problem.

Many people don't know that quad-socket and higher socket systems have been available for over a decade, way before quad or dual-core processors were being designed. I own an ALR Revolution 6X6 which is a hex socket-8 dating from the mid-late 90s. I can add any number of processors from 1 through 6, odd or even, I like and the system works fine. If software took advantage of any combination of 'cores' way back then, it shouldn't be a problem now.
 
I doubt that would be a problem.

Many people don't know that quad-socket and higher socket systems have been available for over a decade, way before quad or dual-core processors were being designed. I own an ALR Revolution 6X6 which is a hex socket-8 dating from the mid-late 90s. I can add any number of processors from 1 through 6, odd or even, I like and the system works fine. If software took advantage of any combination of 'cores' way back then, it shouldn't be a problem now.

Yeh, why would you code software to work only in 2, 4 8 and so on. Seems easier to just have it dynamic to use what ever is there.
 
I doubt that would be a problem.

Many people don't know that quad-socket and higher socket systems have been available for over a decade, way before quad or dual-core processors were being designed. I own an ALR Revolution 6X6 which is a hex socket-8 dating from the mid-late 90s. I can add any number of processors from 1 through 6, odd or even, I like and the system works fine. If software took advantage of any combination of 'cores' way back then, it shouldn't be a problem now.

You're right of course, I hadn't thought of that. I take exception to your "most people" statement though. You'd have to be pretty noobish to be unaware that SMP existed long before multi-core CPUs ;)
 
You're right of course, I hadn't thought of that. I take exception to your "most people" statement though. You'd have to be pretty noobish to be unaware that SMP existed long before multi-core CPUs ;)

Most people are noobish. Ask 90% (a made up statistic of course) of computer owners not on this forum if they even know what SMP or dual-core, or 2S means and they'll most likely say "Yeah, my computer has dual-cores and it's faster than not dual-cores," or "Oh yeah, I heard about that Pentium Duo Core 2 thingy. I want to put on in my Athalon to make it go faster."
 
Yeh, why would you code software to work only in 2, 4 8 and so on. Seems easier to just have it dynamic to use what ever is there.
Sometimes you might have to time threads?
F@H SMP client beta is running 4 threads right now, but once beta is over they plan to spawn as many threads as there are cores.
 
F@H SMP client beta is running 4 threads right now, but once beta is over they plan to spawn as many threads as there are cores.
Really? I hadn't known about this. Well, I assume there will be a minimum thread requirement to distinguish the SMP client from the single threaded standard client. The tri cores will be good processors to run the SMP client on, providing the clockspeed is over the 2GHz threshold. I heard that Stanford will eventually release a higher-end SMP client with 8 threads, but I guess that's no longer in the works.
 
dual socket tri core motherboards anyone?

What do you call an AMD 4x4 with two tri cores :p

A unimog?
 
What I'm drawing from all this is that AMD will be doing two things to introduce the X3:
a) disable a functionally defective core on a X4 so the remaing three can be sold. (what they'll be doing at first)
b) disable a core that won't scale to meet the speeds their X4 line is selling at. (what they'll be doing later when yields improve)

While this sounds like a good idea for now, what about six months to a year from now when yields and scaling have improved to the point that AMD's quadcores are consistently rolling off the line at 2.5-3.0GHz and the defect rate is low? Would it still make sense to disable a perfectly good core that scales well to maintain an artificially created market segment?

I think your conclusions are correct. What happens when yields improve, however? The 3-core procs go away and the price of the quads fall into the gap, making room for the next line of CPU's that will be coming out right about then to be priced at the top. Considering these triple core procs will be mostly sold to OEMs like Dell, Hp etc, this is a great plan, both as far as economics, marketing, and price/product cycle.


EOL them all.

Really? I hadn't known about this. Well, I assume there will be a minimum thread requirement to distinguish the SMP client from the single threaded standard client. The tri cores will be good processors to run the SMP client on, providing the clockspeed is over the 2GHz threshold. I heard that Stanford will eventually release a higher-end SMP client with 8 threads, but I guess that's no longer in the works.


Heck 8 core aware ? Stanfords software has been very much stagnant of late,None of the clients have had any real major changes in forever. :( Even the SMP client isnt that hot.
 
Most people are noobish. Ask 90% (a made up statistic of course) of computer owners not on this forum if they even know what SMP or dual-core, or 2S means and they'll most likely say "Yeah, my computer has dual-cores and it's faster than not dual-cores," or "Oh yeah, I heard about that Pentium Duo Core 2 thingy. I want to put on in my Athalon to make it go faster."

The inclusion of the common misspelling of "Athlon" definitely completed that joke for me :D
 
dual socket tri core motherboards anyone?

What do you call an AMD 4x4 with two tri cores :p

A unimog?

Either a 6x4, 3x4 or 3x3.

4x4 was a reference to 2 dual cores processors (4 cpus) x 2 dual GPU cards (4 GPU's) right?

So once you go from dual to tri its a 6x4, then if you went to quad it'd be 8x4 ?

confused?! :confused:
 
Lol. I think you guys missed the joke.

A Unimog is a famous world war 2 , 6 wheel drive pickup truck. Kinda like a 4 x 4 but its 6 wheel drive.
 
Has AMD confirmed that the tri cores can be used in dual processor configurations??
 
just wanted to say thanks to all for keeping this thread civil and close to on topic.....

and we all need amd to make money,stay alive and competitive.
 
Has AMD confirmed that the tri cores can be used in dual processor configurations??

There is virtually no chance this would happen. AMD only allows "server cpus" to run on multisocket boards. The fact that one core would be disabled I would guess would automatically cause it to fail the validation process for a server CPU.

 
There is virtually no chance this would happen. AMD only allows "server cpus" to run on multisocket boards. The fact that one core would be disabled I would guess would automatically cause it to fail the validation process for a server CPU.
Figured as much, but AMD does have the FX series that works in the 4X4 platform, which aren't officially server products.
 
Figured as much, but AMD does have the FX series that works in the 4X4 platform, which aren't officially server products.

Yes, but the FX line is also supposed to be the very best AMD has to offer, which a tri-core would not be.
 
Sorry I didn't read every page of replies to the article but something has me confused.

Enough people are seeming to get upset that AMD might be selling some (not all) of the X3's that were slated to be X4's, but had a bad proc on it so it is technically crippled.

This is new? AMD has been doing this for a long time now. As I understand some A64 3200+'s that were released were just X2 4200+ with a core shut down because it didnt perform.

I'm pretty sure Intel has been doing it to...wouldn't suprise me in the least.

Of course they are trying to make money off of it, they would be idiots not to. The X3's will still perform very well and, as everyone else has said, add more to their line up.

If you don't like it then don't buy one. They are made more for the masses who buy their computers from Best Buy and Dell, and i'm sure people will overclock the snot out of them too :p
 
Are you fucking stupid? All current operating systems scale to at least 32 processors.

Woah...

You do realize that most applications don't use more than 1 "core/processor" right? So faster speed > than slower w/more cores/processors.
 
Woah...

You do realize that most applications don't use more than 1 "core/processor" right? So faster speed > than slower w/more cores/processors.

The reason is that you run more than one program on your computer.
 
The reason is that you run more than one program on your computer.


Me thinks you need to clue in and calm down. First of all, most of the programs / applications / games people use do not use more then 1 core, and for running more then 1 thing at a time, HT was good enough for that, Dual Core is good enough for that, never enough to justify 3 or 4 cores, only certain apps will take advantage of multi-core chips, and those apps make the chips worth the $. Video / photo editing, 3d creation / rendering etc. Supreme commander is the only game that really needs more then 2 cores =p
 
Me thinks you need to clue in and calm down. First of all, most of the programs / applications / games people use do not use more then 1 core, and for running more then 1 thing at a time, HT was good enough for that, Dual Core is good enough for that, never enough to justify 3 or 4 cores, only certain apps will take advantage of multi-core chips, and those apps make the chips worth the $. Video / photo editing, 3d creation / rendering etc. Supreme commander is the only game that really needs more then 2 cores =p

Do you still use DOS? The OS on my computer has sundry threads doing a ton of more or less useful tasks. In particular, the one that displays graphics on your screen uses much processing. If supreme commander manages to run two threads by itself, then immediately we can identify the graphics driver as a task for the third core. Add on other shit that inevitably gets left running in the background, such as IM clients, agents for music players, and there are plenty of tasks to occupy processors.

(Also SMT was a good hack to make a terrible processor a little less useless, but it doesn't deserve any recognition now. Notice that no one uses it now)

I can't figure out why there is complaint over having MORe processing resources.
 
Do you still use DOS? The OS on my computer has sundry threads doing a ton of more or less useful tasks. In particular, the one that displays graphics on your screen uses much processing. If supreme commander manages to run two threads by itself, then immediately we can identify the graphics driver as a task for the third core. Add on other shit that inevitably gets left running in the background, such as IM clients, agents for music players, and there are plenty of tasks to occupy processors.

(Also SMT was a good hack to make a terrible processor a little less useless, but it doesn't deserve any recognition now. Notice that no one uses it now)

I can't figure out why there is complaint over having MORe processing resources.

You make some good points. unfortunatly its not the case as you think. Software is not designed to run on "many" processors yet. its designed to run on a faster single core, and it will run faster on 1 core then it will on 16, once software goes into the direction of multi-core env then we can see a real benifit. (yea im talkin about your OS :p)

btw no one is complaining about more processing resources, people are saying they would rther have 3 3ghz cores over 4 2ghz cores.

and SMT was GREAT for its time. and it is being used in intels NEXT generation chip, so I guess its not as useless as you say ^^
 
You make some good points. unfortunatly its not the case as you think. Software is not designed to run on "many" processors yet. its designed to run on a faster single core, and it will run faster on 1 core then it will on 16, once software goes into the direction of multi-core env then we can see a real benifit. (yea im talkin about your OS :p)

btw no one is complaining about more processing resources, people are saying they would rther have 3 3ghz cores over 4 2ghz cores.

and SMT was GREAT for its time. and it is being used in intels NEXT generation chip, so I guess its not as useless as you say ^^

I prefer to compare 3 3ghz cores with 4 3ghz cores so that
 
I prefer to compare 3 3ghz cores with 4 3ghz cores so that

*sigh*.. the point is, one of the cores is "weak", it won't go past 2ghz (or whatever), so they disable it, and now you can get up to 3ghz on the other 3.
 
You make some good points. unfortunatly its not the case as you think. Software is not designed to run on "many" processors yet. its designed to run on a faster single core, and it will run faster on 1 core then it will on 16, once software goes into the direction of multi-core env then we can see a real benifit. (yea im talkin about your OS :p)

btw no one is complaining about more processing resources, people are saying they would rther have 3 3ghz cores over 4 2ghz cores.

and SMT was GREAT for its time. and it is being used in intels NEXT generation chip, so I guess its not as useless as you say ^^

You're missing a key element in your logic. If you're running two programs and the OS on two cores then you'll have some contention for processor time. If you're running them on three cores then the OS can have its own processor resources, program one its own, and program two its own. There's no contention. That's why it's faster even if the individual programs aren't SMP aware. The OS is SMP aware and will assign each thread the most available processor at the time it's run.
 
You're missing a key element in your logic. If you're running two programs and the OS on two cores then you'll have some contention for processor time. If you're running them on three cores then the OS can have its own processor resources, program one its own, and program two its own. There's no contention. That's why it's faster even if the individual programs aren't SMP aware. The OS is SMP aware and will assign each thread the most available processor at the time it's run.

I never said it multi-tasking would not be faster, I simply said multi-tasking doesn't justify the cost of a quad core chip vs a faster dual or 3x core.
 
I never said it multi-tasking would not be faster, I simply said multi-tasking doesn't justify the cost of a quad core chip vs a faster dual or 3x core.

Quad cores are at the same price point that dual cores were three months ago. Hell to get a faster dual core than a Q6600 you'd have to pay more money for it.
 
Quad cores are at the same price point that dual cores were three months ago. Hell to get a faster dual core than a Q6600 you'd have to pay more money for it.
Which should be great news for these triple core procs, as prices fall on the quads, the release price of these triple core systems should as well.


Could someone back me up/refute me on this idea:
I think these 3x cpus should be MORE stable for OC'ing than a 4x system. It seems that by disabling a core on a 4x chip, you have more over all resources available, lower heat, and more bandwidth available to the remaining three cores.
Am I wrong?
 
Which should be great news for these triple core procs, as prices fall on the quads, the release price of these triple core systems should as well.


Could someone back me up/refute me on this idea:
I think these 3x cpus should be MORE stable for OC'ing than a 4x system. It seems that by disabling a core on a 4x chip, you have more over all resources available, lower heat, and more bandwidth available to the remaining three cores.
Am I wrong?
If that's the case, then why even bother getting anything over a single core?

That statement is rather foolish, atleast in my opinion.
 
Which should be great news for these triple core procs, as prices fall on the quads, the release price of these triple core systems should as well.


Could someone back me up/refute me on this idea:
I think these 3x cpus should be MORE stable for OC'ing than a 4x system. It seems that by disabling a core on a 4x chip, you have more over all resources available, lower heat, and more bandwidth available to the remaining three cores.
Am I wrong?


It depends on how the core is disabled. If the completely disconnected every trace going to the core, then it wouldn't be any different as if the core hadn't been there. But if they just soft disable it then it'd still use power and could cause stability problems for the other cores.
 
It depends on how the core is disabled. If the completely disconnected every trace going to the core, then it wouldn't be any different as if the core hadn't been there. But if they just soft disable it then it'd still use power and could cause stability problems for the other cores.


maybe not, doesn't barc have advanced power state or w/e which can completely shut off 1 core?
 
Back
Top