Before you go and buy a conroe.. think about this

[undeclared]

AMD 4x4 is coming out soon.

Conroe e6600 - $400 or so, right?

X2 3800 - $150
take two of them - $300
OC all 4 cores to 2.5ghz

Pop them on a 4x4 motherboard, for 4 cores total.

Add in reverse multithreading.

Do you guys think that 4 cores acting as one can outperform any conroe solution out there? Cause I sure as hell do.

Furthermore, just remember that on these sockets or what not, they will be able to get stuff like video gpus, except performing a helluva lot faster because it's directly on the motherboard rather than on an expansion slot.

I really like conroe, but I'm waiting to see what 4X4 has in store.

The conroe X6800 supposedly has reverse multithreading, but that is the only one. I want to see reverse multithreading on 4 cores, spending as little as $300 for the cpus.. and probably beating the living crap out of a conroe ;/

Simply cause 4 > 2

;p

 

[imported_Questar]

Originally posted by: xtwells
AMD 4x4 is coming out soon.

Conroe e6600 - $400 or so, right?

X2 3800 - $150
take two of them - $300
OC all 4 cores to 2.5ghz

Pop them on a 4x4 motherboard, for 4 cores total.

Add in reverse multithreading.

Do you guys think that 4 cores acting as one can outperform any conroe solution out there? Cause I sure as hell do.

Furthermore, just remember that on these sockets or what not, they will be able to get stuff like video gpus, except performing a helluva lot faster because it's directly on the motherboard rather than on an expansion slot.

I really like conroe, but I'm waiting to see what 4X4 has in store.

The conroe X6800 supposedly has reverse multithreading, but that is the only one. I want to see reverse multithreading on 4 cores, spending as little as $300 for the cpus.. and probably beating the living crap out of a conroe ;/

Simply cause 4 > 2

;p

Ah, the missinformed. 4x4 will be $1,000 without the MB.

Reverse HT doesn't exist

 

[undeclared]

I saw a screenshot of a bios option to do it

Secondly, we honestly have no idea how much it will cost (otherwise why not just buy a server motherboard with nforce chipset supporting pci-e x16?)

 

[Henny]

It might be out in 2H'07

 

[undeclared]

That long? aw

Well, I look forward to two 4-cores then =0

 

[Keysplayr]

Just because Intel hasn't "announced" it, doesn't mean they aren't working on something to steal thunder from AMD's 4x4. Intel likes to be first at everything. Doesn't always work out that way, but that trend has not gone unnoticed. How far off is 4x4 anyways.

EDIT: Just saw. 2H '07. At least 7 months not including the rest of August. Well, hopefully they can speed that up a bit, and when it gets here, I hope it will be awesome.

 

[thilanliyan]

Originally posted by: Questar
Ah, the missinformed. 4x4 will be $1,000 without the MB.

Reverse HT doesn't exist

Exactly, reverse HT doesn't exist, AMD said they are working on it but may come in like 2010 or when K10 comes.

To the OP, how much was gained going from 1 core to 2 for gaming?? Almost nothing except in certain games that required a patch. Four cores are gonna be even less useful than two for a while yet and by the time 4 cores are almost mandatory, AMD will have something better than 4x4.

Waiting for the next best thing means you'll be waiting forever.

Also, Intel will have Kenstfield (quad-core) out in the beginning of next year...there are already people over at xtremesystems that have tested and posted results for the Kentsfields using currently available motherboards.

Originally posted by: xtwells
I saw a screenshot of a bios option to do it

Are you talking about the "core multiplexing" option on the Intel Badaxe?? I believe that was so Core 2 Duo CPUs can turn off 1 core so that all the cache can be used by the remaining core.

 

[imported_Questar]

I saw a screenshot of a bios option to do it

You were lied to.

we honestly have no idea how much it will cost

http://www.dailytech.com/article.aspx?newsid=3509

 

[Kromis]

Reverse Hyperthreading doesn't exist...because I didn't make it yet! Lol. J/k. All this technology is hard stuff.

 

[Pabster]

You can already run 4x4 with a pair of dual core Opterons.

4x4 isn't gonna be cheap. And if 2Qish '07 is an accurate ETA, Intel will have Kentsfield cranking. In fact they're already out there.

 

[harpoon84]

4x4 is gonna be expensive, consume a shitload of power and have next to 0 application support. Only now are we starting to see apps benefit from dual core. Quad core or 4x4 doesn't mean a doubling in performance over dual core you know.

I think someone is upset that Intel has stolen AMD's thunder of late.

Too much reading of http://sharikou.blogspot.com/ can do that to you. Check it out, very funny AMD fanboy blog. He is a lunatic, but he seems so serious he reminds me of the Iraqi Information Minister: http://www.welovetheiraqiinformationminister.com/

 

[imported_Mongrel]

By the time the motherboards hit the streets X23800's will be selling for $85 (or less...)

I guess that makes it even a better deal....

 

[dexvx]

Only FX processors, not regular X2's will be allowed 4x4 support, according to DailyTech.

 

[Pabster]

Originally posted by: dexvx
Only FX processors, not regular X2's will be allowed 4x4 support, according to DailyTech.

So much for the $85 x 2 idea

Doesn't seem like a surprise to me.

 

[twjr]

Originally posted by: harpoon84
4x4 is gonna be expensive, consume a shitload of power and have next to 0 application support. Only now are we starting to see apps benefit from dual core. Quad core or 4x4 doesn't mean a doubling in performance over dual core you know.

I think someone is upset that Intel has stolen AMD's thunder of late.

Too much reading of http://sharikou.blogspot.com/ can do that to you. Check it out, very funny AMD fanboy blog. He is a lunatic, but he seems so serious he reminds me of the Iraqi Information Minister: http://www.welovetheiraqiinformationminister.com/

Totally agree with the sentiment. 4x4 is going to be for an elite and indugent few that have to have the bleeding edge. it is simply to much money, really i think it should be named $X$. For most people it is a complete waste of money and if you do need 4 cores there are opteron and xeon options.

Checked out that site, crap he has smoked some serious crack that one.

 

[gobucks]

4x4 is not the answer for AMD because:
1) games barely use 2 cores, much less 4
2) only FX's will have 4x4 support (in fact, this is probably one reason why AMD discontinued 2x1MB X2's, so that they can have an FX core with the necessary extra HTT links, and an X2 core without them)
3) because of Conroe's longer pipeline, they are getting about 20% higher clocks than K8s, and they are also 20% more efficient per MHz. This will likely negate any performance benefit for the extra cores.
4) AMD has stated that there is no reverse hyperthreading in development.

K8L won't do it either, because:
1) Intel's Kentsfield will be coming out in the same timeframe
2) K8L's cache scheme is retarded. L1 and L2 caches are staying separate, and in fact are being cut in half (64K L1 and 512K L2, the former of which will hurt performance significantly. AMD's 128K L1 boosted performance immensely). L3 will be unified, but from what i've heard it's only gonna be 256K, and L3 is pretty slow.
3) other performance improvements won't help much either. increased HTT speeds won't help except in large scale, 4P server environments.

This is the first time in a while i have felt AMD has lost its way. they need to re-engineer their chips; boost the IPC, lengthen the pipeline slightly and improve branch prediction, and UNIFY THE ENTIRE CACHE, NOT JUST 256K L3.

 

[goinginstyle]

Originally posted by: harpoon84

Too much reading of http://sharikou.blogspot.com/ can do that to you. Check it out, very funny AMD fanboy blog. He is a lunatic, but he seems so serious he reminds me of the Iraqi Information Minister: http://www.welovetheiraqiinformationminister.com/

You know he was fired from Intel with cause, will not allow unmoderated posts on his site, and changes his stories once something occurs to look as if he suggested or talked about it first. He refuses to prove his claims about websites being paid off, the same websites that have done nothing but praise AMD for the last four years or more. He does not actually own the hardware he raves about, well he does have an older S939 3500+ system with a 7300. Ask him where his degree came from and you will find out he is not actually a Ph. D, the guy lies everytime he speaks.

 

[harpoon84]

Originally posted by: goinginstyle

Originally posted by: harpoon84

Too much reading of http://sharikou.blogspot.com/ can do that to you. Check it out, very funny AMD fanboy blog. He is a lunatic, but he seems so serious he reminds me of the Iraqi Information Minister: http://www.welovetheiraqiinformationminister.com/

You know he was fired from Intel with cause, will not allow unmoderated posts on his site, and changes his stories once something occurs to look as if he suggested or talked about it first. He refuses to prove his claims about websites being paid off, the same websites that have done nothing but praise AMD for the last four years or more. He does not actually own the hardware he raves about, well he does have an older S939 3500+ system with a 7300. Ask him where his degree came from and you will find out he is not actually a Ph. D, the guy lies everytime he speaks.

Yeah, I kind of figured. Only a disgruntled ex Intel worker could hate the company THAT much. How did people find out he was fired though?

Anyway, I only mentioned the site because it's one of the causes of the brain explosions as proven by the OP's example.

 

[goinginstyle]

Originally posted by: harpoon84

Originally posted by: goinginstyle

Originally posted by: harpoon84

Too much reading of http://sharikou.blogspot.com/ can do that to you. Check it out, very funny AMD fanboy blog. He is a lunatic, but he seems so serious he reminds me of the Iraqi Information Minister: http://www.welovetheiraqiinformationminister.com/

You know he was fired from Intel with cause, will not allow unmoderated posts on his site, and changes his stories once something occurs to look as if he suggested or talked about it first. He refuses to prove his claims about websites being paid off, the same websites that have done nothing but praise AMD for the last four years or more. He does not actually own the hardware he raves about, well he does have an older S939 3500+ system with a 7300. Ask him where his degree came from and you will find out he is not actually a Ph. D, the guy lies everytime he speaks.

Yeah, I kind of figured. Only a disgruntled ex Intel worker could hate the company THAT much. How did people find out he was fired though?

Anyway, I only mentioned the site because it's one of the causes of the brain explosions as proven by the OP's example.

He was exposed last spring by some former co-workers (who were glad to see him go) after he posted his usual used diaper trash at a couple of websites. He was fired with cause due to "system abuse" while he was at Intel. System abuse is network or theft lingo at Intel. He has a bad habit of posting flame bait at all of the websites now and on high traffic blog sites. He does this to increase traffic to his site since even some of his zealots have left him now. His crusade for months that Intel was paying off review sites about Conroe's performance came back to haunt him as has most of his "predictions". Even his supporters had a hard time backing him after seeing him change previous posts to ensure it appeared he was right. His calculations are always off and he will take one sentence from a variety of web articles to create a story that supports his viewpoint. He does not have a PH.D. and uses that to try and gain credibility with people. I hear he spams the finance sites to try and get someone to post his side of the AMD/Intel results. He finally got someone at Yahoo Finance to link to his site. He does not own any of the hardware he likes to compare and hence the reason why he always does calculations. He stated in April that Conroe would not run on 64-bit windows, Anand among others were paid Intel pumpers, and that Intel did not actually have Conore available for review as the numbers were nothing more than overclocked P4s and of course all of this has been removed from his blogs. I think he posts here as OCHungry and maybe Snorre as their posts are always quoting his statements and they typically have to "calculate" a result since they never have actual hardware to backup their statements.

 

[goinginstyle]

Sorry about hijacking the thread... Will get back on subject now. 4x4 will be interesting, mainly from the viewpoint of seeing what systems will look like in two years at the upper mainstream performance segment.

 

[Pederv]

Originally posted by: goinginstyle
Sorry about hijacking the thread... Will get back on subject now. 4x4 will be interesting, mainly from the viewpoint of seeing what systems will look like in two years at the upper mainstream performance segment.

Just imagine 2 quad core CPU's and 4 dual core GPU's (8 X 8), bench that!

 

[myocardia]

Originally posted by: goinginstyle
He was exposed last spring by some former co-workers (who were glad to see him go) after he posted his usual used diaper trash at a couple of websites. He was fired with cause due to "system abuse" while he was at Intel. System abuse is network or theft lingo at Intel. He has a bad habit of posting flame bait at all of the websites now and on high traffic blog sites. He does this to increase traffic to his site since even some of his zealots have left him now. His crusade for months that Intel was paying off review sites about Conroe's performance came back to haunt him as has most of his "predictions". Even his supporters had a hard time backing him after seeing him change previous posts to ensure it appeared he was right. His calculations are always off and he will take one sentence from a variety of web articles to create a story that supports his viewpoint. He does not have a PH.D. and uses that to try and gain credibility with people. I hear he spams the finance sites to try and get someone to post his side of the AMD/Intel results. He finally got someone at Yahoo Finance to link to his site. He does not own any of the hardware he likes to compare and hence the reason why he always does calculations. He stated in April that Conroe would not run on 64-bit windows, Anand among others were paid Intel pumpers, and that Intel did not actually have Conore available for review as the numbers were nothing more than overclocked P4s and of course all of this has been removed from his blogs. I think he posts here as OCHungry and maybe Snorre as their posts are always quoting his statements and they typically have to "calculate" a result since they never have actual hardware to backup their statements.

Speaking of being paid by Intel, how long have you been an employee there? Four posts on this site, and at least three of them pertaining to this guy whom you've illegaly disclosed things from his employment at Intel. Hmm, I smell a rat. Please stop posting on this forum, unless you have something to add to a discussion, or some product knowledge to share.

And to the OP, get over it. Intel has finally retaken the performance crown, and will likely retain the performance lead for the next one to two years, minimum.

 

[Viditor]

Originally posted by: gobucks
4x4 is not the answer for AMD because:
1) games barely use 2 cores, much less 4

You mean as of today (and not next year), right?

2) only FX's will have 4x4 support (in fact, this is probably one reason why AMD discontinued 2x1MB X2's, so that they can have an FX core with the necessary extra HTT links, and an X2 core without them)

The 2x1MB X2's were discontinued to cut costs during the price war (cache is very expensive!).

3) because of Conroe's longer pipeline, they are getting about 20% higher clocks than K8s, and they are also 20% more efficient per MHz. This will likely negate any performance benefit for the extra cores.

They are getting ~20% higher clocks because they're 65nm, not because of the pipeline...

4) AMD has stated that there is no reverse hyperthreading in development.

? Where did they state that?

K8L won't do it either, because:
1) Intel's Kentsfield will be coming out in the same timeframe

Kentsfield is an MCM (2 dualcores glued together using the FSB for cache coherency) and K8L is a native quadcore with on-die cache coherency.

2) K8L's cache scheme is retarded. L1 and L2 caches are staying separate, and in fact are being cut in half (64K L1 and 512K L2, the former of which will hurt performance significantly. AMD's 128K L1 boosted performance immensely). L3 will be unified, but from what i've heard it's only gonna be 256K, and L3 is pretty slow.

This is just plain wrong...they aren't cutting the L2 or L1 in half, and they've ALWAYS been seperate (as they are in C2D). K8L will be L1=64k+64k just as X2 is now, and L2=512k+512k...L3 is expected to be 2-4MB!

3) other performance improvements won't help much either. increased HTT speeds won't help except in large scale, 4P server environments.

I don't think you've read the other enhancements very closely...
K8L specs

 

[StrangerGuy]

4x4 is going to be a bust, K8L or not.

 

[myocardia]

Originally posted by: StrangerGuy
4x4 is going to be a bust, K8L or not.

Well, there's no way for us to know for sure, but yeah, it looks like it to me.