Sign in with
Sign up | Sign in
Your question

AMD 4x4 = 15 % faster than CONROE in gaming

Last response: in CPUs
Share
July 7, 2006 9:51:24 PM

http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)

More about : amd 4x4 faster conroe gaming

July 7, 2006 9:56:15 PM

Crap that sucks cause for the bang for the buck conroe is still better.
July 7, 2006 9:58:59 PM

it depends on if they let us put cheapie x2'3800's and over clock the hell out of them on the 4x4 platform

then its viable...maybe


4X4 will have some things it will just rule in...and dont forget...some big name games are be optimized for quad cores

Crysis is one of them

its a fun time to be a tech nerd....
Related resources
July 7, 2006 9:59:44 PM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)


Nice try to cover your ass but... nothing.
July 7, 2006 10:04:09 PM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)
That's all conjecture. Anyone could write that, as there are no
supporting links, or anything. I think Multiplexing will increase performance 400%. prove i'm wrong. :roll:
July 7, 2006 10:04:09 PM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)
Do yourself a favour-kill this thread NOW!!
July 7, 2006 10:09:37 PM

well tbh its cheaper to buy a conroe instead of 2 cpus and ONLY get 15% improvement. 2 Graphics cards, 2 cpu's what next? 2 computers to run 1 game?
July 7, 2006 10:11:29 PM

i think we should consult the inquirer on this topic before going any further.
July 7, 2006 10:20:21 PM

Quote:
i think we should consult the inquirer on this topic before going any further.
:trophy: Post of the day post of the day man good job!
July 7, 2006 10:30:01 PM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)
Do yourself a favour-kill this thread NOW!!

:lol:  Shot down by a sock puppet, oh the humiliation.
July 7, 2006 10:34:53 PM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)
Do yourself a favour-kill this thread NOW!!

:lol:  Shot down but a sock puppet, oh the humiliation. You think thats embarrising? Try losing your testicles in a knife fight to a midget.
July 7, 2006 10:48:53 PM

As much as I'd like to give you 5 for the family guy call, I'm afraid I can only give 3.
July 7, 2006 10:50:58 PM

Quote:
As much as I'd like to give you 5 for the family guy call, I'm afraid I can only give 3.
I know but atleast it's from you the master of wit and sarcasm on these forumz so ehh i cant complain.
July 7, 2006 10:51:13 PM

Quote:

:lol:  Shot down but a sock puppet, oh the humiliation.


Thanks for that, Action_Capon!! :lol: :lol: 
July 7, 2006 10:51:57 PM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)


Only 15% wow for 2 extra cores that isn't too much.
July 7, 2006 11:52:06 PM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)


Quote:

Thus we can say that theoretically, AMD’s 4x4 Platform can outperform Core 2 Extreme by 10-15 percent in a condition where Reverse Hyper Threading is implemented.


1. its theoretical, and IF RHT is implemented
2. the website doesn't state which processors they used for 4x4, when they concluded that 10~15% can be gained. i'm assuming its 2 x FX-62
3. if the above assumption stays true, then if $2000CPU can't outperform one processor @ $1000, isn't it a little bit too pathetic?

someone mentioned putting two X2 3800+ into a 4x4, and OC to kill Conroe.

no shit.. you're comparing two processors OCed to the extreme, with one non-OCed processor. PD can own a FX-62 by severly OCing it. don't forget that Conroe has better OC ability than any of X2 or FX lineups, due to its 65nm.

to be honest, i think 4x4 is a total joke, unless you multi-task the hell out of the computer. that's a different story.
July 7, 2006 11:54:55 PM

Quote:
i think we should consult the inquirer on this topic before going any further.

i think i've got better sources than INQ.

sharikou.blogspot.com
madmodmike.blogspot.com
July 8, 2006 12:36:04 AM

Quote:
i think we should consult the inquirer on this topic before going any further.


and... Shot down by a Noob... :p  :lol:  :mrgreen: :tongue: :trophy:
July 8, 2006 12:38:33 AM

yeah well 4x4 is a quad core so of course it'll be faster than a dual core. just wait until intel releases their own quad core.




case closed.
July 8, 2006 1:13:05 AM

Or.. you could just O/C a conroe to 3.6 Ghz, on air i might add, and beat two FX 62s for hmm 300 dollars?


Not trying to be a fanboy of intel, but 4x4 is a total waste of money.
July 8, 2006 1:31:23 AM

Yeah right now the fastest chip cost 1,000.00 which is a shit load of money, especially if you buy 2!!!! For the performance upgrade it is not much. But you go to remember the AMD is cutting there prices down for there CPU's So if you buy two FX-62's you can be looking at 900.00


If you over clock them that system will be one fast mother!!!!!
ONLY TIME WILL TELL WHO WILL HAVE THE BETTER SET UP. Even then someone else will have somthing faster in a couple of months!!! LOL
July 8, 2006 1:34:09 AM

That site is just speculating, they have no hard evidence proof of what they say.....there's no actual hardware to test this yet!!

Besides, two dual core CPUs to beat a single dual CPU by ONLY 15%???..................That performance gap aint that BIG!!!!!......I be interested in what that single dual core is on to only fall behind 2 dual cores by 15%!!!

If so, if Intel were to come up with similar stuff (2 dual core socket motherboards for Conroe).........we can also assume that the 2 of the fastest Conroes in tandem will out perform the fastest FXs by 50% or more?????

But then again.......its just some "Reverse Speculation"!!!!
July 8, 2006 1:46:33 AM

errr....


2-Dual Core FX-62 against 1-Dual Core Core2Extreme... ($2000vs.$999 = 10-15%)


RHT are not proven to run succesfully on AM2-Cores... they've got lil' problem coz' their multi-core chip does'nt share its cache...

Q:what would happen when non-shared cache are running RHT?

A:well, expect a lot of cache miss, branch mis-predict, when a single threaded apps start to run on multi-core chip.

CMT is not finished.. well, i guess when they perfect "Mitosis" an IA-64's (Itanium) technology to their current line-up then it would be interesting..

and last.. Core2Extreme does'nt stop at 2.93Ghz.. oh wait until Q4 and you will see lot of Core@Extreme.. (i guess = X6800-2.93Ghz / X6900-3.2Ghz / X8000-3.5Ghz)
July 8, 2006 2:18:43 AM

Quote:
(...) to be honest, i think 4x4 is a total joke, unless you multi-task the hell out of the computer. that's a different story.


Not necessarily a joke and not necessarily for intensive multitasking, only; theoretically, the 4x4 concept is very appealing, since it could allow CPUs & co-processors (ASICs, FPGAs, ...) to work together.

The whole point is: r-HT hasn't yet seen the light, the 4x4 is just a plain concept, for the moment & a 10 to 15% estimated performance increase (compared to a Core 2 Duo Extreme - and assuming 2xFX-62), is but utter speculation... about a concept. For the time being, that's about it.


Cheers!
July 8, 2006 3:02:40 AM

Quote:
i think we should consult the inquirer on this topic before going any further.


LOL
July 8, 2006 3:10:32 AM

Maybe this is off-topic, but here goes.

Instead of using the 4X4 just for more CPU's, would it be possible to use the different sockets to do different things? Some Ideas I can think of would be physics chips, graphics chips, or some sort of bundle chip, like a second southbridge, that could do network, sound, SATA etc. Or, here's an idea: some REAL integrated graphics so that a mainstream buyer could play some games without the need for a $200+ GPU.
July 8, 2006 3:16:14 AM

whats the point of 4x4? Its not like its a special processor. All it is is 2 dual core processors and 2 dual core graphics cards. Do you REALLY think that you can't do the same EXACT thing with 2 conroes? You may need to get a woodcrest instead of conroe, but whats the difference? You can even get a DP motherboard, use one dual core CPU, and then upgrade later. This is all the platform and has nothing to do with a CPU. You still need to pay for 2 EXTREMELY expensive graphic cards that will cost more than the 2 CPUs.
July 8, 2006 3:35:16 AM

That site doesnt even have benchmarks, and its statements are completly theoretical. Ill guarantee the 4x4 platform wont beat conroe, because of the simple fact that most games/software arent even dual core optimised, so what is the capability of doing 4 hardware threads going to matter when not even 2 software threads are going to be present... its not going to matter. And even if it did win by 15 % is that supposed to be a feat ? 4 cores versus 2... thats a complete loss in my books not a win, its like nVidia and their lame 7950 they took the hillbilly aproach to beat ATI and just strapped two cards together and called it one. Amd is doing the same thing, they know they have been beaten archetecturally, so they are just going to add a socket and bamshabam they have good performance with an extremly high pricetag. Sorry if its not in the utmost appealing. It must be entertaining for Intel watching AMD scramble like this, because I would be laughing right now if I was a part of intel.
July 8, 2006 3:49:43 AM

Quote:
its like nVidia and their lame 7950 they took the hillbilly aproach to beat ATI and just strapped two cards together and called it one.
So even when it's competitively priced like Nvidia's solution, it’s still a bad solution?
July 8, 2006 3:51:32 AM

I find it kind of lame that they couldnt beat ATI, so the took the lazy way out and strapped two cards together, because their archetecture isnt fast enough to beat out the X1900XTX without doing what they did.
July 8, 2006 4:00:53 AM

not real news. Just a bunch of crap and opinion that does not count. 4x4 is just a media hype for now. No prototype available, just a concept so far. Even if it is real, not everybody can afford it. I would imagine the motherboard would be pricey to handle 4x4. 4 video cards inside one case is some hot stuff. I wonder if there be issue in the long run for running such as setup.
July 8, 2006 4:19:05 AM

i dont think INTEL will be laughing at quad core dual socket K8L's

thats said...there are big name games being optimized for 4 cores

whos to see CRYSIS cant be coded to offload AI routines to CORE 3 and 4


and let core 1 and 2 run the engine with no AI routine bog down..

you guys think small.....
July 8, 2006 4:22:15 AM

Quote:
i dont think INTEL will be laughing at quad core dual socket K8L's


They'll be laughing in 2008 when dual socket K8L's come out.
July 8, 2006 4:42:32 AM

What games are these, because I certainly dont know about them, and there hasnt been any announcement of the cryengine 2 being heavily multithreaded. We can think outside of the box and say ok they could do this and this, but they simply arent. I was just talking about Intel laughing atm, I didnt mean for it to be a future when K8L comes out which isnt goign to be a monster, but we will just have to wait and see I guess.
July 8, 2006 4:46:43 AM

Quote:
i dont think INTEL will be laughing at quad core dual socket K8L's

thats said...there are big name games being optimized for 4 cores

whos to see CRYSIS cant be coded to offload AI routines to CORE 3 and 4


and let core 1 and 2 run the engine with no AI routine bog down..

you guys think small.....


Lordpope when you buy AMD's 4x4 system, I for one will be
interested in hearing from you. Did I hear you say it's too expensive
for you. Well! Dah! AMD stating "Performance at any price". Is not
a statement I would be proud of. Unless, it meant, Them spending the
money on R&D or manufacturing facilities.
AMD had a great product. It's still a good product. I follow a lot of
sites that build custom gaming machines. And a lot of them, stop
offering the Intel package. For example, overdrivepc.com.
Now it's Intel's time with conroe.
July 8, 2006 4:58:52 AM

Quote:
They'll be laughing in 2008.

if they can control cache yields on die :wink: and keep yield production high enough to make it pay.(joking)scary thing is intel may have figgured out how to do it,if so AMD's crying will last longer than the energizer bunnies freakish commercials,,LOL I mean a long time.

what does this have anything to do with yields? 4x4 is NOT a new processor. It is just taking 2 of their top of the line processors, and throwing them into a DP system that supports SLI video cards. Nothing here is a new concept, just that AMD is pushing it to be the next best thing.
July 8, 2006 5:06:11 AM

that a damn estimate... now shut up you stupid fanboy!! Lets do the math...
2 AMD FX-62.... $1000 each = $2000...... 1 Intel Conroe.... $1000 FOR ONE = $1000..... 2x the price and only 15% performance increase.... that is the greatest deal i have ever heard of.... OMG :roll:
July 8, 2006 8:07:00 AM

Quote:
2x the price and only 15% performance increase.... that is the greatest deal i have ever heard of.... OMG :roll:


Sounds like the Core 2 6700 versus Core 2 Extreme. :roll:
July 8, 2006 9:16:57 AM

You are happy with posting news that shows how a system delivers 15% higher performance than a system that costs 1/4?
Dude, youre uber-pathetic...
July 8, 2006 9:20:03 AM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)

OMG............ :roll:
What a BS!
A conclusion made without benchmark in Shakira's style. Why don't you PM Shakira?
He will put this into his blog as a fact and a prove(with some numbers increased of course to benefit his love)
July 8, 2006 9:46:56 AM

Does anybody really understands the 4x4 concept ?
I think NOT. This is not about games or price, it’s all about performance.
Is 4x4 better than single socket ? Of course it is, it has better memory bandwidth and good scalability.
The fact that current AMD processors lag behind INTEL’s is irrelevant. The fact remains that 4x4 is a very good concept for AMD because it makes the best of their processors.

Is 4x4 the best way to go for games … probably not.
Is 4x4 the best way to go for workstations ? Hell YES !

Can Intel do the same with it’s current dual core ? … Not successfully and Woodcrest already shows it has a FSB bottleneck.

My conclusion … 4x4 it an excellent concept from AMD and if better processors become available the thing will be the next best thing since slice bread.
July 8, 2006 10:15:52 AM

Quote:
Does anybody really understands the 4x4 concept ?

Yes, but seems you are not the one
Quote:
I think NOT.

no body can stop you, you can think whatever you want.
Quote:
This is not about games

It is about games. Read the topic.
Quote:
or price, it’s all about performance.

sort of the second(perfomance) and the relation of the two: performance/price
Quote:
Is 4x4 better than single socket ?

No it is not better.
Quote:
Of course it is

no, it isn't.
Quote:
, it has better memory bandwidth and good scalability.

Single socket quadcore is better solution than dual socket dualcore. The two socket mainboard is also much more expencive than the single socket. Memory bandwidth is relative factor for performance. For example, sAM2 K8 has much more memory bandwidth than Core2 Duo has, but it is not faster.

Quote:
The fact that current AMD processors lag behind INTEL’s is irrelevant.

No, it is relevant!

Quote:
The fact remains that 4x4 is a very good concept

No, 4x4 is very very bad concept: milking money for not reasonable performance boost on the desktop/workstations. Who needs a good 2P server can buy Woodcrest and who needs a good 4P or 8P can buy s940 or wait for socketF.

Quote:
for AMD because it makes the best of their processors.

It makes the worse of their procesor, the performance/price ratio of the 4x4 is catastrophic compared to any other AMD solution
Quote:
Is 4x4 the best way to go for games … probably not.

I agree, but it is aimed for the gamers, the very rich and very stupid ones.
Quote:
Is 4x4 the best way to go for workstations ? Hell YES !

HELL no.

Quote:
Can Intel do the same with it’s current dual core ? … Not successfully and Woodcrest already shows it has a FSB bottleneck.

They allready done that and wiped the floor with the best 2P Opteron solution.

Quote:
My conclusion … 4x4 it an excellent concept from AMD

Your conclusion != general conclusion.
general conclusion = 4x4 is an extremly expencive concept.
Quote:
and if better processors become available the thing will be the next best thing since slice bread.

If......then....else....
time will tell and we'll have to wait a least a year before we can expect to see any better sAM2 CPU.
July 8, 2006 10:17:20 AM

Quote:
Does anybody really understands the 4x4 concept ?
I think NOT. This is not about games or price, it’s all about performance.
Is 4x4 better than single socket ? Of course it is, it has better memory bandwidth and good scalability.
The fact that current AMD processors lag behind INTEL’s is irrelevant. The fact remains that 4x4 is a very good concept for AMD because it makes the best of their processors.

Is 4x4 the best way to go for games … probably not.
Is 4x4 the best way to go for workstations ? Hell YES !

Can Intel do the same with it’s current dual core ? … Not successfully and Woodcrest already shows it has a FSB bottleneck.

My conclusion … 4x4 it an excellent concept from AMD and if better processors become available the thing will be the next best thing since slice bread.


i dont think so.... woodcrest uses 2-independent 1333Mhz FSB.. that was a total of 21.2GB/s

the HT 2.0 are only 11.2GB/s

more here:TipidPC.com
July 8, 2006 11:41:50 AM

Well … theoretically maybe.
The single FSB is limited in practice at 1,333 with a 1,600 FSB troreticaly.

Quote:
„However, with the antiquated FSB approach, two dies on that chip have to share the bus, imposing a bandwidth bottleneck, not only because of twice the "data hunger" on that FSB, but also slowing it down by at least 20% due to three loads instead of two.
How to solve that? Two ways: One is to bring out BOTH FSBs (one per core) out to the dual-FSB north bridge, and require a new package in the process. It is a quick fix, but needs a new socket! „

The link you provided is based from an Intel official press release.
Nobody can say that the 2-independent 1333Mhz FSB is a good idea or not. I think it’s not.
It’s more of a quick fix than a real solution for the FSB problem.

Also with more cores you can’t just add another independent FSB for each core since it requires a new north brige. And if we talk about multi socket it is clear that each core with it’s own FSB is not going to happen.
Can anyone think about a north brige for a 4 soket/4 cores + independent FSB ? It just doesn’t look good at all. Intel has a problem with it’s north brige and independent FSB don’t scale at all.
July 8, 2006 12:29:12 PM

Quote:
Well … theoretically maybe.
The single FSB is limited in practice at 1,333 with a 1,600 FSB troreticaly.

Quote:
„However, with the antiquated FSB approach, two dies on that chip have to share the bus, imposing a bandwidth bottleneck, not only because of twice the "data hunger" on that FSB, but also slowing it down by at least 20% due to three loads instead of two.
How to solve that? Two ways: One is to bring out BOTH FSBs (one per core) out to the dual-FSB north bridge, and require a new package in the process. It is a quick fix, but needs a new socket! „

The link you provided is based from an Intel official press release.
Nobody can say that the 2-independent 1333Mhz FSB is a good idea or not. I think it’s not.
It’s more of a quick fix than a real solution for the FSB problem.

Also with more cores you can’t just add another independent FSB for each core since it requires a new north brige. And if we talk about multi socket it is clear that each core with it’s own FSB is not going to happen.
Can anyone think about a north brige for a 4 soket/4 cores + independent FSB ? It just doesn’t look good at all. Intel has a problem with it’s north brige and independent FSB don’t scale at all.


Do you even know what you are talking about?

two dies on that chip have to share the bus.....

No they do NOT have to share the bus, that is the whole purpose of independant FSBs. When you go to quad core, then yes, you need a new socket, or you have to start sharing again, but the current dual core processor does have independant FSBs. Do you think AMD can really fit 4 cores onto the current package today? Oh wait, it doesn't matter because we wont know anytime soon. So there is a good chance that quad core will have a new socket anyways. And do you really think it scales that much different than HTT? If FSB has a new FSB per CPU, that means that dual core will be 21.2GB/s and quad core will be 42.4GB/s, while AMD will still be running quad core on 11.2GB/s. Will it take more pins? YES! But will the bandwidth scale better? Absolutely!
July 8, 2006 12:37:13 PM

Quote:
The fact that current AMD processors lag behind INTEL’s is irrelevant.

Acctually current AMD beats intel. (Core 2 duo [stupid name] isn't out yet.)
July 8, 2006 1:22:40 PM

Dude, you are posting totaly OT BS!
Quote:
Well … theoretically maybe.
The single FSB is limited in practice at 1,333 with a 1,600 FSB troreticaly.

Who cares for single/dual socket solutions?
It provides enough bandwidth and it is not bottlenecking the system.
Quote:
Nobody can say that the 2-independent 1333Mhz FSB is a good idea or not. I think it’s not.

If you are nobody, yes you can say whatever.
It is not an idea, it is realised solution, there is huge difference. It is sucessfull solution for 2P Intel platforms, not the best.
Quote:
It’s more of a quick fix than a real solution for the FSB problem.

Yes it is a quick fix until Intel bring CSI. And since there are no problems with it WHO CARES?


Quote:
Also with more cores you can’t just add another independent FSB for each core since it requires a new north brige.

No. Core2 is glueless dualcore with shared L2, no FSB and no northbridge involved.

Quote:
And if we talk about multi socket
We were not talking about your OT thoughts, we were talking about the topic: "AMD 4x4 = 15 % faster than CONROE in gaming".
Quote:
it is clear that each core with it’s own FSB is not going to happen.

I don't understand what you want to say.
Quote:
Can anyone think about a north brige for a 4 soket/4 cores + independent FSB ?

Again, I don't understand what you want to say.

Quote:
It just doesn’t look good at all. Intel has a problem with it’s north brige and independent FSB don’t scale at all.

Intel have no problems with the norhbridge. The independend FSB works and scales good, the result is beaten Opteron which has better CPU2CPU comunication.
Intel have no good solution for connecting more than 2P, AMD have no good solution for connecting more than 4P. AMD have better 4P servers, Intel have better 2P servers.

Now can you please try to be on topic or STFU becouse no one cares about your OT dillemas and lack of knowledge.
July 8, 2006 1:39:02 PM

Quote:
http://xpentor.com/mamboserver/content/view/76/1/

according to this site...



(Disclaimer: the statements made by xpentor.com are the opinions of said site. It is neither implied or stated by me that i agree in whole or in part with anything in said site. Furthermore , predictions of unreleased hardware performance and price modeling with no benchmark data or channel price data from a reputable site is prohibited on general principle in this forum)


"The tough part about predictions is that they're always about the future."

Yogi Berri
July 8, 2006 1:45:37 PM

Quote:

And do you really think it scales that much different than HTT? If FSB has a new FSB per CPU, that means that dual core will be 21.2GB/s and quad core will be 42.4GB/s, while AMD will still be running quad core on 11.2GB/s. Will it take more pins? YES! But will the bandwidth scale better? Absolutely!


You don’t understand, let me explain:
dual core +independent FSB => north bridge 21.2GB/s
quad core +independent FSB=> north bridge 42.4GB/s

the north bridge -> memory bandwidth is limited, to satisfy the need for memory bandwidth Intel introduced quad channel and FB-DIMM. So in quad core setup the memory bandwidth/core actually decrease.

Quote:
„The 5000 chipset runs DDR2-533 at FSB1066 speed and DDR2-667 at FSB1333 and performs similar to what you may be used from desktop solutions. Although the quad channel memory controller introduces higher bandwidth, it does not scale that well. Expect gross bandwidth increases in the area of 50% when compared to a dual channel controller, as latencies grow as well.”
5000P/X and 5000V MCH (northbridge)
• FC-BGA package with 1432 balls
• Supports two processors with dual independent busses (DIB)
• FSB1066 (266 MHz, 16.6 GB/s) and FSB1333 (333 MHz, 20.8 GB/s)
• Quad channel DDR2 memory controller for FB-DIMMs (Fully Buffered DIMMs), DDR2-533 or DDR2-667

PS: gOJDO Sorry about OT but I wanted to respond to the fact that Woodcrest don’t has a bottleneck in quad core or dual socket/dual core setup. And insulting me won’t make you look smarter.
!