New Nehalem socket is... large.

[PCXL-Fan]

Can one quantify the gains AMD saw when going from the amd64 to the amd x2 with the IMC?

Originally Posted by skoreanime

One of my friends was mailed an engineering sample Pentium 4 chip once (once they were starting to roll out) . Stuff like that can happen

Thats pretty cool.

I have a korean canadian friend on a forum who himself has a friend who got an engineering sample Pentium 4 chip.

[runawayprisoner]

Pshh... you can get Engineering Samples for free (once you request it), you know.

[cottonvibes]

Originally Posted by runawayprisoner

Hint: be a seller.

but even if you're a seller, to get a chip this early?
are the chips even finalized?

Originally Posted by Borisz

IMC is the one thing AMD has over Intel right now. Intel is still pounding them in the back, even though the core 2 is bottlenecked by the old FSB standard. Remove that bottleneck and gangraping will commence. AMD will be the butch side, unless they can make the Phenom suck less phenomenally.

i don't think Intel's chips are being bottlenecked because they don't have IMCs.

i think not having IMCs have helped them overclock better and gain more performance, because its 1 thing less they have to do.
and the less they have to do, the less transistors the chip has to have, and the less transistors, the easier it is to overclock.

[runawayprisoner]

Originally Posted by cottonvibes

but even if you're a seller, to get a chip this early?
are the chips even finalized?

Uhm... ever heard of B1-stepping Q9450? Or any Bx stepping 45nm processor to begin with? Hehe... yeah. The difference between an ES and a retail processor might be stepping... or might be something else... so you can't really say much about the performance of an ES compared to that of a retail one. However, there are cases when the stepping on an ES is superior to that of a retail processor... and...

Engineering Samples are meant to be... experimental CPUs. People receive them to see how they perform, and... they report benchmarks and stuffs to Intel. Most of these steps are just to find the best stepping, or manufacturing steps for the retail processor.

[Borisz]

Originally Posted by cottonvibes

i don't think Intel's chips are being bottlenecked because they don't have IMCs.

Intel themselves criticized the FSB for being a dinosaur antique component thats only holding them back.

i think not having IMCs have helped them overclock better and gain more performance, because its 1 thing less they have to do.
and the less they have to do, the less transistors the chip has to have, and the less transistors, the easier it is to overclock.

No, Intel just designs better chips. AMD never overclocked as good as Intel did, with or without IMC.

[Phil]

DDR3 only......um the high latencey of DDR# is going to hold this chip back in the beginning...just like DDR2 did when it first cae out.......im waiting for another 2 years

[tuanming]

It's large because its has integrated technologies such as memory controller, Quick Path Interconnect (QPI), etc. All of these technologies need extra room/space to fit inside a cpu.

[John Aiton]

Originally Posted by Phil

DDR3 only......um the high latencey of DDR# is going to hold this chip back in the beginning...just like DDR2 did when it first cae out.......im waiting for another 2 years

the current yorkfield cpus are fast enough for your average pc user,
I can see them needing more in the next 10 years

nehalem is more suited to servers and crunchers with al the cores

tbh, do any of us gamers actualy need nehalem anytime before ray-tracing?

[SCHUMI-4-EVA]

Well that depends on how "round" the next "wheel" of a game will be...if it ends up being another cube like crysis then yes...gamers will need it...if however they actually round the damn thing off propperly from now on then no.

[Cheesus]

We need faster Graphics cards before we can even begin to consider such CPUs worth it.

[SCHUMI-4-EVA]

Unless we get another Oblivion or NWN2....

[PCXL-Fan]

Quake 3 Arena had rounding algorithms. Its not that hard to do.

[SCHUMI-4-EVA]

...you know what I was trying to say....

[Thanakil]

Originally Posted by SCHUMI-4-EVA

Unless we get another Oblivion or NWN2....

Were those games CPU-intensive? Cause I sure as hell don't remember them being that...

[cottonvibes]

Originally Posted by John Aiton

tbh, do any of us gamers actualy need nehalem anytime before ray-tracing?

i would buy it so i can play PCSX2 with almost all games at full speed.
i say "almost all games" because Refraction said that there might be games that even a 5Ghz dual-core wouldn't run at fullspeed.

[Phil]

In 4 years, 5ghz will be comomplace......like Cheesus said, we have to wait till video cards catch up now....but mostly coding...I mean we barely have code optimized for 4 cores...just think about 8 and 16..as mudlord would say, optimization is the key

[cottonvibes]

coding won't get more optimized for multi-cores.
every application is different, and requires a different strategy to divide the workload between the different codes.

saying "we need better optimized coding" is basically saying "we need a world full of Genius and Skillful Coders".

in 5 years, we'll have chips so fast, we don't need multi-threading for CPU's.

for GPU's its different though, since lots of Graphical algorithms can be easily preformed in parallel.
It makes more sense to focus multi-cores in GPUs than in CPUs.
Game coders don't even handle most the work done by the GPU, thats done by the GPU's drivers and Hardware Accelaration Libraries.



but i think probably in 5-15 years, we'll start seeing a muli-functional CPU, that does all the work and eliminates the Graphics Card.
Then everything would just be done by a super CPU.

[mudlord]

saying "we need better optimized coding" is basically saying "we need a world full of Genius and Skillful Coders".

No, we don't. We just need coders that know how to write effective code to target what we have. Of course, programming skill plays a huge role, but learning more effective methods is the key. Plus, optimization should never be a last priority.

in 5 years, we'll have chips so fast, we don't need multi-threading for CPU's.

Yet again we use the MS/Vista argument - "Why optimize when stuff is fast enough"....
Sheesh.

[cottonvibes]

multi-cored cpus have already been out for some time, i doubt coders are going to be writing more effective code in the future.
its just too much of a pain, especially in games where the GPU is usually the limiter.

maybe if coders start learning more effective methods like you said, it will help.
but when technology is evolving so fast, its easier to just wait till faster PCs comeout.

coders already have enough problems with their programs, no need to make it harder when in a couple years, all their optimizations would be pointless...



but what i'm saying is mostly for games.
when it comes to an OS like vista, it should be as optimized as much as possible.
mainly because its an OS, and kindoff the brain of the computer.
a game isn't as important as an OS, so a high emphasis on optimization in a game isn't as important.

[StriderVM]

Then why is the PS3 getting traction nowadays? Performance wise it's like a supah computer! But then, how about games....?

Anyway, back to the topic at hand...... For me, in reality, it's currently unlikely to see processors in the 9Ghz range or something. It's because before, getting to lower processes helps more than hurts them. Now it's starting to do the opposite. So this means there is a limit on the Ghz thing.

More likely to happen are dual to quad and more cores. And also, besides the actual programming for more than one core, I think there will be a time when the OS might be able to distribute the workload completely between cores/processors.