• Welcome to SC4 Devotion Forum Archives.

Planned Ultra computer

Started by Mulege, March 24, 2009, 09:55:14 PM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

Mulege

Right now I a working on a cool computer Imagine this 16GB of Ram 16ghz Octacore Prossesor A DVDRW CDRW Combodrive and Bluray Drive Nividia FX Gforce 9600 Graffics card

I sure can't wait till they come out with the Octacore Procssesor in the next two years!

choco

just more resources for vista or 7 to hog.....

i'd much rather see some advancement in the time constants of the chips rather than just to keep adding cores.  with exception to non-x86 based chips (like the cell), this doesn't really do much other than provide new products built on the same old technology.  this way, you feel like you getting something new.

adding additional cores is analogous to adding more lanes to a highway.....you add a bit more capacity to teh cpu, but you maintain the same speed limit.

until chip makers get ahold of some reliable low-k dielectrics and CMP technology catches up, im afraid x86 architecture is at its feasible speed limit.

Meastro444

Go for it, but that 9600 will bottleneck the whole system, that's for sure.
Friend of the Certified Drama Queen :)

Jonathan

This is semi relevant,
If I have a quad core and it's at 3Ghz
Does that mean it's 3*4 = 12GHz?

Another semi-relevant question,
Is the GeForce 260 better than the 9 series?

Not sure what you talking about Choco(I don't understand the actual insides of a CPU(something I want to though)) but isn't 64-bit processors better?

Jonathan

nerdly_dood

A 3GHz processor isn't actually more like 12 GHz - it's 3GHz no matter how many cores it has. HOWEVER, it would function as fast as a single-core 12 GHz processor.

At least that's my understanding.

I know next to nothing about nVidia video cards, though - I just know about ATI and their number-rating system.

Yes, 64-bit processors are better because they handle twice as much data per clock cycle in each core as a 32-bit one.
My days here are numbered. It's been great and I've had a lot of fun, but I've moved on to bigger and better things.
—   EGO  VOBIS  VADELICO   —
Glory be unto the modder and unto the fun and unto the city game!

choco

well.....im not a scientist or circuit designer, but i am a semiconductor engineer.  my company builds tools that are used in the manufacture of computer chips.  i've spent about 2 years of my life in a fab (literally), and have had considerable time to discuss some of these things with process engineers.

essentially, there's too much stock put into these numbers like clock speed.  the real speed of a chip is defined as 1/(RC) where R is the resistance of the copper lines and C is the capacitance of the dielectric.  until these values can be lowered, the fsb of a chip will not increase from where its at now.  the XGHz is just a marketing tool.

a quad core 3GHz processor is not the equivalent of a single core 12GHz processor.....it would be equivalent to 4 3Ghz single core processors.  however, these 4 3Ghz cores individually are not as fast as a 3Ghz single core either, since core have to share resouces like memory and some cache.  this means the data has to be switched and managed, which added miles of extra copper lines which in turn slow the processor.  in otherwords, muticore processors use a larger footprint of silicon, which makes them slower.

the benefit we see to multicore is parallel processing......having 2 or more "pipelines" to process the data.  so you can process more data with a multicore than you can with a single core in any given amount of time, but the speed that the data is processed is the same. 

couple this withthe fact that vista and 7 are using more and more resources for all the bloatware included, are you really getting a faster computer.....or are you just perceiving a faster computer?

now, sony developed the Cell on an entirely new architecture....not x86 based.  its a fascinating chip really.....the bench unit hit 5.64Ghz on air cooling.  despite its own manufacturing challenges, it can lay to waste any other chip on the market with regards to performance.  a single cell will outperform the fastest Intel chip (or AMD for that matter) as well as the fastest GPU's currently available.......at the same time.  not only does it have more cores (most have 8 active cores), but it doesn't use cache....ultimately removing the slowest logic in an x86 chip.






Dino007

Quote from: Warrior on March 26, 2009, 02:20:16 PM


Another semi-relevant question,
Is the GeForce 260 better than the 9 series?

Yes, it is, but for such supercomputer I would take gtx 295.
TIGR

Kitsune

I know octo xeon's come out in q1 2010, that could make one heck of a skulltrail system (2xoctocores, 6x4gb for 24gb of ram, run crossfire or sli depending on if the 5000 or 300 series is the king). Have a ssd or a cheetah as a main drive, heh, very nice system.
~ NAM Team Member

cameron1991

Xeon processors are geared more towards workstations and servers aren't they?
How would a Xeon processor stack up compared to say a roughly equivalent I7 processor?

Kitsune

I believe better - theres a reason why skulltrail systems are very popular with gamers and hackers alike.
~ NAM Team Member

frdrcklim

#10
Quote from: Mulege on March 24, 2009, 09:55:14 PM
Right now I a working on a cool computer Imagine this 16GB of Ram 16ghz Octacore Prossesor A DVDRW CDRW Combodrive and Bluray Drive Nividia FX Gforce 9600 Graffics card

I sure can't wait till they come out with the Octacore Procssesor in the next two years!

When it comes to imagine computer hardware, I'd imagine something higher than a 9600GT if you want to play those really hard core games or buy two 9600GTs on SLI.

Quote from: choco on March 26, 2009, 12:27:49 PM
just more resources for vista or 7 to hog.....

i'd much rather see some advancement in the time constants of the chips rather than just to keep adding cores.  with exception to non-x86 based chips (like the cell), this doesn't really do much other than provide new products built on the same old technology.  this way, you feel like you getting something new.

adding additional cores is analogous to adding more lanes to a highway.....you add a bit more capacity to teh cpu, but you maintain the same speed limit.

until chip makers get ahold of some reliable low-k dielectrics and CMP technology catches up, im afraid x86 architecture is at its feasible speed limit.

Ever heard of the 1 Terahertz processor being made by MIT? I'm sure it's gonna lighten your day if advancements in technology is what you want ;D.
300... 200... 100... 50... 40... 30... 20... 10

Yep, I still got it.

Andreas

It's a waste of money in the first place.  :P  No typical application (or game) will be able to use the power of such a system effectively; those workstations are designed for massive parallel tasks, such as rendering 3D animations, calculating scientific data, or running large database systems.
Andreas

XiahouDun

Quote from: frdrcklim on March 28, 2009, 01:34:14 AM
Ever heard of the 1 Terahertz processor being made by MIT? I'm sure it's gonna lighten your day if advancements in technology is what you want ;D.
Still not strong enough to play Crysis? :D

I'd need to see the specs for the RAM being used, as well as the OS, to tell you where the weakpoint would be. From this info given though, the video card or the case will be holding the system back. Spending $15 more can get you a 9800GT which will have a notable performance increase, or just go for one of the new GTX video cards that are as long as a runway.
Current project: Movies 14

You may have meant to search for Houdini. (result of searching for XiahouDun on SC4D)

JoeST

The limitations in Crysis are probably due to the coding of the game being inefficient...

Joe
Copperminds and Cuddleswarms

frdrcklim

Quote from: XiahouDun on March 28, 2009, 02:26:02 AM
Still not strong enough to play Crysis? :D

Try that in quadcore configuration :D.
300... 200... 100... 50... 40... 30... 20... 10

Yep, I still got it.

choco

Quote from: frdrcklim on March 28, 2009, 01:34:14 AM
Ever heard of the 1 Terahertz processor being made by MIT? I'm sure it's gonna lighten your day if advancements in technology is what you want ;D.

a bit....but as i said...its not x86 based, and even different materials than silicon......

i didn't say faster cycles wasn't possible....only that its not possible with and x86 based chip.   ;)  there's optic chips as well that run much, much faster, but commercial viability is a problem (same as with the MIT chip).

cammo2003

Well, those nonstandard materials will eventually become standard. We're starting to hit the limits of silicon as well as current manufacturing processes (32nm processors will have to be manufactured within a submerged chamber - 45nm is the limit of manufacture in air).

Quantum computing's where the future's at though.

Oh, and I'd imagine you'd want SSD drives for your supercomputer. Current hard drives limit modern PC performance quite significantly.