Overclock.net banner
1 - 20 of 30 Posts

·
Registered
Joined
·
1,560 Posts
Discussion Starter · #1 ·
Quote:


The worldwide attempt to increase the speed of supercomputers merely by increasing the number of processor cores on individual chips unexpectedly worsens performance for many complex applications, Sandia simulations have found.


Read the rest here.


From the article it looks like anything over 8 cores on a single chip hurts performance.
 

·
QA Engineer
Joined
·
3,016 Posts
Irony?
 

·
Registered
Joined
·
3,746 Posts
Good thing I stopped at 4.
 

·
Tator Tot Enthusiast
Joined
·
3,177 Posts
Quote:


The problem is the lack of memory bandwidth as well as contention between processors over the memory bus available to each processor. (The memory bus is the set of wires used to carry memory addresses and data to and from the system RAM.)

Simple problem to fix IMO.
 

·
PC Gamer
Joined
·
1,777 Posts
Quote:


Originally Posted by Tricky
View Post

Simple problem to fix IMO.

your face is a simple problem to fix oh wait .......

(just kidding mate)
 

·
Premium Member
Joined
·
7,187 Posts
Quote:


Originally Posted by Drizzt5
View Post

Good thing I stopped at 4.


Yup, I'm sticking with my Q6600 until it's completely obsolete.
 

·
Banned
Joined
·
8,805 Posts
I thought this was discussed months ago and it was because we are having an increase in cpu's, but we are not increasing the speed in which data can get to them.
 

·
Premium Member
Joined
·
16,928 Posts
anyways, no surprises there. just like how they realised that insanely high clockspeeds weren't the way to go during the P4 era and then started adding chips. they'll figure out something. intel's Tick Tock strategy comes to mind
 

·
Premium Member
Joined
·
4,513 Posts
We don't even need super computing, we just use super computers to simulate dumb things to see what this or that would do etc.
 

·
Premium Member
Joined
·
65,162 Posts
Quote:


Originally Posted by joemaniaci
View Post

I thought this was discussed months ago and it was because we are having an increase in cpu's, but we are not increasing the speed in which data can get to them.

Yup... it's a repost from Jan 15th, 2009.
 

·
Premium Member
Joined
·
15,202 Posts
Quote:

Originally Posted by Tricky View Post
Simple problem to fix IMO.
Not as much as you'd think. It would require more complex PCB and larger interfaces to the memory, as well as faster memory itself.
 

·
Registered
Joined
·
1,832 Posts
well, gaming and graphic intense applications will soon be GPU reliant and not on the CPU so supermulti-CPU would really be useless unless its a super intense app that requires alot of calculations.
 

·
Registered
Joined
·
1,101 Posts
Quote:

Originally Posted by squall325 View Post
well, gaming and graphic intense applications will soon be GPU reliant and not on the CPU so supermulti-CPU would really be useless unless its a super intense app that requires alot of calculations.

I think you have that bachwards.
 

·
Premium Member
Joined
·
4,513 Posts
Quote:


Originally Posted by Sasquatch in Space
View Post

I think you have that bachwards.

I think you both have it backwards.

It's pretty balanced if you ask me.

CPU does physics and geometry calculations and other crap, gpu does pixel calculations and video displaying and other crap as well.
 
1 - 20 of 30 Posts
Top