Wait for cheaper quad core to upgrade?

I have some money to upgrade soon, but I know AMD is releasing their quadcore in a few months(?). Do you think it would be worth waiting? How expensive do you think AMD’s lowest quad core CPUs will be? All major gains I’ve seen of multiple core benchmarks so far are mostly compiling, or rendering, etc stuff I probably won’t be doing often. Are more cores(four instead of two) going to be a real benefit for software/games in a year or three?

Short answer: No.

Long answer: No.

Get the fastest dual-core you can buy.

Unless you’re playing Supreme Commander ;-)

Benchmarks pls?

You must ask yourself: what software takes advantage of quadcore? Yeah…

Isn’t quadcore also hot as hell/really power-hungry currently?


But it also enlarges your e-penis by 4!!

Also, if you’re going quad core why not also go with quad SLI!!!1111oneone

Compilers! :-)

Until more multi-thread apps are common, a quad core is a waste of money right now.

I have been waiting for a price drop in Quad cores. Right now, Ill agree they are not worth the money.

I have heard that Intel plans on dropping thier price sometime in Q2 2007, to 530 USD (down from 870 USD). If and when this happens, I think it would be a good buy.

I’m actually working on that.

Supreme Commander has two primary threads, one for rendering one for the “simulation” (pathfinding, physics, etc.) Each thread can consume the resources of an entire core.

There are other threads for audio, driver, etc. Those get moved to additional cores if possible.

Intel worked with Gaspowered on the threading model. The built-in performance test was designed based on a spec I gave Gaspowered.

I haven’t followed it too close, but Intel’s Bearlake 1333mhz FSB is supposed to be due out mid year. So if you’re going to wait, it might be a good idea to pick it up then.


Yeah, there was a thread about it on the GPG.net forums. I am not sure they are really optimized as much as they say. Really, 1 thread for rendering isn’t really N-Core optimized code. I suppose a good algorithm would count the number of CPU’s and create that many threads for each heavy-weight process.

It would be interesting to rethink standard single-thread algorithms in terms of N-Threads.

Lets take a simple A* path finding algorithm. One way to optimize it is simply to divide the number of objects you need to path find for by the number of cores you have and make that many threads with smaller ques of paths to find.

However, that isn’t in the spirit of the question. How can you take the single algorithm and make it mutli-threaded so that one object gets the benefit of N cores?

I wouldn’t even know where to begin to think about that kind of stuff.


(Kevin is referring to the same chipset, I believe.)

To quote my favorite movie of all time, Firefox: To code multi-threaded, you must think multi-threaded. Or something.

I’m actually working on that.

Cool, I’m looking forward to reading it! I’m sure it’ll be another exemplary Case article, as I’ve come to expect.

Also, +1 Firefox reference. Who here played the arcade version?

make -j16

There’s the quadcore benefit :)

One of the few times the arcade game was actually better than the movie.

Like I was sayin’…

The most striking thing about these results is, relatively speaking, how little separation there is between the fastest and slowest systems. One of the reasons for that is clear. The Task Manager plot tells us (and the results confirm) that, despite rumors the game would put more than two CPU cores to good use, Supreme Commander doesn’t use four cores very effectively.

Same advice applies. Unless you do a lot of highly specialized video rendering or raytracing… get the fastest dual core system you can buy, and forget quad even exists for the next 2 years at least. By then it’ll be cheaper anyway.