Sign in with
Sign up | Sign in
Your question

Multi-core/thread article

Last response: in CPUs
Share
a b à CPUs
March 14, 2005 12:42:59 PM

Very interesting <A HREF="http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=237..." target="_new">Anandtech Multi-Core article</A>.

__________________________________________________
<font color=red>You're a boil on the arse of progress - don't make me squeeze you!</font color=red>
March 15, 2005 12:48:58 AM

Some great information. Too bad it gets a little hard to follow at times. The guy from Unreal helped a lot.
a b à CPUs
March 15, 2005 2:13:11 AM

LOL! I started to go cross-eyed a couple of times, but I thought it was a good read, too.

__________________________________________________
<font color=red>You're a boil on the arse of progress - don't make me squeeze you!</font color=red>
Related resources
March 15, 2005 2:42:50 AM

I will jump on the dualcore bandwaggon just to be able to encode a dvd and play my favorate game. That realy compells me alone. Maybe another system for my game servers. How would the pricing compare to a comporable operton system?

"If youre paddling upstream in a canoe and a
wheel falls off, how many pancakes fit in a doghouse? None! Icecream doesn't have bones!!!"

"Battling Gimps and Dimbulbs HERE at THGC"
March 15, 2005 2:46:51 AM

Expect a dual core 3200+ to hit the streets at about the price of a 3800. The dual core @1.8 ghz should be about the price of a 3500+ The 2.2 will probably be top speed, and price.
March 15, 2005 4:11:22 AM

"You can expect games to take advantage of multi-core pretty thoroughly in late 2006 as games and engines also targeting next-generation consoles start making their way onto the PC.

Writing multithreaded software is very hard; it's about as unnatural to support multithreading in C++ as it was to write object-oriented software in assembly language. The whole industry is starting to do it now, but it's pretty clear that a new programming model is needed if we're going to scale to ever more parallel architectures. I have been doing a lot of R&D along these lines, but it's going slowly."

<pre><font color=red>°¤o,¸¸¸,o¤°`°¤o \\// o¤°`°¤o,¸¸¸,o¤°
And the sign says "You got to have a membership card to get inside" Huh
So I got me a pen and paper And I made up my own little sign</pre><p></font color=red>
Anonymous
a b à CPUs
March 15, 2005 6:57:05 AM

So, somewheres around $400 US? Wonder if it will be worth it...

<font color=green> Woohoo!! I am officially an <b> Addict </b>!! </font color=green>
<i> <font color=red> One new Firefox fan </font color=red> </i>
March 15, 2005 12:15:34 PM

Bah use a Intel compiler.

-Jeremy Dach
a b à CPUs
March 15, 2005 12:43:14 PM

Wraith: I hesitate buying $200+ processors. Unless there are significant improvements that I <b>need</b>, then I won't upgrade just to upgrade. It may be in a future system somewhere down the line, but probably not my next purchase.

Rich: I think that quote is probably one of the most significant statements in the entire article. I took away that dual-core programming will require a paradigm shift for programmers and significant innovation to make the process feasible/easier.


__________________________________________________
<font color=red>You're a boil on the arse of progress - don't make me squeeze you!</font color=red>
March 15, 2005 4:29:12 PM

Quote:

Bah use a Intel compiler.

-Jeremy Dach


AnandTech: Did you make use of auto-parallelisation compiler technology (like the auto parallelisation found in Intel C++ compiler) to make the engine multithreaded?

Tim Sweeney: Auto-parallelization of C++ code is not a serious notion. This falls in the same category as the Intel compiler's strip-mining optimizations and other such tricks, which are designed to speed up one particular loop in one particular SpecFP benchmark. These techniques applied to C/C++ programs are completely infeasible on the scale of real applications.
March 15, 2005 11:08:23 PM

Bah!

-Jeremy Dach
March 15, 2005 11:11:27 PM

More likely, the low end will start around $250, with the mid ver about $450, and the top dog around $750.
Anonymous
a b à CPUs
March 16, 2005 2:11:11 AM

Well I won't be affording one of those for awhile :smile:

<font color=green> Woohoo!! I am officially an <b> Addict </b>!! </font color=green>
<i> <font color=red> One new Firefox fan </font color=red> </i>
March 16, 2005 4:29:03 AM

Not a big deal really. Even with Sweeney on the job, it's going to take a whole lot of coding to get a dual core to overcome the difference in actual clockspeed. I'm thinking they will be of some value in about a year. No point in jumping on the band wagon too soon.
a b à CPUs
March 16, 2005 9:54:10 AM

Especially considering those who jump on early get to pay the R&D costs for the companies...

__________________________________________________
<font color=red>You're a boil on the arse of progress - don't make me squeeze you!</font color=red>
March 16, 2005 5:57:01 PM

Quote:
Writing multithreaded software is very hard; it's about as unnatural to support multithreading in C++ as it was to write object-oriented software in assembly language. The whole industry is starting to do it now, but it's pretty clear that a new programming model is needed if we're going to scale to ever more parallel architectures.

He's not kidding about this. Actually, IMHO he's very much understating the complexity involved. Besides the pure development issues, multithreading also really hinders debugging. Also timing bugs will be very difficult to catch before beta testing as you can only provoke those that appear in your own hardware, so many slip through the cracks until other hardware is in use.

So the more programs use multithreading, the more bugs there are likely to be in initial product launches. And that translates into a lot more after-market expense in fixing those bugs. So it's not just a 'writing' issue. It's a serious maintenance and customer service issue as well, and is going to require an awful lot of companies to significantly change the way that they do business to not only do much more pre-release testing, but also to cover the inevitable after-market costs.

And in fact, consoles are going to rip a great disparity here. Having an unchanging hardware platform of a console is going to hide even more bugs than before now (thanks to the additions of multithreading issues) that would be provoked by differences in hardware. PCs on the other hand, having generally no standardized platform, are going to see each and every one of these bugs pop up. So if multithreading catches on, it's going to make porting titles from consoles to PCs a hell of a lot more involved.

And that's still just the game industry. Try writing multithreaded Word, Firefox, or even an mpeg compression utility in any way useful. :\ Even if you can manage to find something to run in a seperate thread, you'll still just end up losing more performance from timing issues than you'll gain from splitting the workload onto more hardware.

And then there's the issue of the <i>loss</i> of performance that single-cored single-CPU users are going to see if multithreading somehow catches on.

<pre>Antec Sonata 2x120mm
P4C 2.6
Asus P4P800Dlx
2x512MB CorsairXMS3200C2
Leadtek A6600GT TDH
RAID1 2xHitachi 60GB
BENQ 16X DVD+/-RW
Altec Lansing 251
NEC FE990 19"CRT</pre><p>
March 16, 2005 5:58:22 PM

Quote:
Especially considering those who jump on early get to pay the R&D costs for the companies...

**ROFL** Those would be the enthusiast gamers running dual-CPU boxes. ;) 

<pre>Antec Sonata 2x120mm
P4C 2.6
Asus P4P800Dlx
2x512MB CorsairXMS3200C2
Leadtek A6600GT TDH
RAID1 2xHitachi 60GB
BENQ 16X DVD+/-RW
Altec Lansing 251
NEC FE990 19"CRT</pre><p>
!