• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

When is AMD going to release real 8 core chips?

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Edrick said:
Is it me or was the OP just baiting with a thread like this?
Click to expand...

Oh yeah, pretty classic troll that got plenty to play along 🙂

Still some interesting points discussed here though, so reading the thread was worthwhile despite the possible original intent.
 
If the OP counts a pair of int cores (as x86 cores were up to 386) as only one core, he should look at Interlagos with 16 cores, if he doesn't count the Magny Cours.

BTW if some cores share L2 or L3, would they count as a single core? ;-)
 
Dresdenboy said:
If the OP counts a pair of int cores (as x86 cores were up to 386) as only one core, he should look at Interlagos with 16 cores, if he doesn't count the Magny Cours.

BTW if some cores share L2 or L3, would they count as a single core? ;-)
Click to expand...
Or worse: what if the GPU shared the L3 with 4 hyperthreaded cores? There's no way that they would count as real cores!

/s
 
Forget 8-core, when are the mainstream cpu's going to get 6 core? I've heard the mainstream Haswell are still 4-core only, which makes sense I guess. I hope the die-shrink has 6-cores though.
 
StinkyPinky said:
Forget 8-core, when are the mainstream cpu's going to get 6 core? I've heard the mainstream Haswell are still 4-core only, which makes sense I guess. I hope the die-shrink has 6-cores though.
Click to expand...

When the manufacturers think there's enough profit to be made to justify the investment.
 
StinkyPinky said:
Forget 8-core, when are the mainstream cpu's going to get 6 core? I've heard the mainstream Haswell are still 4-core only, which makes sense I guess. I hope the die-shrink has 6-cores though.
Click to expand...
/me looks at sig

/me looks back at post

I believe we have tried that already
 
performance wise in non-gaming applications Bulldozer consistently beats a Phenom II X6 1100T by the amount of the clockspeed difference or more, unless the app isn't well threaded. Thus, i feel safe calling the BD a "6-core equivalent"; obviously none of it's physical functions are in 6, but it's got the IPC of a 6 core Phenom II at the minimum in well threaded apps, and in a couple well threaded apps, blows right by the 6 core phenom II by the clockspeed difference + 10-32%. Even in Sandra GFLOPS it beats the Phenom II X6 by the clockspeed difference, meaning it's theoretical FP performance is that of a 6-core Phenom II. As expected, in Sandra GIPS, it blows by the Phenom II X6 by 32% + the clockspeed difference. And some results that could be explained by 4 fpus can also be questioned: In After Effects it's 14% faster than the 4 core Phenom II X4 in IPC but slower than the X6, yet in Blender which also is a FPU-centric app, it blows by both, beating the X6 by 30% in IPC.
 
Last edited:
Anarchist420 said:
Intel will probably be the first to release a true 8 core CPU although there will be no 8 core Haswell if I'm not mistaken.
Click to expand...

What?!

AMD released a 12 core Magny-Cours (12 actual cores) 2 years ago. How is Intel going to be first?

Unless you are meaning 12 cores on one die. But then thats kind of splitting hairs.
 
Last edited:
StinkyPinky said:
Forget 8-core, when are the mainstream cpu's going to get 6 core? I've heard the mainstream Haswell are still 4-core only, which makes sense I guess. I hope the die-shrink has 6-cores though.
Click to expand...
When Intel has fully integrated everything onto an SoC. That's why you're not seeing more than 4 cores... integration is the focus currently, and for good reason.
 
Phynaz said:
When the manufacturers think there's enough profit to be made to justify the investment.
Click to expand...

There is not much investment to be made. They already make the hex cores. They (Intel) just has to position them in the mainstream, with price proportionate to the quads. Since the 3570 is 280.00, they should be able to make a hex core for around 450.00 instead of the 600.00 for the E chips
 
Idontcare said:
The issue with bulldozer is not core-count, or power consumption, or IPC...it is all of the above when combined into a singular product.
Click to expand...

Very well said. That sums it up.
Bulldozer has the core count, hence the strong performance in multi core intensive apps but when IPC on a single core comes into play, Bulldozer sucks (just as P4 did)
 
Last edited:
Sorry to burst your bubble, but AMD will release a "fake" 10 or 12 core desktop chip before they release a "real" 8 core desktop chip, which is what they are calling AMD FX.
 
frozentundra123456 said:
They (Intel) just has to position them in the mainstream, with price proportionate to the quads. Since the 3570 is 280.00, they should be able to make a hex core for around 450.00 instead of the 600.00 for the E chips
Click to expand...

You always pay a premium for top of the line parts. It is rarely linear like you suggest. Marketing 101.
 
BenchPress said:
They can still compete against Intel, if they want to. I hope they grow a pair.
Click to expand...

No they can't. And there's no reason to compete in CPUs, x86 efforts are increasingly futile anymore with the rise of ARM. I'd focus on bang/buck server chips, low power and APU integration, those things will pay off down the road. Having a power hungry, insanely powerful CPU that doesn't get utilized by 95% of the population is stupid. You can always toss in moar coars and go parallel for content creators.
 
LeftSide said:
I don't get why people want more cores. I want more MHz and IPC.
Click to expand...
It's prohibitively expensive to aggressively increase IPC or MHz. And a lot of people think adding more cores is the best/only alternative.

It's not though. A more efficient way to increase parallelism right now is by using wide vector instructions with gather support. That's what AVX2 is for. It will make many scalar code loops run up to eight times faster.

CPUs should balance ILP, TLP and DLP for best performance across a wide range of applications. ILP is exhausted and DLP is cheaper than TLP for now. But eventually the number of cores will go up again. And fortunately Haswell is preparing for that as well with TSX.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top