HASWELL-E 8Core/16T How much faster will it be, because of DDR4 rather than DDR3 ?

[lamedude]

Ignoring the GPU issue... has there ever been a review posted, anywhere, showing a significant difference in performance between two machines based on slight differences in RAM speed or technology?

P4 was slower with SDR and a bit faster with dual channel DDR(875) or RDRAM(850) over single channel DDR (845).

 

[Charles Kozierok]

P4 was slower with SDR and a bit faster with dual channel DDR(875) or RDRAM(850) over single channel DDR (845).

Interesting. My question was rhetorical and I didn't expect an actual answer.

I guess that shows that it's possible, but then again RDRAM versus SDR is a far bigger difference in technology than any other set of alternatives we've seen recently.

 

[yottabit]

P4 was slower with SDR and a bit faster with dual channel DDR(875) or RDRAM(850) over single channel DDR (845).

I was going to say, about the only time I ever noticed a significant difference was going from SDRAM to DDR

To a lesser degree going from DDR to DDR2

DDR3 was a wash

It depends on when you adopt the technology IMO. If you adopt DDR4 out of the gate you probably won't see an increase. If you are upgrading to DDR4 a year or two after launch there will probably be much higher frequency RAM and you'll see a difference.

 

[SOFTengCOMPelec]

Thanks for all the voting and posts in this thread.

They have helped me much better understand what this new DDR4 is all about, and how it (going by majority of feelings) probably won't speed up Haswell-E 8-core, future systems at all, or only by a small amount.

Especially on the early DDR4 releases for (non-Server) Home/Gaming uses.

But, Servers MAY benefit by DDR4, especially many core (> 8 core), and when/if faster versions of DDR4 come out.


-----------------------------------------------------------------

I voted no significant increase, but that is strictly for "workstation" use, as quad channel DDR3 already offers MASSIVE bandwidth and 6 or 8 cores are not enough to chew it in real world tasks that are run on those workstations.

For servers DDR4 is gonna be great, for example 2 socket servers:

1) Currently server DDR3 is stuck on 1600, and if you go beyond 16 slots speed drops fast. While IVB based server processors will increase that to 1866, same 16 slot max speed limitation will remain. So if DDR4 will start @ 2133 and allow filling 24 slots at that speed it can offer substantial advantage in density and speed.
2) Once IVB-EP is out, 12 core / 24 T CPU can get bottlenecked by memory, you don't have to run anything special, enterprise app running on JVM will do, as garbage collectors scale almost perfectly with threads and memory bandwidth.

Thanks.
Your post has (also, the other posts were moving me quickly in this direction as well) significantly improved my understanding of why going from DDR3 ==>> DDR4, is not expected to improve the speed of Haswell-E 8-cores.

It's because existing (later generation) DDR3 memory (assuming you have enough banks of them e.g. Socket2011, 8 RAM slots full) is NOT constraining the processor speed. We are not memory speed constrained, at the moment (assuming non server/workstation usage).

If everyone was using ancient DDR(1), and it was still running at the sort of frequencies it was at, when DDR was first released, and the final limitation, was that there was just ONE ram socket.

THEN, going to DDR4, would (probably) have brought a huge speed improvement, now (DDR(1) ==>> DDR4).

------------

After all these posts, I now much more fully understand, so thanks again (to ALL).

 

[tweakboy]

Haswell E is in 2014 almost 2 years from now. Why talk about it like its out already...

Also to clarify. Ivy Bridge E desktop enthusiast will be 4930k the 600 dollar chip. All 3 models are 6 core. 4960x 6c12t.

Also to clarify there is no 6 core Haswell on earth and dont expect any more cores unless you go to server line and pay 1k per CPU and 1k for mobo.

Haswell quad cores for rest of 2013. Ivy E 4960k the fastest desktop processor will be 6c12t and thats how it squashes a 4 core haswell. gl