New Haswell-E benchmarks

I know this is not the hardware forum... but!
usual Dawbench
 
yup its time again

bear in mind these are preliminary and stock only for
right now. OCed number later.

still have much tweaking to do. and more
stability.

5960x @3.00 (8 core)
Turbo @ 3.5
Win 8.1
Cubase
7.5.2
Hdspe
32 buffer 208
64 buffer 229
128 buffer 233
256 buffer 240
 
ok just got OCed numbers @ 4.5GHz

32 buffer 281
64 buffer 319
have to stop here as we are OUT of space. (exceeding test)



5930k @ 3.5 (stock) only turbo's to
3.7
32 buffer 162
64 buffer 178
128 buffer 189
256 buffer 197

basically a 20% jump from 4930K

over clocked 5930K @ 4.5GHz
32 buffer 197
64 buffer 213
128 buffer 224
256buffer 236


and 5820k @4.5GHz
32buffer 198
64 buffer 212
128 buffer 224
256 buffer 230



compared to:

Single 2697 v2 12 core 2.7ghz
RME
HDSPe
RXC-EXT
Cubase 7.5

@32 buffer 215
@64 264
@128 270
@256 279

for comparison

4930k @3.4 16gb RAM
RME HDSPe/
Digiface
Cubase 7.0.4

32 buffer 135
64 buffer 144
128 buffer 156
256 buffer 157

Hey, Scott.  Thanks for the numbers.  Is the 5930 a four or six core?
 
Kind regards,
 
Living Room Rocker

I am really surpirsed about the result of OCed 5930 6 core comparing to OCed 5820 4 core. Does that mean you better off with 5820? The numbers are almost identical but the prices are big difference when adding Motherboard and RAM.

Paul P
 
I'm curious as to how this relates to X3, especially since the tests are using Cubase.
 
In what situation would you need so much power ?
 

its irrelevant what DAW you use as long as that daw is multi-thread ability. (some don't do it well)
so the % ability would be the same in sonar vs Cubase vs PT

www.dawbench.com
 
number of RCX compressor.  he offers a few options.

Hey Sam,
no not really as far as 5930K and 5820K (5960X is obviously a different league, 8 Core.)
right now we are still sussing out the memory issues with this platform (why the systems are not up on our site yet)
we also don't have our own ram yet (be at least another week) so we are having to use the what I consider crappy ram off the shelf to test with. the 5820K is behaving differently that the other 2.
the 5820k is only stable with @2133, @ 2400 its a tad wonky. (we are using DDR4 8 gig sticks 2666 to test with)
once we have our own ram in stock we will be better suited to tell whats really up. (our ram is DDR4 2800 grade 1 A)
FYI 2400 is the highest we could go with any of the 3 chips.  and 2133 is really the only 100% stable one..
funky memory controllers and not the best ram
this has been a very rocky launch in many ways..

Thanks for the info Scott.

Do these CPUs need big noisy fans when OC'd?

Mike,
we are using liquid cooling and air cooling so no.. they are quiet.
 
 
Sam,
they all seem to OC the same. still not sure about the 5820k, wont know until I see good ram.
I have a feeling it might be limiting in a few ways, (mower memory bandwidth for one)
 
as to the other, really depends at this point both GHz and core count matter depends more on which libraries. orchestral would do well with both.  

well best I can do is 3770K
the benchmarks I have for the 2600K is from the older test and it does not 100% translate.
 
32 buffer 103
64  116
128  123
256 129
if I recall correctly the 3770k was not a big jump at all from the 2600k
 
basically the 5820K is double that of the 3770k
not to mention vastly better memory bandwidth