Sandy Bridge Preview

i have several sandy bridge processors and boards. some are beta others are retail release.
 
the 2300 is a 2.8GHz and it does not even beat the 2.8GHz 760. in fact is about on par with the 750. or the AMD 1090T.

the improvements are more seen in Turbo boost and unfortunately sandy bridge turbo is worse for audio than previous.
turbo was highly suggested to be off previously now its a must.

so so far they suck... we have had maybe 6-8 hours in front of a sandy bridge with communication between several board manufactures and Intel as well trying to resolve the turbo thing..
this is the first time in a good while i have been disapointed in an Intel release.

on a better note the memory bandwidth is about 10% better but still far behind that of an X58 platform. 

Thanks for the update Scott. Do you have any contacts at Intel that you can lean on to hopefully get things improved for DAWs?

Do you have any contacts at Intel that you can lean on to hopefully get things improved for DAWs?

 
FWIW, We DAW users (as a whole community) are an extreme small part of the Intel market share.
For the most part, we're along for the ride...  
The situation is even worse with laptops... for the same reason.

Eotm


Thanks for the update Scott. Do you have any contacts at Intel that you can lean on to hopefully get things improved for DAWs?

we are premier intel partners so we have 3rd level engineer contacts yes.. (had them long before being premier however) i have had the chips and intel boards for almost a month. took 10 days just to get it to be able to post and load windows.
then several back and forths over a few things. (very common when you get alpha or beta boards) during that time is when i emailed my contacts with a few mobo manufacturers to get retail releases..
 
funny thing is i got slightly better base #s with the beta intel boards. (turbo off)
 
i love new platforms....not!
 
 

just a quick update to this.
without getting totally techy on you...

this architecture is unlike anything else we have ever tested/validated
when you change motherboards windows is not happy in fact we had to reinstall. (at first we assumed a ram issue or it just got corrupted somehow)

bear in mind when you swap out mobos with the same chipset in the past it was never an issue, might have a few drivers to change (usually lan.)

that was oddity #1.

changing cpus in the same board. windows EXPLODES again. we are like huh?

1) there are so many components on die (on the processor) now thats its like changing motherboards.

2) each processor is different by a good amount.

now to the best part which is going to make me end up eating some of my words i have posted concerning Sandy Bridge.
(foot in mouth)

please bear in mind we are no where done validating this platform and still a week away from being ready to release it as known "working perfect" and discoving all the tricks. we are however newly energized to get it done after first thinking, well it sucks..

the 2600k acts completely different from the 2300 (wishing i had tried it sooner i have had these well over a month ago)

as known Turbo boost can create bad DPC spikes when left on.
the 2300 was some of the worst we seen. far worse than P55/X58 platform
however the 2600k is so far NOT
turbo is also very different with sandy bridge

with a heavy project using Dawbench the turbo was hitting 3.9GHz

and was basically keeping up with a 980x. yes i just said 980x pick you jaw up off the floor.

again we discovered this @ 7pm last night so much more validation to do... including video testing...

needless to say we are excited again to test this platform... should have more today

The sandy bridge NDA lifted today and the reviews are starting to post.  I'm still wading through the Anandtech review.

Lots of info but of course no audio/DAW specific info.

Thanks Scott and company for all your hard work doing the tests and posting
for all to see

LOL how about 10 core (20 with HT x 4) quad server... so thats 80 cores.. 40 real and 40 HT... got 20k? each cpu is $4k

4 options
SB 2600k OCed to 4.5GHz
950 oced to 3.6GHz
980x stock
980x @ 4ghz <---- sick power

the X58 still has far better memory bandwidth.

but basically yes to your answer...

hi,

first thing.. stock 2600 is self ocing as in Turbo mode. up to 3.7GHz.. i had mine turboing to 3.9 with a few tweaks..
(nice to see turbo working and not causing issues... at least on the 2600. the 2300 was horrid...)

i now have the 2600k running at 4.7GHz rock solid, trying for 5GHz.
i will be selling them @ 4.5GHz..
no numbers yet for audio.. the video geeks have the systems and wont let me touch it right now...

2) onboard video.. i dont trust it it seems worse than onboard video thats on a H55.
whilst supposedly enhanced (better) i think they put all the eggs into DVD Playback?
large track count was slow on redraw and zoom in/out was a tad slow as well..
onboard would be ok for lite duty users.. past that not recommended.

way to diificult to explain in a forum post there may be a how to out there somewhere but it is rather early..

980X - $999
[link=http://www.newegg.com/Product/Product.aspx?Item=N82E16819115223]http://www.newegg.com/Pro...x?Item=N82E16819115223[/link]
SandyBridge 2600 $329
http://www.newegg.com/Product/Product.aspx?Item=N82E16819115070

Am I missing something?  By the benchmarks the 2600 is slightly ahead of the 980X but is only $329 while the 980X is still 1K.  hmm....

Interesting stuff, aside from the speed and performance... you say memory bandwidth is not as good? What about heat or power consumption?

Scott-
    If you were to unofficially recommend a top of the line DAW build (using only one computer) would you recommend the 2600 over the 980X?  Performance vs $$$ of course

What kind of cooling is needed to OC the 2600K to 4.7GHz?

Thanks.
Dave