The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Prelim Haswell Benchmarks Recorders, Players & Tape Machines
Old 12th April 2013
  #1
Lives for gear
 

Prelim Haswell Benchmarks

that time again!

got the new Haswell up and running!
its showing way better than expected and better than other benchmarks on it i have seen around the interweb
again very preliminary
but wow holy crap will grab OCed # soon had to change the heatsink this puppy is running hot

on the other hand the when the 2600 came out(budget original sandy) it was benchmarking right behind the 980X ( intels flagship at the time) so this is kinda a repeat..


Dawbench RXC test DAW Bench : DAW Performance Benchmarking

4770k @3.5GHz (stock)
8gb Ram
RME HDSPe
RXC Ext
Cubase 7.0.3 (also tried 6.5 and it was the sme)

120 RXC @32 Buffer
134 @64
142 @128
143 @256

for comparison

3770k @ 3.5GHz (ivy)
70 RXC @32 Buffer
80 @64
96 @128
113 @256


3930K @3.2GHz (sandy E)

133 RXC @32 Buffer
143 @64
145 @128
156 @256


Scott
ADK
Old 13th April 2013
  #2
Lives for gear
 
valis's Avatar
Looking good so far...
Old 13th April 2013
  #3
Lives for gear
Nice!

Sent from my SGH-T999
Old 13th April 2013
  #4
Gear Addict
 
Goddard's Avatar
 

That's pretty impressive! Thanks for the info.

Have fun with the o/c!
Old 15th April 2013
  #5
Lives for gear
 

should have the OCed #s by end of day
Old 15th April 2013
  #6
Gear Maniac
 

Wow, that's a nice step up. I've read about haswell having improved floating point performance, I think this could be the reason it does so well.
Old 15th April 2013
  #7
Lives for gear
 
valis's Avatar
Haswell improves floating point performance with specific new instructions, but the software has to be able to use the new instructions. So while you can support either AVX2 or FMA3 AND COMPILE THAT SUPPORT IN using a recent compiler, it still only improves code that takes advantage of vectorization. Floating point operations using AVX or plain old SSE3 see more modest increases in performance, not double as with the new instructions. And existing plugins certainly won't magically get double just by running on Haswell, even if they can handle AVX2 or FMA3 (for floating point multiply speed gains) they'll need to be compiled to support the new instructions.

In any case looking forward to more specs from scott...
Old 16th April 2013
  #8
Here for the gear
 
userfriendly's Avatar
 

Quote:
Originally Posted by valis View Post
In any case looking forward to more specs from scott...
Me too. Updated graphs especially. Pretty shiny graphs.
Old 16th April 2013
  #9
Lives for gear
Scott... have you heard anything new about LGA 2011? Any new SB-E's or is Ivy-E still on the way?

Or does it sound like they're jumping straight to Haswell, and a 3930K will end up being the max CPU on it (max AFFORDABLE CPU lol)
Old 16th April 2013
  #10
Here for the gear
 
userfriendly's Avatar
 

Quote:
Originally Posted by CPhoenix View Post
Scott... have you heard anything new about LGA 2011? Any new SB-E's or is Ivy-E still on the way?

Or does it sound like they're jumping straight to Haswell, and a 3930K will end up being the max CPU on it (max AFFORDABLE CPU lol)
They will not be jumping straight to Haswell-E - expect that chip in late 2014. Later this year there will be a Core i7-4820K (successor to the current entry level 3820, note they added a "K" to this one now too), as well as IBE successors to the other current SBE models. Core count will be the same, prices and TDP will be roughly the same as well, and there will be a slight increase in clock speed (i.e. 200 MHz more).
Old 16th April 2013
  #11
Lives for gear
^vverryyy nice.

So looks like no word on an 8-core LGA2011 huh? haha.


Thanks for the word on the street too man! It's tough to find info like that unless you're constantly in the know, which I definitely am not.
Old 16th April 2013
  #12
Here for the gear
 
userfriendly's Avatar
 

Quote:
Originally Posted by CPhoenix View Post
So looks like no word on an 8-core LGA2011 huh? haha..
I've only heard rumours about an 8-core SBE to be released around the same time as the IBE 4- & 6-cores - but that was just that, a rumour. TBH, that always sounded a tad strange to me, considering IBE has the better performance per watt ratio it would make more sense in my eyes if that 8-core part would be based on the IBE die as well. So, take that with a grain of salt. Or a bucket, depending...
Old 16th April 2013
  #13
Gear Addict
 
Goddard's Avatar
 

Haha, there are already 8-core LGA2011. They're called Xeons.

Hope the OCing is proceeding smoothly, Scott.

Old 16th April 2013
  #14
Here for the gear
 
userfriendly's Avatar
 

Quote:
Originally Posted by Goddard View Post
They're called Xeons.
Yes, and for a reason. A hefty price tag related reason.
Old 16th April 2013
  #15
Lives for gear
 

Wow, didn't expect results to be this good.
Old 16th April 2013
  #16
Lives for gear
Quote:
Originally Posted by Goddard View Post
Haha, there are already 8-core LGA2011. They're called Xeons.

Hope the OCing is proceeding smoothly, Scott.

Haha... outside of the overly expensive server-grade chips
Old 16th April 2013
  #17
Lives for gear
 

ok OCing done this processor is danged hot. I could not sell this as it is but at least I got numbers.
I know when the Ivy first came out they ran like scalded dogs. we were able to do some things in the bios to get the temps down.
thus far not so much but again not a lot of face time.

I7 4770k OCed @ 4.5GHz

32 : 121 RXC
64 : 129
128 : 138
256 : 144

pretty much dead on stock? having them re-run numbers again to be sure.
its only turbos to 3.9GHz @ stock so..
Old 16th April 2013
  #18
Lives for gear
 

Wonder if Intel went cheap with heatsink paste like they did with the Ivybridges?
Old 16th April 2013
  #19
Here for the gear
 
userfriendly's Avatar
 

Quote:
Originally Posted by jcschild View Post
pretty much dead on stock? having them re-run numbers again to be sure.
Yeah, that does seem a bit... odd.

Quote:
Originally Posted by Bassmankr View Post
Wonder if Intel went cheap with heatsink paste like they did with the Ivybridges?
Even if so, I still would have hoped for better overclocking. Seeing how much they were hyping that aspect.
Old 16th April 2013
  #20
Lives for gear
 
TAFKAT's Avatar
 

With the OC results , sounds to me like it was throttling due to overheating.

That would make sense.

Not at all happy to hear about the heat, bring on IBE... :-)
Old 17th April 2013
  #21
Gear Addict
 
Goddard's Avatar
 

Yikes! Would not have expected that, even with an engineering sample processor. Something must be haywire?
Old 17th April 2013
  #22
Lives for gear
 

De-liding an Ivybridge of mine and using decent heatsink paste went from a 4.3 Ghz to a 4.6 Ghz overclock at the same temps. Maybe it's time to take the lid off the sample and see what kind of job they did.
Old 19th April 2013
  #23
vfr
Gear Maniac
 
vfr's Avatar
 

The benchmarks are intriguing, but I'd be concerned that test results using engineering samples won't translate to production CPUs...

Sent from my Nexus 4
Old 19th April 2013
  #24
Gear Head
 

But there were lots of reports last week that Intel had started sending out Haswells to system builders with the official release due in June, so these are not going to be early 'engineering samples'.

In fact, when you add in the reports that the first batch of chips that have been made with the USB3 wake-from-sleep bug are not going to be recalled but sold, then that makes it look like what Scott is testing is the real thing.
Old 19th April 2013
  #25
Lives for gear
 
projektk's Avatar
 

Worth an upgrade from sandy? Doing everything I can to hold off until ddr4 is standard, from sandy that should be one hell of a system speed upgrade in all aspects of the build.

Sent from my SAMSUNG-SGH-I317
Old 19th April 2013
  #26
Here for the gear
 
userfriendly's Avatar
 

Quote:
Originally Posted by projektk View Post
Worth an upgrade from sandy? Doing everything I can to hold off until ddr4 is standard, from sandy that should be one hell of a system speed upgrade in all aspects of the build.
DDR4 is just barely on the horizon - and it being not just available and usable in widely available mainboards, but also reasonably priced compared to DDR3, might take another two years.
Old 19th April 2013
  #27
Gear Nut
 

I hate to be slightly dumb - but could someone translate what real-ish word speed improvements over a 2013 i7 there would be in percentage terms with these new set of processors?
Old 19th April 2013
  #28
Gear Head
 

Quote:
Originally Posted by Craig32768 View Post
I hate to be slightly dumb - but could someone translate what real-ish word speed improvements over a 2013 i7 there would be in percentage terms with these new set of processors?
Well, that's what is going to be revealed with the testing, but it's too early to say for sure now. Scott's results show big improvements at stock speeds, but there are lots of issues, eg. do these translate to other areas, why are the overclocking results so similar, is the chip running too hot to make this sustainable?
Old 19th April 2013
  #29
Lives for gear
 

just a quick FYI about this that semi explains the heat.
got an email from Intel that went to all manufacturers of motherboards and platinum intel members..

it has to do with power supply minimum loads and C6/C7 power states.. revealing in part some of the heat issues.

as Vin mentioned its most likely "turbo'ing" down due to the heat (shutting down cores etc..)
I will try to run it up again with several cpu monitors running at the same time
to see if we have 8 cores @ 4.5GHz or not.. thinking no.. even water cooling with fans blowing directly on the processor did no good. it will probably be a bit before we have updated bios for the C6/C7 states.
Old 19th April 2013
  #30
Lives for gear
 

So basically you have an overclockable (unlocked CPU) processor that won't let you overclock because of heat LOL. If you are getting so much heat that water won't let you get to 4.6 Ghz, I'd say Intel has a few bugs to work out. I'm still guessing bad contact or cheap paste between die and lid (time for the single sided razor blade Scott).
Topic:
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump