Quantcast
Latest i7 Benchmarks Including 6950x 10 Core - Gearspace.com
The No.1 Website for Pro Audio
Latest i7 Benchmarks Including 6950x 10 Core
Old 4th June 2016
  #1
Gear Nut
 
🎧 10 years
Latest i7 Benchmarks Including 6950x 10 Core

Hi all, Dave at ADK. We have been doing some benchmarks with the new i7 chips, and I know there has been some interest in their performance, so figured I should start a new thread for these. Right now I only have ten core numbers. Other chips are here, just have to wait on the video benchmarks to get finished with the 10 core. I will update as I get to benchmark the remainder.

All testing is done using the Dawbench-EXT test, Cubase 8.5, and an RME HDSPE.

2687w V4
320 @ 32 Buffer*

* Out of instances running at about 90% CPU load. See later post regarding performance accross more tracks.

6950x
280 @ 32 buffer
320* @ 64 buffer

*This score is probably higher, but I am out of instances on this test.

6900k
227 @ 32 buffer
272 @ 64 buffer
293 @ 128 buffer
317 @ 256 buffer

For comparison:

Single 2697V3
264 @ 32 buffer
292 @ 64 buffer
318 @ 128 buffer

5960X
208 @ 32 buffer
229 @ 64 buffer
233 @ 128 buffer
240 @ 256 buffer

5930k
162 @ 32 buffer
178 @ 64 buffer
189 @ 128 buffer
197 @ 256 buffer

And our latest current generation quad i7 6700k:
121 @ 32 buffer
155 @ 64 buffer
161 @ 128 buffer
170 @ 256 buffer

Last edited by Bondo21; 10th June 2016 at 06:10 PM.. Reason: Added 2687W
Old 4th June 2016
  #2
Gear Guru
 
🎧 10 years
Interesting. The 5820k really does seem like a great value when doing some rough calculations on how many instances one gets per cpu relative to price (@32 buffer).
Old 4th June 2016
  #3
Gear Addict
 
🎧 10 years
Thanks for taking the time to benchmark these! Very impressive that 6950x. Pretty steep price at over $1700, though. I was hoping for a price drop for the 8 core, but it seems that that's not happening any time soon....
Old 4th June 2016
  #4
Lives for gear
 
Sounddesigner's Avatar
 
🎧 10 years
Impressive benchmarks!
Old 4th June 2016
  #5
Lives for gear
 
Finnish's Avatar
 
🎧 15 years
What's that current generation quad i7?
Old 4th June 2016
  #6
Gear Addict
 
🎧 10 years
Thanks!
How 6900, 6850 and 6800 look?
Old 4th June 2016
  #7
Lives for gear
 
rockreid's Avatar
 
🎧 10 years
Wow...Broadwell-E isn't f***ing around is it?

Well, looks like I'm in for my next 5 year build. 6950x here I come. Every multi-core bench test on this monster just crushes. Anyone have any info on that new Gigabyte X99 MB w/ Thunderbolt 3? Not much in the way of real world reviews as far as I can tell.
Old 4th June 2016
  #8
Lives for gear
 
jlaws's Avatar
 
🎧 10 years
Thanks for doing this, Dave! It looks like a monster. Any chance of overclocked results? Tomshardware got theirs up to 4.3 ghz which I think would be quite a boost...

I'd probably be all over this except that kaby lake will be the first platform that supports 3D xpoint, so it's probably worth it to wait for kaby lake x or skylake x.
Old 4th June 2016
  #9
Lives for gear
 
rockreid's Avatar
 
🎧 10 years
3dx point or Optane tech. Sounds great but I think I'm going to wait it out for the tech to mature a bit,and besides, my guess it will be quite expensive for awhile. Just a hunch, but this point in time is a sweet spot as far as upgrading now instead of waiting later. I have a feeling next year's Skylake-E (12 core guessing) x top tier chip will be even more expensive than the 6950x with diminishing rate of returns, especially accounting for overclock potential and perhaps problems with die shrink thermal issues.
Old 4th June 2016
  #10
Lives for gear
 
🎧 15 years
Have you also done DPC latency testing with each CPU to see how much of a latency hit you take with additional cores?
Old 4th June 2016 | Show parent
  #11
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by Finnish ➑️
What's that current generation quad i7?
6700k.

Fixed in the list.
Old 4th June 2016 | Show parent
  #12
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by jlaws ➑️
Thanks for doing this, Dave! It looks like a monster. Any chance of overclocked results? Tomshardware got theirs up to 4.3 ghz which I think would be quite a boost...

I'd probably be all over this except that kaby lake will be the first platform that supports 3D xpoint, so it's probably worth it to wait for kaby lake x or skylake x.
Technically this is mildly OC'd. It is running at 4.0ghz with 3000 RAM. We have been shipping our 5xxx series 2011 chips in the same ballpark since release and we have done similar with prior gens, so the results are comparable. For us that is stock. Per Eric, our head computer geek, this was not pushing the system in any way. Very little change to the voltage to get that. I will see if he has had time to push it any further.
Old 4th June 2016 | Show parent
  #13
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by Bassmankr ➑️
Have you also done DPC latency testing with each CPU to see how much of a latency hit you take with additional cores?
Short of running LatencyMon after doing my usual list of audio tweaks to the OS, no I have not. There was nothing sticking out at me that I remember on that part.

I won't get to the other chips until sometime next week.
Old 4th June 2016
  #14
Lives for gear
 
🎧 5 years
I'd like to try this on my Hoswell i5 just to see where I am power wise! Can I get the test file?

Since it's Cubase il love to run the test on both Mac and PC!
Old 6th June 2016 | Show parent
  #15
Gear Nut
 
🎧 10 years
Quote:
Originally Posted by classictunz ➑️
I'd like to try this on my Hoswell i5 just to see where I am power wise! Can I get the test file?

Since it's Cubase il love to run the test on both Mac and PC!
Test files are here:
DAW Bench : DAW Performance Benchmarking

Obviously results will vary based on which DAW you used and which audio device.
Old 6th June 2016
  #16
Gear Nut
 
🎧 10 years
I was able to get some preliminary results for the 6900k 8 core chip. This is essentially the update to the 5960x. 8 cores, similar clock speed, similar price point. Results were about what I expected with this not being a major architecture change (As in requiring new board/ chipset/ socket)with a small jump in performance over the similar product it is replacing.

Also, i am going to add these are preliminary and I may be adjusting them a bit once I can test on the exact same computer.

6900k
227 @ 32 buffer
272 @ 64 buffer
293 @ 128 buffer
317 @ 256 buffer
Old 6th June 2016
  #17
Lives for gear
 
Finnish's Avatar
 
🎧 15 years
What an amazing amount of power
Old 6th June 2016 | Show parent
  #18
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by Bondo21 ➑️
6900k
227 @ 32 buffer
272 @ 64 buffer
293 @ 128 buffer
317 @ 256 buffer

5960X
208 @ 32 buffer
229 @ 64 buffer
233 @ 128 buffer
240 @ 256 buffer
Interesting. While on small buffer size the difference is not that big (9%), it increases quite substantially on higher buffer settings (32%!).
Why is that?
Old 7th June 2016 | Show parent
  #19
Lives for gear
 
🎧 5 years
Quote:
Originally Posted by Bondo21 ➑️
Test files are here:
DAW Bench : DAW Performance Benchmarking

Obviously results will vary based on which DAW you used and which audio device.
Thank you!
Old 7th June 2016 | Show parent
  #20
Lives for gear
 
Finnish's Avatar
 
🎧 15 years
Quote:
Originally Posted by meloco_go ➑️
Interesting. While on small buffer size the difference is not that big (9%), it increases quite substantially on higher buffer settings (32%!).
Why is that?
32 samples is quite low, it's obvious that the system does not have a lot of headroom compared to 64 samples which is quite usual buffer setting these days. 32 samples is still something that requires a powerful and a stable system, i.e. optimized and tuned for DAW
Old 7th June 2016 | Show parent
  #21
Gear Addict
 
🎧 10 years
Quote:
Originally Posted by Finnish ➑️
32 samples is quite low, it's obvious that the system does not have a lot of headroom compared to 64 samples which is quite usual buffer setting these days. 32 samples is still something that requires a powerful and a stable system, i.e. optimized and tuned for DAW
Maybe, but anyway, 6900k performance increases from 32 to 256 by 39%, while on 5960 only by 15%.
Look at the pic -- 32 indeed puts a lot of strain on the system, but the difference increases as buffer size grows. I wonder why it happens, and also how 6850k and 6800k work in that regard. If same tendency exists, it might happen that 6850k could compete with 5960 at higher buffer sizes.

Old 7th June 2016
  #22
Lives for gear
 
Finnish's Avatar
 
🎧 15 years
Crushing power in that 6900k!
Old 8th June 2016
  #23
Lives for gear
 
🎧 10 years
Ok, I'll toss a hat in to this.

In fact Dave's results made me go back and recheck half of mine this morning, but they look solid, so I'll up them.

Setup = Asus X99 deluxe 1st gen (latest bios) / NI KA 6 (latest firmware / drivers) / Windows 10 - latest build.

Right, so I had to do a stack of retesting on earlier chips. Why? Well during testing (in response to Daves feedback I believe) Vin released a new beta, as these chips slay the current channel count.

(in fact see Dave's notes on this in the original post)

What I noticed this time. With the new firmware / drivers / W10 build I was seeing a stright increase of track count or 20 extra when I retested the 5820 / 5960X, when compared to my older results pulled in the last build of DAWBench, and using older drivers sets under W7.

I'm explaining this because my results are looking higher than Daves across the board, but they seem to be constantly better so I'm outlining some of the reasons the might be drift between setups.

End of the day however everything in the system can skew DAWBench results which is why you need a software install that's tweaked once and everything needs to be tested using it. My W7 drive got retired last week (not on purpose, some driver update nuked it), so now I suppose is a good a time as ever to make the jump.

I can't state if this is W10 specific due to some update tweak they've released since launch (we didn't see a jump in performance at launch, this may well come down to third party factors), or if it'd down to other driver changes (more likely), but its enough that it's invalidated my older results and all the results below are tested using the same OS/software install across all of them.

I'll also note that I've always found Reaper more efficient than Cubase, so that might account for some of the performance differences, I suspect ASIOGuard might have a impact here too but this isn't the thread to go diving into software at this point.

The important thing to point out is that my playing field is level, as I'm sure Dave's is but you can't draw direct comparisons between them.

Right... now I've posted the disclaimer of the day, on with some figures.

Komplete Audio6 interface - windows 10 - 44100Hz
Reaper as the DAW of choice using the DAWBench 2016 EXT beta build.

6800K @ Stock
32 - 209
64 - 229
128 - 235
256 - 245

6800K @ 4.2
32 - 238
64 - 256
128 - 275
256 - 286

6900K @ Stock
32 - 304
64 - 336
128 - 354
256 - 371

6900k @ 4.1
32 - 316
64 - 343
128 - 362
256 - 379

6950x @ stock
32 - 345
64 - 376
128 - 394
256 - 414

6950x @ 4.0
32 - 355
64 - 385
128 - 403
256 - 421

5930K @ Stock
32 - 187
64 - 201
128 - 211
256 - 219

5930K @ 4.2GHz
32 - 210
64 - 232
128 - 246
256 - 252

6700K @ Stock
32 - 162
64 - 174
128 - 182
256 - 186

6700K @ 4.4Ghz
32 - 168
64 - 180
128 - 191
256 - 197

6600K @ Stock
32 - 89
64 - 101
128 - 109
256 - 119

6600K @ 4.4Ghz
32 - 100
64 - 114
128 - 121
256 - 129

6700T @ stock
32 - 115
64 - 133
128 - 140
256 - 145
Old 8th June 2016
  #24
Lives for gear
 
rksguit's Avatar
Numbers Look Utterly amazing,however it ain't no $300 i7 2600K[my last desktop-,but I mainly use a W540 lap]here in "rip off-for audio and just about everything else land" that CPU is AUD $1,649-i7 6900K https://www.pccasegear.com/products/...w&gclsrc=aw.ds
Then Mobo,PSU,RAM,etc etc etc- I'm such an up vibe guy huh.

PS:Pete Kaine's results using the good but decently priced KA6 interface,are pretty amazing!

I speak too soon as always as the 6800K 6 core is far more affordable

Last edited by rksguit; 11th June 2016 at 11:56 AM..
Old 8th June 2016
  #25
Lives for gear
 
b0se's Avatar
 
2 Reviews written
🎧 5 years
That 6950x is a beast.

Pete - slightly OT, Scan rejected an RMA of mine weeks ago and have yet to send me back the item (mobo). Their online system is a pain to use, could you PM me to help perhaps?
Old 8th June 2016 | Show parent
  #26
Lives for gear
 
Finnish's Avatar
 
🎧 15 years
2600k here, too, I bought it five years ago. If everything stays just fine (no hardware failures etc) I have no need to upgrade hw. Software yes, occasionally. It's cool to see the progress in the hardware land, happily being and staying a bit behind

Quote:
Originally Posted by rksguit ➑️
Numbers Look Utterly amazing,however it ain't no $300 i7 2600K[my last desktop-,but I mainly use a W540 lap]here in "rip off-for audio and just about everything else land" that CPU is AUD $1,649-i7 6900K https://www.pccasegear.com/products/...w&gclsrc=aw.ds
Then Mobo,PSU,RAM,etc etc etc- I'm such an up vibe guy huh.

PS:Pete Kaine's results using the good but decently priced KA6,are pretty amazing!

I speak too soon as always as the 6800K 6 core is far more affordable
Old 8th June 2016 | Show parent
  #27
Lives for gear
 
🎧 10 years
Quote:
Originally Posted by rksguit ➑️
PS:Pete Kaine's results using the good but decently priced KA6,are pretty amazing!
Yeah, It's performing pretty well still to be fair, although of course if you happen to have invested in a better performing interface already (not that the KA6 is a slouch at the price point), then you can look forward to even more overhead to play with.

Quote:
Originally Posted by b0se ➑️
Pete - slightly OT, Scan rejected an RMA of mine weeks ago and have yet to send me back the item (mobo). Their online system is a pain to use, could you PM me to help perhaps?
Yeah sure, it's a different warehouse from the one I'm based in, but I can chase it over the internal system. I shot you a PM, so follow up over there with your details and I'll see what can be done.
Old 8th June 2016
  #28
Gear Addict
 
🎧 10 years
So it looks like 6800k beats 5930x? Looks promising.
Old 8th June 2016
  #29
Lives for gear
 
rockreid's Avatar
 
🎧 10 years
Just for laughs I'd like to see how many instances of Diva on Divine Mode the 6950x and the 6900k will crush out. Well, I'll get to see for myself I guess once these new CPUs are finally released.
Old 8th June 2016
  #30
Lives for gear
 
matucha's Avatar
 
1 Review written
🎧 15 years
It's interesting how little you gain by overclocking 6900k and 6950x, while 6800k seems to respond better.

I expected less difference between haswell-e and broadwell-e, the power consumption is also considerably less. Hmmm....
πŸ“ Reply

Similar Threads

Thread / Thread Starter Replies / Views Last Post
replies: 4616 views: 612375
Avatar for smoke
smoke 7th May 2021
replies: 98 views: 37658
Avatar for dfghdhr
dfghdhr 2 weeks ago
replies: 835 views: 176817
Avatar for jlgrimes11
jlgrimes11 30th August 2020
replies: 518 views: 41212
Avatar for audslu
audslu 28th December 2015
Post Reply

Welcome to the Gearspace 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…

Forum Jump
Forum Jump