Kaby Lake processors: bor-ing ! - Page 2

mackerel · 2016-12-18, 10:32

The official public show was the "New Horizons" stream/video. You can see it at http://www.amd.com/en-us/innovations/new-horizon as well as download the Blender test they used. Note there was a mistake in an earlier provided download where samples were set to 200 not 150. That will impact timing significantly, and be cautious of this if comparing with results found elsewhere.

AMD want to give the impression that Ryzen 3.4 GHz = 6900k stock, and will probably undercut it by some not insignificant amount on launch and trumpet that. Of course, the full statement should include the phrase "in this Blender test." This then swings around to another question, how does the AMD Blender test actually behave? I tried to answer that with limited testing. The summary would be:

Ram speed doesn't seem to make a significant difference (<1%, 6700k 4.2 GHz, 2666 vs 2133 dual channel dual rank. I didn't try crippling the ram further)
6MB to 8MB L3 cache doesn't seem to make a difference (<1%, 6600k vs 6700k HT off)
HT gives a significant boost, about 52% faster (or, takes 66% of the time compared to HT off)
Haswell to Broadwell, and Broadwell to Haswell gave about 3% IPC boost each. Caution limited test samples, I'll see if I can use the other linked sheet to verify this.
Pentium G4400 IPC 4% lower than 6600k. The G4400 is Skylake without AVX extensions, suggesting those are not used as I'd expect a much bigger difference if it was.

So although this Blender test does use all available threads, it doesn't seem to be a particularly heavy stress.

For the audience here, it doesn't give any clues as to possible FMA performance which is where things really get interesting to me. I've seen conflicting info on this in the past, and my current position is it'll be half IPC of Intel. I hope I'm wrong on this point and they manage to match Intel.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Hyperthreading broken in Skylake and Kaby Lake?	GP2	Hardware	4	2017-06-26 02:08
Kaby Lake Memory Speed	airsquirrels	Hardware	12	2017-06-22 14:48
Kaby Lake / Asrock disappointment, RAM weirdness	Prime95	Hardware	17	2017-01-27 21:09
Kaby Lake chip	Prime95	Hardware	0	2016-10-26 23:23
64 bit processors	brandon_2003	Hardware	52	2005-03-27 11:19

2016-12-18, 10:32	#12
mackerel Feb 2016 UK 448₁₀ Posts	The official public show was the "New Horizons" stream/video. You can see it at http://www.amd.com/en-us/innovations/new-horizon as well as download the Blender test they used. Note there was a mistake in an earlier provided download where samples were set to 200 not 150. That will impact timing significantly, and be cautious of this if comparing with results found elsewhere. AMD want to give the impression that Ryzen 3.4 GHz = 6900k stock, and will probably undercut it by some not insignificant amount on launch and trumpet that. Of course, the full statement should include the phrase "in this Blender test." This then swings around to another question, how does the AMD Blender test actually behave? I tried to answer that with limited testing. The summary would be: Ram speed doesn't seem to make a significant difference (<1%, 6700k 4.2 GHz, 2666 vs 2133 dual channel dual rank. I didn't try crippling the ram further) 6MB to 8MB L3 cache doesn't seem to make a difference (<1%, 6600k vs 6700k HT off) HT gives a significant boost, about 52% faster (or, takes 66% of the time compared to HT off) Haswell to Broadwell, and Broadwell to Haswell gave about 3% IPC boost each. Caution limited test samples, I'll see if I can use the other linked sheet to verify this. Pentium G4400 IPC 4% lower than 6600k. The G4400 is Skylake without AVX extensions, suggesting those are not used as I'd expect a much bigger difference if it was. So although this Blender test does use all available threads, it doesn't seem to be a particularly heavy stress. For the audience here, it doesn't give any clues as to possible FMA performance which is where things really get interesting to me. I've seen conflicting info on this in the past, and my current position is it'll be half IPC of Intel. I hope I'm wrong on this point and they manage to match Intel.