[Eurogamer] CPU Bottleneck, perhaps its actually RAM?

[TheELF]

Fact is no matter what DDR4 memory you pair with an i3 6100 in Ryse, it'll continue to be a bottleneck for Ryse in CPU demanding sections. For this game you need an i5/i7/FX 8-core to have any hope of maintaining 60 fps.

For those who have been following i5 bottlenecking in CryEngine games, this wouldn't be surprising. CryEngine games LOVE 6-8 threaded CPUs + IPC. If you do not have an i7 in CryEngine games, even if you use DDR4 2800, it's not going to help the i5 overcome the lack of sufficient threading capability:

390x @ 1140/6200
i7-6700K@4.6
i5-6600K@4.6
https://www.youtube.com/watch?v=AbsXPjVcc6w

No, fact is that the bottleneck=CPU demanding parts are the parts that only use two threads/cores to a high degree and not the parts that use more threads, look at the video.
2 threads (looking straight ahead) at ~30% with ~85% CPU usage for the whole game = 25-30FPS
4 threads (looking at the sky) at ~10-20% with ~80% CPU usage for the whole game = over one hundred FPS
It works much like what I said earlier about mantle/Dx12,it breaks up frame rendering into multiple threads but the core of the game ,in Crysis 3, still only uses two threads.

Sure the i7 at the same speed = huge cost difference is able to run those two threads a bit better then the i5 but still a faster i3/i5 will run those two threads faster than a slower i7 but may have lower FPS while looking at the sky/ground.

https://www.youtube.com/watch?v=lEafN99Q4K4

 

[Flapdrol1337]

For those who have been following i5 bottlenecking in CryEngine games, this wouldn't be surprising. CryEngine games LOVE 6-8 threaded CPUs + IPC. If you do not have an i7 in CryEngine games, even if you use DDR4 2800, it's not going to help the i5 overcome the lack of sufficient threading capability:

Not all cryengine games, mechwarrior online barely scales over 2 cores.

 

[Head1985]

What's it's showing is how poorly optimized/outdated the game engine is and sensitivity to memory speed on Skylake vs. other systems isn't logical either since HW-E destroys Skylake in overall memory bandwidth.

My theory is that skylake can use faster memory for better fps.
Haswell-e maybe have more memory bandwidth but cant use it in games for better Fps.

Skylake have big architectural changes vs haswell intel even didnt metion alot of them.
We already know this:

And few big that intel didnt mention

There're some new details on Skylake:

* Front end now has 5 decoders from the usual 4.
* Micro-ops cache can deliver 6 m-ops/cycle instead of 4.
* loop-buffer size is now 64 m-ops
* bigger OoO structures (but no official numbers cited, IIRC ROB size is 224 entries, RS size is 97)
* page split load penalities from 100 cycles to 5 (that's an improvement!)
* longer idle time for the PAUSE instruction
* faster L3, 2-cycles per line now

http://www.realworldtech.com/forum/?threadid=154100&curpostid=154100
http://www.intel.com/content/dam/ww...4-ia-32-architectures-optimization-manual.pdf

 

[Deders]

Not sure what speed those test are done at but the results differ from my experience.

This was done with lots of things happening in the background.

I concur with Head1985 that it looks like Skylake can make better use of wider memory bandwidth than Haswell.

 

[Headfoot]

I'll be buying faster RAM on my next build, that's for sure

 

[bystander36]

My theory is that skylake can use faster memory for better fps.
Haswell-e maybe have more memory bandwidth but cant use it in games for better Fps.

Skylake have big architectural changes vs haswell intel even didnt metion alot of them.
We already know this:
http://abload.de/img/1d3kgm.jpg
http://abload.de/img/2vjk64.jpg
http://abload.de/img/3czjou.jpg

And few big that intel didnt mention

http://www.realworldtech.com/forum/?threadid=154100&curpostid=154100
http://www.intel.com/content/dam/ww...4-ia-32-architectures-optimization-manual.pdf

There is also a DDR3 vs DDR4 difference in that comparison. Comparing Haswell-E in dual channel mode might be a more direct comparison. Or Haswell-E in quad to see if Haswell-E holds an advantage.