Contribute
Register

Performances issues with a 3930k

Status
Not open for further replies.
Joined
Mar 8, 2012
Messages
16
Motherboard
Gigabyte Z390 UD
CPU
i9-9900K
Graphics
RX 590
Mac
  1. MacBook Air
Hello all,

first of all, thanks to tonymac and its community. I'm a happy owner of three great hackintoshes (server, workstation, render cow) thanks to your help. :)

My first post is about a performance issues. (geekbench 32, single / multi-core score)


Let me explain : on a 4770 based setup (B85, 12gigs) i got 3228 / 12681 on OS X and 3389 / 13129 on Windows 7. That's an expected difference, no problem at the moment.

But on my main setup (3930k, X79, LGA2011, 32gigs) the difference is huge : 2915 / 21236 on OS X and 3853 / 22793 on windows.

In real world use I can feel that After Effects is a bit more laggy when rendering a single frame (in a live composition, when just a thread is used frame by frame) by switching to OS X / Windows.
That's not so slight to be forgotten.

So anyone has a clue ?

----------

An other problem not related to the first, occuring on OS X AND windows (Believed at first time that OS X only was affected, but that's not OS dependant).

Basically the 3930k get destroyed by the 4770 on a frame by frame basis (when only mono-thread is engaged).

I knew that the sandy bridge architecture was out-performed by Ivy but by a small margin (15 or 30% at max)

But actually it's more one a 3 to 4 ratio, 3930k is muuuuch slower to render my composition (noticeably with heavy effects like Trapcode Particular, for those who knows After Effects world).

When I inspect Task Manager on win or Activity monitor on OS X, I can clearly see that the 4770 setup use much more that 100% on the thread (300 to 400%) when the 3930k never go upon 100% on the single thread.

It's like the 4770 knows how to distribute a mono-thread on 3 or 4 cores where the 3930 is capped to 100% (one core used at full)

Actually my main working beast is a slug when working with AE compared to the much lighter 4770 setup (a lot less ram, not overclocked, no RAID, etc)

Do you know if intel added a specific "killer" feature that paralellize a single thread across multiple core or something like that ?

It's seems to be after effects related, but not sure at all. (Other appz never seems to be capped to 100% like this on a single thread)

Of course, this performance gap remains even if I tried different BIOS version, defaut or custom BIOS settings etc.

I'm a bit stuck, if you know...
 
Some benches that sum up everything,
total time to render the same after effect project, multi processing disabled. Fresh installs for OSes (I swap boot drives in both setup for consistency)

4770 / OS X : 6m20
4770 / WIN : 5m58 (+6% over OS X)

3930k / OS X : 19m3
3930k / WIN : 14m57
 
Status
Not open for further replies.
Back
Top