can someone explain this?

[Ashleyyyy]

so i've been trying to solve my pc's bottlenecking problem. 

i have a cpu bottleneck. my specs are in my signature. 

 

i wanted to play the original crysis again, but it was unplayable. the mouse stuttered, and it dipped down into 30fps range on low settings. 

 

now, just as an experiment to see what it would do i turned hyperthreading off. and what do you know. 50fps consistent (vsync on limits it at 50fps for some reason) with everything on low except texture detail (might be something else, the first setting in advanced graphics) turned up to ultra, and 8x antialiasing. and no stuttering mouse. 

 

but, when i ran geekbench 4 single core score was up by about 400 points... not that much. 

 

(crysis ran in a Windows VM with 6 core's allocated and my R9 290X passed through to it, geekbench 4 was run on my host OS, Fedora Linux.)

 

can somebody tell me why disabling hyperthreading made such  a massive difference in crysis? 

[spartaman64]

windows got confused with resource management for so many threads? idk

[Ashleyyyy]

25 minutes ago, spartaman64 said:

windows got confused with resource management for so many threads? idk

in both cases it had 6 core's allocated. only with hyperthreading on it had 12 out of 24 threads allocated so still 6 core's. 

[spartaman64]

1 minute ago, firelighter487 said:

in both cases it had 6 core's allocated. only with hyperthreading on it had 12 out of 24 threads allocated so still 6 core's. 

with hyperthreading the threads are probably weaker than without HT with HT on you are getting 6 logical cores while with it off you are getting 6 physical cores 

[Ashleyyyy]

38 minutes ago, spartaman64 said:

with hyperthreading the threads are probably weaker than without HT with HT on you are getting 6 logical cores while with it off you are getting 6 physical cores 

not by much... geekbench 4 multi-thread went from about 17000 with HT on to about 16000 with it off (i can't remember the score's exactly)

[spartaman64]

1 minute ago, firelighter487 said:

not by much... geekbench 4 multi-thread went from about 17000 with HT on to about 16000 with it off (i can't remember the score's exactly)

synthetic benchmarks dont always transfer well to real world applications

[Ashleyyyy]

2 minutes ago, spartaman64 said:

synthetic benchmarks dont always transfer well to real world applications

oh ok. i didn't know that. i assume turning off hyperthreading has some kind of performance hit in multi core tasks... 

[thedude4bides]

I just played through crisis a few months ago.  To get max FPS you have to play in windowed mode.  Full screen limited things for some reason.  I then installed a borderless windows app to make it look clean.  FPS shot from 50 to over 100 (vsync off).

 

Not sure what your hyperthreading issue is but this is what I did to play crysis.

[DildorTheDecent]

Hyperthreading on is only worth about 30% perf gain from having it off. 

 

Maybe Crysis or Windows wasn't a huge fan of being ran in a VM (I've no idea btw just speculation). Or maybe the GPU was unhappy with it. 

[Ashleyyyy]

9 hours ago, DildorTheDecent said:

Maybe Crysis or Windows wasn't a huge fan of being ran in a VM (I've no idea btw just speculation). 

it's still being ran in a VM... 

[DildorTheDecent]

1 hour ago, firelighter487 said:

it's still being ran in a VM... 

Which means...what?

[Ashleyyyy]

28 minutes ago, DildorTheDecent said:

Which means...what?

if crysis wasn't a fan of being run in a VM it would run like crap regardless of hyperthreading...