CPU; # of threads matter much?

[TorTorden_VR]

And a funny thing.
The main difference between the default VR medium and VR ultra is extra SS and AO.
Pretty much all the rest of the settings are left at medium.

 

[TorTorden_VR]

Thats a great example of AO in action. I just can't seem to notice how it works in ED

It is a very subtle softening on the lighting, mostly in cockpit and inside stations.

It is like AA not a very good implementation, passable for 2014/2015 game but nothing special.
And it sucks up performance like a sponge for exceedingly little benefit in this game.

 

[Juniper]

[anecdotal]

I have a workstation PC running a 2950x Threadripper (16 core, 32 thread), with 32GB of RAM and a 1080Ti.

I also have a "regular" gaming PC with an FX8350, 16GB of RAM and also a 1080Ti.

Whilst this isn't the case for every piece of software I run, in the case of VR with E: D I've noticed a surprising lack of difference between the two machines in terms of performance / frame rate. With Task Manager open, I have to say that on the TR, E: D doesn't come anywhere close to utilising all 16 cores never mind the threads.

Of course, the Threadripper is not designed for games (although it does run them quite well thank you), but I was surprised by this. I was expecting a much bigger performance hike over the older FX8350 system.
[/anecdotal]

 

[SimonNQ]

I concur with most of what has been reported, My testing is with an 8700K @5Ghz and ProcessLasso ( program used to set process settings )

I tested by doing the following:-
Set Elite Dangerous exe to use the first 6 cores only, set VR compositor and all steam exe's to last 6 cores, set Elite exe process to HIGH as there are some windows processes that cannot be set to only use specific cores etc.. ALL other processes were set to use cores 6-11, with cores 0-5 used for essential windows and elite exe only.

Result, didn't really notice much difference at all I expected any spikes in the game exe to be minimized, however they still remained. FPS was about the same, loading times about the same. Did notice slightly better loading speed once returning to normal space from SuperCruise especially when dropping out at stations etc, but only very minimal. Overall, not worth the effort.

My current setup is to NOT assign any specific cores to Elite exe, but to set the process to HIGH. Using PocessLasso, this is automatically set every time I start the game etc.. This has had a more positive affect on the game, you can also use ProcesLasso to reduce the priority of other exe's at the same time etc..

ProcessLasso https://bitsum.com/?f

 

[Vtskr]

You really don't need to change game process priorities to high especially in VR gaming. Elite dangerous might have some background threads running that do not affect FPS and have low priority set by devs. If you set ED process to HIGH these background threads might end up having higher priority than compositor. So threads that do not affect FPS will steal CPU time from threads that actually affect FPS

 

[SimonNQ]

You really don't need to change game process priorities to high especially in VR gaming. Elite dangerous might have some background threads running that do not affect FPS and have low priority set by devs. If you set ED process to HIGH these background threads might end up having higher priority than compositor. So threads that do not affect FPS will steal CPU time from threads that actually affect FPS

this is why we can spread processes over specific cores using affinity.. So they cannot interfere with more important processes etc..

 

[IronmanG]

I concur with most of what has been reported, My testing is with an 8700K @5Ghz and ProcessLasso ( program used to set process settings )

I tested by doing the following:-
Set Elite Dangerous exe to use the first 6 cores only, set VR compositor and all steam exe's to last 6 cores, set Elite exe process to HIGH as there are some windows processes that cannot be set to only use specific cores etc.. ALL other processes were set to use cores 6-11, with cores 0-5 used for essential windows and elite exe only.

Result, didn't really notice much difference at all I expected any spikes in the game exe to be minimized, however they still remained. FPS was about the same, loading times about the same. Did notice slightly better loading speed once returning to normal space from SuperCruise especially when dropping out at stations etc, but only very minimal. Overall, not worth the effort.

My current setup is to NOT assign any specific cores to Elite exe, but to set the process to HIGH. Using PocessLasso, this is automatically set every time I start the game etc.. This has had a more positive affect on the game, you can also use ProcesLasso to reduce the priority of other exe's at the same time etc..

ProcessLasso https://bitsum.com/?f

Are you saying above 6 cores does little and threads make hardly any difference?
When I had my rift and was struggling slightly with performance I emailed occlulus tech support. They told me that it didn't really utilise threads and 4 cores was all it was OK for as that's what companies in the main were making back then. Could try and find the email. In the end I got a clocked 8600k at 4.8 and that is fine

 

[SimonNQ]

Are you saying above 6 cores does little and threads make hardly any difference?
When I had my rift and was struggling slightly with performance I emailed occlulus tech support. They told me that it didn't really utilise threads and 4 cores was all it was OK for as that's what companies in the main were making back then. Could try and find the email. In the end I got a clocked 8600k at 4.8 and that is fine

This depends completely on how the software is written... There is no hard and fast rule.. yeah some older software and or game engines are still using 32bit as well..... More modern games will utilise your hardware far more than older game engines especially.

Further to this, I can say my testing has not really shown any advantages or disadvantages in playing with cores or hyper threading, obviously, if you turn cores off it will etc... 4 Cores/8 Thread is fine for most games as most games will want more GHZ than cores per se'

 

[Benji85]

Elite loves threads/cores will use 16

 

[IronmanG]

The long and short if vr is you need a 2080ti and 9900k maxed out and even them you won't be reaching full potential is some titles all of the time. The next gen might be closer but they they will change the HMD to use it more. The circle of life

 

[Arguendo]

It's been a while, so this is a semi-necro, but I thought it was worth an update.

Finally got my hands on an i7-7700k at a price that was too good to let go, and got to test it in the game today.
My old i5-6600k was running overclocked at 4.4GHz and in stations with other cmdrs it gave me an FPS of 44-45 with ASW off. Running the "new" i7-7700k stock with the exact same graphics options never goes below 70 FPS, and most of the time 80+, in the same scenario.
My GPU is now the clear bottleneck, running at between 80 and 100% depending on what's going on, while the CPU is running at a nice 60-ish%. That gives my CPU the overhead to run other programs that are non-GPU intensive in the background, without affecting Elite, which is exactly what I was looking for.

The i7-7700k is running at 4.4GHz according to GPU-Z and RealTemp GT when playing, which should be exactly what the i5-6600k was running at as well with my OC. The 7th gen vs 6th gen obviously helps, but it also does look like number of threads matter as well. Atleast when you're running a sub-par GPU.
I for one am happy with my upgrade and can live with 70+ FPS with a 1060 6GB

 

[ziddan]

Thank you for this latest report, i had been considering an upgrade from my i5-4690k@4.5ghz to an i7 for a while but finding real world examples of results to be expected for doing so has been somewhat annoying.
I found a cheap 4770k last wednesday and decided to risk it, its on its way currently and while i wont be too sad if this one wont go up to 4.5ghz, i do expect at least a slight performance boost for the higher thread count.

 

[Milner]

Thank you for this latest report, i had been considering an upgrade from my i5-4690k@4.5ghz to an i7 for a while but finding real world examples of results to be expected for doing so has been somewhat annoying.
I found a cheap 4770k last wednesday and decided to risk it, its on its way currently and while i wont be too sad if this one wont go up to 4.5ghz, i do expect at least a slight performance boost for the higher thread count.

I've got the same setup so will be interested in your experience with the i7. My current plan is to see what AMD comes out with mid year and then do a new build with whatever offers the most bang for buck.

 

[ziddan]

I will be reporting back with before/after frame timings docked in the same station, expecting the i7 to arrive within a day or two.

 

[Arguendo]

Thank you for this latest report, i had been considering an upgrade from my i5-4690k@4.5ghz to an i7 for a while but finding real world examples of results to be expected for doing so has been somewhat annoying.
I found a cheap 4770k last wednesday and decided to risk it, its on its way currently and while i wont be too sad if this one wont go up to 4.5ghz, i do expect at least a slight performance boost for the higher thread count.

Your processors look pretty much like mine did; same frequency just newer generation and more threads. Curious to see if you see an improvement like I did.

 

[Benji85]

It's been a while, so this is a semi-necro, but I thought it was worth an update.

Finally got my hands on an i7-7700k at a price that was too good to let go, and got to test it in the game today.
My old i5-6600k was running overclocked at 4.4GHz and in stations with other cmdrs it gave me an FPS of 44-45 with ASW off. Running the "new" i7-7700k stock with the exact same graphics options never goes below 70 FPS, and most of the time 80+, in the same scenario.
My GPU is now the clear bottleneck, running at between 80 and 100% depending on what's going on, while the CPU is running at a nice 60-ish%. That gives my CPU the overhead to run other programs that are non-GPU intensive in the background, without affecting Elite, which is exactly what I was looking for.

The i7-7700k is running at 4.4GHz according to GPU-Z and RealTemp GT when playing, which should be exactly what the i5-6600k was running at as well with my OC. The 7th gen vs 6th gen obviously helps, but it also does look like number of threads matter as well. Atleast when you're running a sub-par GPU.
I for one am happy with my upgrade and can live with 70+ FPS with a 1060 6GB

Yep I saw massive gains going from 2700k to 9900k basically double FPS.

How can you game without asw though? I find without it it's just not fun with all the suttering I'd you drop below 90.

Got an Odyssey plus the other day and man is it good. (When I mod it for comfort)
Blows the rift out of the water.

 

[Arguendo]

How can you game without asw though? I find without it it's just not fun with all the suttering I'd you drop below 90.

People have said this before, and I don't understand it. On my system there is stuttering if I leave it on, and that goes away when I turn it off. Have no clue why that is, as others are reporting the opposite

 

[ziddan]

Ive received my i7 now but im kinda hesitant to put it in since i wont be able to get a reliable "before" reading with my i5 as Elite performance is currently messed up by the patch.

EDIT: Reporting back, my ingame settings seem to have been mostly gpu bound so there was a negligible change for what i checked, some flying about round stations.
On benchmarks i gained about 5% by going from an i5-4690k to an i7-4770k.
This chip doesnt go as high in ghz tho, my i5 was comfortable at 4.5ghz while this i7 only does 4.2ghz so at the exact same clock the gain might have been a little higher.

 

[Milner]

Ive received my i7 now but im kinda hesitant to put it in since i wont be able to get a reliable "before" reading with my i5 as Elite performance is currently messed up by the patch.

EDIT: Reporting back, my ingame settings seem to have been mostly gpu bound so there was a negligible change for what i checked, some flying about round stations.
On benchmarks i gained about 5% by going from an i5-4690k to an i7-4770k.
This chip doesnt go as high in ghz tho, my i5 was comfortable at 4.5ghz while this i7 only does 4.2ghz so at the exact same clock the gain might have been a little higher.

Shame to head the i7 didn't make a big difference for the effort. You've probably said elsewhere but what's your gpu? Thanks for the feedback this confirms my earlier plan and see what's available after AMD launch their next CPU lineup.

 

[ziddan]

Im running on a 1080 for now, im hoping they somehow managed to backport foveated rendering but im not all that optimistic.
It runs most thing good enough still tho.