6 cores or more

Does anyone have any actual data on core loads for Elite: Dangerous in VR with 6 or more cores?

I've got a 6600k @4.2, and running HWiNFO64 I see max load on all 4 cores while parked in the station. So I was wondering if Elite: Dangerous actually utilises 6 or more cores efficiently.

Please note: I'm not interested in virtual hyperthread loads. I'm talking about actual core loads.

So if anyone has actual readings for 6 or more cores under E: D load, I would be much obliged. :)

Thanks in advance!
 
It does seem to spread its load across all cores, slightly favouring the actual cores vs the hyperthread ones.

For some reason I cant use printscreen when in vr... but heres a quick snap from my taskmanager when sat in a station in the regular game.. Just imagine the usage at closer to 35-40% for VR rather than the 15-20% in the shot, the load distribution is virtually identical between vr/non-vr.

hzVJH6q.jpg
 
It does seem to spread its load across all cores, slightly favouring the actual cores vs the hyperthread ones.

For some reason I cant use printscreen when in vr... but heres a quick snap from my taskmanager when sat in a station in the regular game.. Just imagine the usage at closer to 35-40% for VR rather than the 15-20% in the shot, the load distribution is virtually identical between vr/non-vr.


Thanks Jon. Nice CPU you got there. So, do you have a Vive Pro to go with that? :)

Ok, so new motherboard and cpu is on the table. Damn. I was kind of hoping for limited results. That way I could have opted for second hand 7700k and kept my motherboard instead. Oh well... :)

I have an i5 8600k and ED seems to use all cores. Its a big improvement over my old i5 2500k.

Cheers Max. :)
 
Cheers Max. :)

No Probs. I was thinking of getting an i7 3770k as an upgrade, but decided to do the whole lot. I can upgrade to an 8700k if needed in the future. I also got my CPU at a budget price of £170 off ebay which meant I could spend a bit more on my motherboard and ram.

The only thing I need to update now is the GPU for when my Pimax comes. I will see if I can use my AMD fury with it, if not I will sell and upgrade.
 
Righto, finally coaxed the printscreen button into life while under an actual VR load.. heres a new screenie, it does actually seem to slightly favour one core (although its still spread across all logical cores.)

IJzkDYA.jpg
 
You can easily use 16 Cores according to this thread
I'd say a 6 core is the bare minimum for VR, happy with my 8 core.

and you'd better wait another 2-3 weeks, the CPU market prices will shift again once the new AMDs are out tomorrow. 2nd hand CPUs should show up plenty as well.
 
Last edited:
I've literally just switched from using a system with a 6600k to an 8600k, and keeping the ram and GPU the same the performance bump was substantial (easily 15 to 20%).
That said, like all VR games, I'm sure you'll get benefits from faster clock speeds and hyper-threading moving to the 7700k. From what I've read, in most VR applications the 7700k performs similarly to the 8600k. For games that aren't optimized for more than 4 cores I'd imagine the 7700k would be substantially better.

The more you crank the settings, the more your GPU is going to be the ultimate bottleneck, so you've really just got to have enough CPU grunt to keep pace with the rest of your hardware. I'd personally be curious to know if anyone's compared a 7700k directly to an 8600k for this usage?
 
...
The more you crank the settings, the more your GPU is going to be the ultimate bottleneck, so you've really just got to have enough CPU grunt to keep pace with the rest of your hardware. I'd personally be curious to know if anyone's compared a 7700k directly to an 8600k for this usage?

The speedy 4 cores are good in peak and average FPS. But the problem is on peak loads with many threads they stutter juggling in an out data on the caches. Every VR game has at least one more thread for the 2nd view and additional threads for the HMD runtime as well as things like ASW and not to forget head tracking. Chances are with speech recognition and other tools even more threads get added. That is way more than the usual gaming benchmarks put to the test that are made to make sponsors happy.

4 Cores are history today, just get rid of that crap as fast as possible, even if it's a 7700k, there is no bright future for those in VR.
 
And from the same article:

la48sBt.jpg

I'm starting to lean towards the much cheaper 7700k upgrade solution. Already did the bios update on my z170 pro gaming, so it should be good to go.
 
Last edited:
And from the same article:


I'm starting to lean towards the much cheaper 7700k upgrade solution. Already did the bios update on my z170 pro gaming, so it should be good to go.


Probably not a bad way to go. Worst case scenario, you sell it after a while and upgrade to something else. In my experience CPU's tend to resell quite well.

On this same topic, has anyone compared a 8700k to an 8600k with ED? I'd imagine that'd give you another good indicator of how well virtual threads perform vs physical cores.
 
...On this same topic, has anyone compared a 8700k to an 8600k with ED? I'd imagine that'd give you another good indicator of how well virtual threads perform vs physical cores.

Indeed. VR specific mind you, as it represents a different kind of load entirely. Based on the article mentioned, threads are better than cores. As of last generation anyway.
 
i7 4770k @ 4.5gh.. gtx1080.. 32gig ram ( 2400 ) windows on ssd - ED on separate ssd. Oculus rift cv1 - i run mostly high settings - some ultra and shadows low. SS @ 1.65 using debug tool.. my gtx runs at 2100/5400 smooth as butter with no stuttering anywhere. ive never seen my cpu go above 50% on any cores while playing ED it usually sits around 30-40% usage.
 
Righto, finally coaxed the printscreen button into life while under an actual VR load.. heres a new screenie, it does actually seem to slightly favour one core (although its still spread across all logical cores.)

On modern cpu's threads are moved from core to core really quickly (I suppose that is so the individual cores stay cooler, they can then have higher turbo clocks or run slightly more efficiently). So looking at task manager to gauge how well something is multithreaded doesn't work anymore.

Look at this single thread of prime95 loading both cores to 50%:
full.png
 
Last edited:
The problem with Steven Burkes testing is he used the training missions. AFAIK these are not CPU taxing, players with bad CPUs struggled in the open game near busy stations. It really needs some experienced ED Players for relevant testing.
 
Back
Top Bottom