i7 8700K looking good!

[Gunner Deckard]

I would agree on my old setup I could see around 30% to 50% usage CPU and GPU but it would struggle in certain situations. I have a feeling it's a game engine problem or limitation and the higher clocks with the new I7 8700k really help feed the data Elite needs. Just a guess.

 

[NeilF]

It is a VR issue. I never had any problems before even on my triple monitor set up. What people some people have noticed is there are instances of very high CPU use and the GPU isn't at 100% but the headset can't maintain 90FPS. So something is holding back the GPU (what exactly is proving a bit elusive) but it looks the 8700K/Chipset/RAM is brute forcing its way out of it. Rendering two separate 1080x1200 images at 90FPS + positional tracking etc seems to be quite a task. My i5 4670K is more than enough for playing on a monitor.

Oh... I'm sure it was inferred it was a general issue... I'll try in VR then... But if the CPU usage increases much I'd be surprised and if it does IMHO that just shows a significant bug in ED.

 

[Gortron]

Oh... I'm sure it was inferred it was a general issue... I'll try in VR then... But if the CPU usage increases much I'd be surprised and if it does IMHO that just shows a significant bug in ED.

I've seen my CPU go to 100% in certain situations and often in the high 80's/90's especially in busy CG stations. Oddly on certain planetary outposts too, even in solo. I seem to remember getting very high CPU use in High intensity combat zones but was hard to monitor anything without getting blown to bits hehe.

I do suspect there is some issues with the code but in VR the CPU needs to make twice as many draw calls to the GPU so CPU usage increasing significantly makes sense.

 

[TorTorden_VR]

Indeed and the thing with trippels, at least when using surround vision is that the AMD\Nvidia drivers are combining the screens into one virtual device for windows, so it only calls for a single draw.

 

[Gortron]

Indeed and the thing with trippels, at least when using surround vision is that the AMD\Nvidia drivers are combining the screens into one virtual device for windows, so it only calls for a single draw.

Yep and V-Sync is normally on @60FPS but with VR there are 2x as many draw calls that need to be made 50% faster to get 90FPS.

 

[Cmdr Kurnek]

Ah, shucks here it is.
To be taken with one of these: https://upload.wikimedia.org/wikipedia/commons/c/c2/Salt_truck_Milwaukee.jpg

https://www.youtube.com/watch?v=3lXGR0VyQPk&feature=youtu.be

Was this with any SS/HMD multiplier? What did you use to track the FPS?

 

[TorTorden_VR]

No supersampling.
SS and hmd-Q at 1x.

As for fps I just used the ingame counter (CTRL-F).

If I engage any SS with my current settings I start to trigger ASW more frequently.
I don't see enough benenfit with SS to decrease settings in favour if it.

The pimax is going to be a tough cookie.

But it won't be the cpu's fault.

 

[NeilF]

It is a VR issue. I never had any problems before even on my triple monitor set up. What people some people have noticed is there are instances of very high CPU use and the GPU isn't at 100% but the headset can't maintain 90FPS. So something is holding back the GPU (what exactly is proving a bit elusive) but it looks the 8700K/Chipset/RAM is brute forcing its way out of it. Rendering two separate 1080x1200 images at 90FPS + positional tracking etc seems to be quite a task. My i5 4670K is more than enough for playing on a monitor.

Well, here's me in that same HazRes for 30-40mins in VR and max detail... What's interesting is, at times my GPU was pushed so went from 90fps to 45fps (it's what it's designed to do of course). I didn't notice in game but only when I came out to grab screen shots.

What was very telling was one time it went from 90fps -> 45fps -> 90fps when an Imperial Clipper simply appeared in my instance, and then when I blew it up! Quite literally just the Imperial Clipper appearing in my instance (not nearby or in my line of sight) to my hit my GPU enough to mean I dropped from 90fps.

Strange - As a rule when a lot of ships appeared that burst my GPU usage so it dropped to 45fps. And if an Imperial Clipper was around that seemingly alone would burst my GPU usage!

Anyway, my CPU looks fine all the time? 40% or so in use?

Here's the death of an Imperial Clipper, and 45fps -> 90fps accordingly.

A long period with lots of ships so GPU overloaded so 45fps all the time...

In this sequence I boosted further out and close in, and as the ships were lost, and re-appeared you can see the GPU flicking between 90 and 45fps.

 

[Wasp0]

Well thats Imperial 'biowaste' ships for you!!

 

[PetroVitallini]

Yes, YES YOU DO!!

Just do it! An early Christmas present to yourself. You know you have worked hard this year, sacrificed plenty, so hey you deserve it. YOU KNOW YOU DO, The forum Knows it, even TJ knows it!!

I expect an answer back informing the forum you have placed an order!!

I actually decided not to upgrade. Good arguments given, but I have to save up some funds for something other I have planed. Might want to put in some overtime in the months to come. Might do it then

 

[NeilF]

Well thats Imperial 'biowaste' ships for you!!

Well, I still don't understand how a single ship appearing (Imperial Clipper) can seemingly use up 5-10% of my GPU power? When it's off in the distance, and I can't even see it?

Feels like some code could be improved there!?

I'm almost tempted to raise it as a bug TBH!

 

[TorTorden_VR]

Slightly off topic.

But I'm not just seeing benefits in ED.
Tested some other games that I before chocked up to "nah plenty good on cpu" like Doom (reboot not vr), and Lone echo.
Almost as much a boon in responsiveness as ED across the board.
In fact Pcars seems perhaps to be one of the least affected by the CPU upgrade but that could be how I have been playing it mostly.

 

[Gunner Deckard]

Did you oc yet your leaving a lot of performance on the table. Elite seams to like really high clocks.

 

[TorTorden_VR]

Did you oc yet your leaving a lot of performance on the table. Elite seams to like really high clocks.

Not yet :S

Something for the weekend perhaps.
Uploading my latest vid with the i7 4790k now.

Still a rather decent machine even though the GPU is downgraded to a 'mere' 980ti ...

 

[NeilF]

Not yet :S

Something for the weekend perhaps.
Uploading my latest vid with the i7 4790k now.

Still a rather decent machine even though the GPU is downgraded to a 'mere' 980ti ...

I'd still be fascinated to see any CPU charts showing the 4790 being significantly worked in ED? I've never noticed it, and indeed even in these fabled Hazres I can't get it over 40-50% usage at VR Ultra detail (see above)?

 

[TorTorden_VR]

I'd still be fascinated to see any CPU charts showing the 4790 being significantly worked in ED? I've never noticed it, and indeed even in these fabled Hazres I can't get it over 40-50% usage at VR Ultra detail (see above)?

It's not a matter of being worked to capacity.
You will never see a computer worked to capacity in VR, its a question of latency between frame generation.
And if Oculus doesn't see enough headroom on any segment be it GPU, CPU or whatever it will trigger spacewarp to maintain responsiveness.
And it won't release itself out of ASW until you have more than half capacity to spare, which is why I saw myself with a gpu running at 25% and a cpu running at 49-60%.

Also my 4790k was not the limiting factor when I had a 980ti, a 1070 is on on par with that, It's main draws are lower TDP, and slightly higher amount of Memory.
ED doesn't use more than 5.5GB or VRAM regardless so this wasn't a bottleneck on the 980ti.

Also, It might just not only be the CPU.
There has been quite considerable speed increases when going from DDR3 to DDR4, and z97 chipsets to z170 or z270.
Everyone seems to latch on the CPU, but we can't really measure a 4790k against an i7 7700 or the new i7 8700k because you can't even post the latter on a z97 board and vice versa.

And the crucial nature of VR is for the entire system to reach and maintain a frame generation time of about 11ms.
And the CPU needs to manage to squeeze not one but two draw-calls inside that time frame.

I shall do some quick cuts on it tomorrow and you can see and make your own mind.

I do however agree that the 4790k isn't bottlenecking a 1070, but it did, and just barely so btw my 1080ti.

I also did a quick review of the video, and one contender for needing optimisations is the HUD.
I engaged reboot repair once during it, and as soon as HUD vanished I saw fps jump back up, as soon is came back online, hello spacewarp.
Another target for optimisation or just plain removal is to get rid of dropped loot.
This tend to build up after a while and just hangs there, clogging up the system.

If I'm not interested in egnineering parts why waste the cpu cycles.

 

[NeilF]

It's not a matter of being worked to capacity.
You will never see a computer worked to capacity in VR, its a question of latency between frame generation.
And if Oculus doesn't see enough headroom on any segment be it GPU, CPU or whatever it will trigger spacewarp to maintain responsiveness.
And it won't release itself out of ASW until you have more than half capacity to spare, which is why I saw myself with a gpu running at 25% and a cpu running at 49-60%.

Also my 4790k was not the limiting factor when I had a 980ti, a 1070 is on on par with that, It's main draws are lower TDP, and slightly higher amount of Memory.
ED doesn't use more than 5.5GB or VRAM regardless so this wasn't a bottleneck on the 980ti.

Also, It might just not only be the CPU.
There has been quite considerable speed increases when going from DDR3 to DDR4, and z97 chipsets to z170 or z270.
Everyone seems to latch on the CPU, but we can't really measure a 4790k against an i7 7700 or the new i7 8700k because you can't even post the latter on a z97 board and vice versa.

And the crucial nature of VR is for the entire system to reach and maintain a frame generation time of about 11ms.
And the CPU needs to manage to squeeze not one but two draw-calls inside that time frame.

I shall do some quick cuts on it tomorrow and you can see and make your own mind.

I do however agree that the 4790k isn't bottlenecking a 1070, but it did, and just barely so btw my 1080ti.

I also did a quick review of the video, and one contender for needing optimisations is the HUD.
I engaged reboot repair once during it, and as soon as HUD vanished I saw fps jump back up, as soon is came back online, hello spacewarp.
Another target for optimisation or just plain removal is to get rid of dropped loot.
This tend to build up after a while and just hangs there, clogging up the system.

If I'm not interested in egnineering parts why waste the cpu cycles.

Huh? But one of the underlying comments in thread is that their old CPU (eg: I4790K) was giving them hiccups in VR - "inexplicably get 100% CPU use from time to time in ED while in VR."

When I've done tests in 1920x1200 and in Oculus VR (in ultra detail), my 4790K has not even really got much past 50% usage. What is hitting limits is my 1070 which at times flicks back to 45fps instead of 90fps due to bursting 100% GPU usage.

What I've noticed about that is strangely ships even in the distance, out of sight, seem to add significant GPU usage. I intend to test this. But IMHO it looks like some questionable coding. ie: An Imperial Clipper arrives in the instance, which is little more than a dot, and out of direct line of sight, and GPU usage goes up by 5-10%?

Note: In my tests in a HazRes, even when I dropped from 90fps to 45fps I didn't notice until I looked at the graphs showing my FPS.

 

[-VR- Max Factor]

Huh? But one of the underlying comments in thread is that their old CPU (eg: I4790K) was giving them hiccups in VR - "inexplicably get 100% CPU use from time to time in ED while in VR."

When I've done tests in 1920x1200 and in Oculus VR (in ultra detail), my 4790K has not even really got much past 50% usage. What is hitting limits is my 1070 which at times flicks back to 45fps instead of 90fps due to bursting 100% GPU usage.

What I've noticed about that is strangely ships even in the distance, out of sight, seem to add significant GPU usage. I intend to test this. But IMHO it looks like some questionable coding. ie: An Imperial Clipper arrives in the instance, which is little more than a dot, and out of direct line of sight, and GPU usage goes up by 5-10%?

Note: In my tests in a HazRes, even when I dropped from 90fps to 45fps I didn't notice until I looked at the graphs showing my FPS.

Your i7 4790 shouldn't be getting to 100% usage even at stock clocks. My old i5 2500k at 4.2ghz doesnt get to 100%. Gets in the 90's though around stations and res sites though.

 

[TorTorden_VR]

Only you can decide wether you have a problem or not.

If you don't you don't. Nothing is better than not having problems.
I'm not here to convert you to a religion or anything.

I'm merely describing my experiences as I have them with this game and various upgrades and observations I have made.

Here's the short list.
I5 4670k bottlencked my 980ti in VR, (out of VR this combo was rocking my 3440x1440 at ultra settings and steady 60fps)

I then added the 4790k, this 'unlocked' the 980ti.

I then upgraded to the 1080ti, also switched RAM from 1600mhz to 2400mhz, and this was also a noticeable upgrade.
Not huge but very noticeable, but then again I'm in the 1% who does see when ASW kicks in even just for a few seconds.
But if it does so because the GPU and not the CPU or other factors it looks a lot smoother and less artefacting than the other way around.

The real problem here if there is one, is the game itself.
There are some really hinky stuff going on in this now rather old engine, I'm really hoping FD considers implementing some of the VRworks features.

There are rumblings now that SinglePass shading etc and other features are making their way into UE4 engine and Unity's main branches and more and games will start to use them and the performance costs of running ED in VR will smell like an old shoe in comparrison.

There are still HUGE software benefits to be found for VR, it is a completely new method and need to meet entirely different demands than.

Only real recommendation I would give now, is to just wait a year or two and buid a new rig.
Don't think I would recommend adding a post Pascal GPU to Devil's canyon era hardware (4790k, z97, DDR3)
Even the new i5 8600k are matching the i7 7700k blow for blow and that's still $100 cheaper than it after the recent discounts.

The only thing I shall make a blatent claim about is the death of the mantra of "Cores\threads don't matter to gaming".
And the new rig is a definite upgrade.

 

[NeilF]

Your i7 4790 shouldn't be getting to 100% usage even at stock clocks. My old i5 2500k at 4.2ghz doesnt get to 100%. Gets in the 90's though around stations and res sites though.

My 4790 isn't getting 100% usage, but we seems to have some people in this thread who are, and people who are upgrading to things like the 8700K to get over CPU issues with such processors.

Personally I cannot get my 4790K over 50% usage even in VR Ultra detail. The only issue I can hit is my GPU exceeding 100% usage and therefore reducing from 90fps to 45fps, but even then I don't witness any real significant issue while using it (I didn't even notice it had happened).

So TBH I'm confused about this whole thread hence me doing testing and working out what is going on. The only thing I have seemingly seen and need to test further is excessive GPU use by ships in an instance. eg: I can't understand why an Imperial Clipper turning up and being rendered as a dot out of sight is possibly using 5-10% of my GPU!? Imagine what happens then when half a dozen+ ships are around... even as distant blips.. (Are there rendering issues in ED such that there's large GPU usage for no real reason?)