mdcurry
The blueroom score is for future true 4k vr. No video card can reach this. I do not think the first Voltas will. In the racing sims, they are so cpu intensive and while they use all threads, they really load up one to two threads, so single core performance is still important. That is why in most games the 7700k wins over the x99 stuff. but the 7700k throws off too much heat. that is why I avoid it. My overclocked 980 Tis at 1500 were truly awesome cards. In vr, in the real world, in the sims they really are almost as good as 1080 TI. It is hard to hit the sweet spot in vr, so some performance is wasted. I have tested a new mystery racing sim. going from 1 car to 30 doubles the cpu load! I have the new Pimax 4k heatset to test. It is great for Movies or 3d photos. 

TRClark911
Good post.  I wasn't sure what the blue room score was for but as someone who VRs with the Rift I can tell you it wasn't for anything currently available today.
 
 https://youtu.be/Y67b9CsniU0
 
That game came out last week and it's phenomenal.  The best looking game currently available in VR IMO... although lots of titles look really good.  Still, it doesn't even phase my system at all while the blue room was a real test.
 

sethleigh
 
I'm really curious about this result, since though my best Fire Strike score beats yours, it only does so because of the cpu tests, where my additional cores come into play. Your graphics tests crushed mine, and my 1080ti was running a pretty good OC. Perhaps single-thread performance from your cpu running 600Mhz faster than mine makes some of the difference, but I doubt it.

sethleigh
ETA: one of the huge benefits of watercooling, as you know, is that you don't cross the temperature thresholds that cause the gpu to throttle down to lower speed steps.

VipergtspaI just got done running heaven and your right, I had to lower my OC to +150 core +600 memory. Core clock stayed at 2050MHz until about 60C then dropped to 2038MHz, most of the run (30+ minute loop) it ran at 2025MHz no lower. Memory speed was 6106MHz and power limit was normally sitting at 115%, sometimes higher sometimes lower.
 
I will try later to see if I can tweak some settings and do more testing.

sethleigh

TRClark911On a side note, I keep reading about at higher resolutions like 4k that the GPU matters more than the CPU... but I'd argue if we were all running the same resolution and same single GPU that the differences in framerate would be noticeable and you'd be able to pick out who had the 7700k... just like you can in this comparison.  I realize this is 1080p and not 4k but I'd be interested in comparing Firestrike Ultra scores just to see if I'm right or wrong.  We all are running a single 1080 Ti so it would be a good CPU comparison.

 
At 4K the framerates are lower, which means the cpu work of generating the content to send to the gpu for each frame is less, and the gpu work of rendering the larger 4K frames is higher, so at 4K the cpu clockspeed difference should be lower. I agree that if all three of our gpus ran at exactly the same gpu/memory speeds, the one with the highest cpu speed should edge out the others, dependent on various factors like IPC for the given cpu generation, cache benefits, etc.
 
For the sake of science and the children, and world peace, a final solution to the Palestinian/Israeli problem, and to give all of us angst-ridden anti-Trumpers more peace of mind, here is my current best Fire Strike Ultra score. I'd love to compare it with yours and Vipergtspa's so we can see the relationship between gpu speed, cpu speed, physics and graphics scores, etc. Given our mutually understood theory about the cpu and gpu burden during these tests, I predict that yours and Viper's advantage in the graphics tests compared to mine will shrink dramatically (at least cut in half), and that my lead in the physics tests will remain. We'll have to argue about whether e-peen girth or length are more important, but that argument can be had later. 

Fire Strike Ultra: 7803 http://www.3dmark.com/fs/12527007

Given our mutually understood theory about the cpu and gpu burden during these tests, I predict that yours and Viper's advantage in the graphics tests compared to mine will shrink dramatically (at least cut in half), and that my lead in the physics tests will remain. We'll have to argue about whether e-peen girth or length are more important, but that argument can be had later. 

At 4K the framerates are lower, which means the cpu work of generating the content to send to the gpu for each frame is less, and the gpu work of rendering the larger 4K frames is higher, so at 4K the cpu clockspeed difference should be lower.

sethleigh

TRClark911That's true... temps aren't a problem anymore but hitting the power limit has been with higher memory OC as I said.  I've dropped from 2076mhz to 1911mhz because of it.   I'm still working on it though to find my happy medium.

 
Yes, as I've mentioned in one of the threads, I'm convinced that the reason some people report their scores going down with higher memory speed is that the higher memory speed eats up more of the 120% power limit, and in places in a given test where the gpu itself is using more power, this can force the gpu to clock down a step or two to preserve the 120% limit. These momentary down-stepping events are enough to cost a few points in the test, because it seems the cost of dropping 12.5Mhz or 25Mhz at times is greater than the benefit of running memory at 6000MHz rather than 5454Mhz or whatever.

When tuning my OC using Heaven, I found the max 100% stable gpu OC, then tuned the memory boost to maximize score by reaching that sweet spot where the memory and gpu shared the 120% power limit with the greatest benefit from both. It turns out that in the Superposition 4K Optimized test, my gpu speed was on average lower (higher cpu overhead on that test?) than it was in Heaven, and I was able to see a benefit to raising the memory boost for that test to take advantage of the 120% power limit.

sethleighIt's so true that the relationship between usable gpu and vram speed varies from benchmark to benchmark due to their own particular patterns of use. One really would have to tune settings for each individual benchmark to maximize all the scores.  These settings (+165 gpu/+625 vram) would have lowered my Heaven scores compared to +165/+450, but in Fire Strike Extreme gave a noticeable improvement.

sethleigh
I think the big takeaway here, at least in these benchmarks, is that for 4K resolution the fastest cpu isn't necessary, as we see with your 4.9GHz 7700 not beating my 4.2GHz 6900 even in the graphics scores. And with 1080p, who cares, since both of our cpus driving a 1080ti could probably peg any 1080p monitor made.
 
That leaves the intermediate realm of 1440p/144/165Hz refresh monitors, where your 7700k would presumably blow my 6900 away. Or in the hypothetical realm where 4K/144Hz monitors come out and I sell a kidney to buy one, my slower cpu speed could limit what I can get, even with 1080ti SLI.
 
I guess the conventional wisdom regarding "gaming performance" and cpu/gpu choice really depends on what monitor a person intends to use. I'm not sure this would have been obvious to me if we hadn't thought this through like this. But for now, I simply don't buy the logic that a fast 4-core is automatically a better gaming cpu than the somewhat slower 6- or 8-core cpus.


ETA: while I was on a roll, I cranked my gpu/vram settings to +165/+625 and re-ran Fire Strike Ultra, and improved my score over the 7803 I'd gotten with +165/500 or whatever it was I'd used. 
Fire Strike Ultra: 7843 http://www.3dmark.com/fs/12527444
 
This is almost making me want to screw around with the voltage curves and see if I can get 2088 Mhz stable in Heaven, then use that in the 3DMark tests... 
 

sethleigh
I do, because I'll often game with Virtual Box running a Solaris or Linux VM, SolidWorks 2016 with some part I'm designing for my 3D printer, Chrome with a zillion windows of a zillion tabs each, Skype, Thunderbird, TeamSpeak, and God only knows what running in the background, and my gaming performance isn't impacted in any way I can detect. In the bad old days when I used to run 2- and then 4-core cpus I'd make sure to shut everything down while gaming to make sure the game got all the cpu grunt there was to be had. Now it's simply not an issue. 

sethleigh
And thanks to you I'm going to go out and find a VR rig to try out.

TRClark911

sethleigh
And thanks to you I'm going to go out and find a VR rig to try out.

Heh... what got me hooked was when my friend showed me the Oculus Dreamdeck demo... it's a free app that you can download once you set up the Oculus Rift.
 
The T Rex from Jurassic Park was bad ass.
 
https://youtu.be/kJx2uS6hd6o
 
Just remember that's a 2D youtube video... with the headset on it's an entirely different experience but that's how VR is.  You can't really show people 2D videos... VR is something you have to experience with the headset for obvious reasons. 

sethleigh
TRClark and DRY_ICE, do you guys know what sorts of plans there are for the popular games out there to implement VR support? I mean things like Battlefield, The Division, Call of Duty, things like that. I'm pretty good at keyboard/mouse stuff where I wouldn't necessarily need to see my keyboard while playing to successfully use all the controls I'd need, so it would be possible to wear a headset while sitting at my desk and just play headset/keyboard/mouse rather than stand up in a room and wave around the special controllers. At least, without having tried it, that seems to be the way I'd want to use it.

I'd love to play games like Warthunder, while driving my tanks around in a VR environment where it looks true 3D rather than just amazingly rendered 2D on a monitor. I don't know why it would be so hard for companies to support this, but I have no idea how close we are to seeing the popular games out with VR support.

sethleigh
TRClark and DRY_ICE, do you guys know what sorts of plans there are for the popular games out there to implement VR support? I mean things like Battlefield, The Division, Call of Duty, things like that. I'm pretty good at keyboard/mouse stuff where I wouldn't necessarily need to see my keyboard while playing to successfully use all the controls I'd need, so it would be possible to wear a headset while sitting at my desk and just play headset/keyboard/mouse rather than stand up in a room and wave around the special controllers. At least, without having tried it, that seems to be the way I'd want to use it.

I'd love to play games like Warthunder, while driving my tanks around in a VR environment where it looks true 3D rather than just amazingly rendered 2D on a monitor. I don't know why it would be so hard for companies to support this, but I have no idea how close we are to seeing the popular games out with VR support.