Jump to content
chiliwili69

Measuring rig performance: Common Baseline (for IL-2 v3.010)

Recommended Posts

On my i7-8700k, total system CPU usage while playing BoS in VR tends to be around 20% with 1 of 12 threads heavily stressed while the others have a light load or are idle.

 

As for how close BoS is to cpu stress tests, the answer is not at all. AIDA64 and CPU-Z stress all CPU threads fully. This results in maximum power draw (and also maximum Voltage droop at default settings). BoS/DCS only stress 1 thread, and so the voltage droop tends to be much more benign as is the heat generation. This is why it is possible to have a high overclock be stable in both games and yet instantly BSOD in a stress test.

 

For an explanation as to how all this works, here's a good youtube video from a well known overclocker:

 

 

 

hmm..that is so interesting. exactly what I experienced.  so If I want to overclock just for the sim, I can up my vcore a tiny bit just for stable game at be at 4.8 and have no issues, (so far as I can tell) but if I hit that stress button on cpuz -ZAP. 

Share this post


Link to post
Share on other sites

On my i7-8700k, total system CPU usage while playing BoS in VR tends to be around 20% with 1 of 12 threads heavily stressed while the others have a light load or are idle.

 

As for how close BoS is to cpu stress tests, the answer is not at all. AIDA64 and CPU-Z stress all CPU threads fully. This results in maximum power draw (and also maximum Voltage droop at default settings). BoS/DCS only stress 1 thread, and so the voltage droop tends to be much more benign as is the heat generation. This is why it is possible to have a high overclock be stable in both games and yet instantly BSOD in a stress test.

 

For an explanation as to how all this works, here's a good youtube video from a well known overclocker:

 

Ahh thanks. This is what I was hoping to hear. With those loads, I feel like I can run 4.8 just fine. If the loads were in the 70%-100% range, I wouldn't feel comfortable. All I really use this PC for is gaming. I don't do any rendering or anything like that. The voltage required to pass these stress tests for over an hour was getting to be too high.

 

Great video BTW. I've just spent that last day or two diving into what LLC does myself. Been learning how to OC manually with this new build.  On my past 2600K build, I just let the auto OC settings in my MB do all the work. So this has all been an interesting learning experience.

 

What CPU did you buy?

 

In this same thread, I put a post comparing IL-2 VR with Prime95:

 

https://forum.il2sturmovik.com/topic/29322-measuring-rig-performance-common-baseline/?p=514455

 

I don´t trust the %CPU or %core load measurements reported by some tools, including Windows performance monitor. Since threads jump from core to core.

At the end of the day, the real effect is in the temperature (more load-->more power-->more temp). And the temp is the real limitation.

I got an 8700K. My experience with Prime95 isn't so good. It runs fine for about 10-15 minutes at reasonable temps, then suddenly takes my CPU to 100C temps. I can run Intel XTU CPU test and memory test under the same settings and manage reasonable temps.

Edited by Warpig

Share this post


Link to post
Share on other sites
My experience with Prime95 isn't so good. It runs fine for about 10-15 minutes at reasonable temps, then suddenly takes my CPU to 100C temps. I can run Intel XTU CPU test and memory test under the same settings and manage reasonable temps

 

Prime95 has three kind of torture test:

- Small FFT

- Large FFT

- Blend

 

FFT means Fast Fourier Transform. I remember this code very well since I programmed it in C about 25 years ago in a supercomputer! (Chapter 12 https://www.amazon.com/Numerical-Recipes-3rd-Scientific-Computing/dp/0521880688) .Nice times!. 

 

FFT is used as a computational intensive code (stress test) just to bring your CPU to the very limit. If you use Blend, it is a mix of small FFT and large FFT.  So, if you leave prime95 running the CPU stress load is not constant, since the code is doing different things which demand more or less computation.

 

I find strange people running 24 hours prime95, when with less than 1 hour the whole rig reach steady-state heating-cooling equilibrium. I also don´t understand well (just my ignorance) why people run prime95 with more threads than what they are going to use later in their game or application.

 

I use my home PC just for playing IL-2 VR with OC (and my son just StarWars BattleFront but not OC). As I showed in the post above, the load required by Prime95 Blend torture test with 1 thread is demanding more than IL-2 VR, so no need for me to go beyond 1 (or maximum 2) threads in Prime95 stress test.

 

But if you use your PC for other games or compute intensive applications (like video conversion) then is a different story.

Edited by chiliwili69

Share this post


Link to post
Share on other sites

Chili I think the 24h att full load in prime95 is just a standard for those enjoying tweaking/overclocking as a hobby in itself. When every ounce of performance affects that ranking list.

 

In a smaller room a Pc at full load and running for 24 hours will increase room temperature. So that is a factor to consider for those guys hovering on the edge with their OC, before they can call it "stable". For us it's just about how far we feel like going.

Edited by a_radek

Share this post


Link to post
Share on other sites

 

 

In a smaller room a Pc at full load and running for 24 hours will increase room temperature.

 

Yeah!  that room will be in a house surrounded by night/day external temperatures as well...they will see night/day CPU temp variations when running their 24hr tests!  :wacko:  :wacko:   

Share this post


Link to post
Share on other sites

Some more interesting tests:

***

 

soooo..I overclocked the 1080GTX FE 10%power, +200 Mhz Gpu and +215 Mem Clock and got:

 

2017-12-19 15:12:11 - Il-2

Frames: 5133 - Time: 60000ms - Avg: 85.550 - Min: 61 - Max: 91

 

2017-12-19 15:31:38 - Il-2

Frames: 5133 - Time: 60000ms - Avg: 85.550 - Min: 61 - Max: 91

 

 

EDIT:chili, dont add to the chart yet though, for some strange reason I cannot replicate min 61/ 85 after going back and checking. not sure what happened

A-ha, I figured out my error and I replicated the above test. I guess when I installed a new tertiary HD for some VR games, etc., my page file changed to this new drive (I happened to pick A:). Reading about some DCS setings re: 'preload radius' I discovered that results were based on your memory aaaaand your HDs page file. So switched my page file back to my SSD and it resolved. (I could only run preload 40% of the way ~65K when everone else at 150K with my same setup). This ended up fixing IL2 as well. Point is make sure your page file is on your fastest drive, SSD in my case.

 

Frames, Time (ms), Min, Max, Avg

5115, 60000, 58, 91, 85.250

 

Frames, Time (ms), Min, Max, Avg

5152, 60000, 62, 91, 85.867

Edited by katdog5

Share this post


Link to post
Share on other sites

 

 

Frames, Time (ms), Min, Max, Avg 5115, 60000, 58, 91, 85.250 Frames, Time (ms), Min, Max, Avg 5152, 60000, 62, 91, 85.867

 

nice results, please, give also your CPU passmark and STmark to public these results with your CPU at 4.8GHz 

Share this post


Link to post
Share on other sites

 

 

For the top of the line hardware players, I do recommend the following. Fly an hour with max settings (ultra), shadows high, oculus Pixel density to 2 or SS-SteamVR to 4, and when starting in game, press and hold ctrl+numpad2 for a second. 45fps locked, no ASW, sharpen off, AA off. Butter smooth movements of camera, no artifacts. Markers and HUD off.

 

I have tried that, but I prefer 90fps. I tell you why.

 

put the right settings to have almost always 90fps (in my case very similar to the VR test settings) and now go to whatever plane in steady flight and while you look to your cockpit gauges and rotating your head press repeatedly Ctrl+Numpad1 (ASW OFF at 90Hz) and Ctrl+Numpad2 (ASW OFF at 45Hz).

 

At 90 fps everything is smooth, at 45fps you will see some judder in the gauges which blurs the image. This also extended to other objects, but the gauges are well noticed.

 

Now, with ASW off (to eliminate propeller effect without using mods) your choice is:

 

- decrease your settings to keep 90 fps almost always

- run at 45fps and increase your settings (keeping always above 45fps)

 

Personally I prefer the first option, but everyone will have their own opinion.

Share this post


Link to post
Share on other sites

nice results, please, give also your CPU passmark and STmark to public these results with your CPU at 4.8GHz

still at 4.7. I will recheck my cpu score tho. this bump based on 1080 oc only. just fyi..(my tests at 4.8 showed virtually no fps improvement)

Edited by katdog5

Share this post


Link to post
Share on other sites

still at 4.7. I will recheck my cpu score tho. this bump based on 1080 oc only. just fyi..(my tests at 4.8 showed virtually no fps improvement)

 

13282 overall CPU, single thread: 2743

 

1080GTX FE +10%power, +200 Mhz Gpu and +215 Mem Clock

 

Frames, Time (ms), Min, Max, Avg

5115, 60000, 58, 91, 85.250

 

Frames, Time (ms), Min, Max, Avg

5152, 60000, 62, 91, 85.867

Edited by katdog5

Share this post


Link to post
Share on other sites

I‘d take benchmark results with a grain of salt once you approach your system limits.

 

On a 7900X I can run individual cores up to 5 GHz when no more than 3 are in use. Gives about 2850 or so Passmark single thread. I can run BoX like that. On Prime95 it will bluescreen after about one minute. At 4.7 GHz I can run Prime95. About 2500 Passmark single thread. I don‘t see much of a difference between 4.7 and 5 GHz in BoX.

 

Being able to boot Windows/run Passmark/run BoX doesn‘t mean that you are not running a setting that may terminate your Bitcoin wallet with one misfired bit, bits that are irrelevant in BoX (aww, a wrong pixel, sad.) but that can have drastic consequences in other applications. Prime95 checks its results, hence it can determine if things went wrong somewhere and stop the processes.

 

Still working on the optimal setting for my cooler. The 7900X is really a piece of work. Once I have that, I will run the VR benchmark to add to Chilis collection.

Share this post


Link to post
Share on other sites

Hi guys,

 

Here are my results (seem very low but I guess maybe that's the 1060 3GB more than anything else)?

 

CPU: i5-8600K stock clocks

User Passmark: CPU: 13644 Single-thread: 2524

16GB DDR4-3200

GTX1060 3GB

Rift

Frames: 2388

Min: 22

Max: 44

Avg: 39.800

 

MSI Z370 SLI PLUS m/b

 

These scores make mine pretty much the lowest on the board but I feel that shouldn't really be the case with the CPU/Mem combo I've got there? Any advice?

I've pushed the clocks up to 4.7 but it didn't make that much difference. I generally play in VR on Balanced with ASW on (as I need it!) and most everything else (Grass, Terrain, Mirrors, etc. at Low-ish.)

Edited by wheeliemonsta

Share this post


Link to post
Share on other sites

I agree wheeliemonsta, your results should be higher. To figure out something's wrong by comparing to others is why this thread was created and you will get lot's of help. Let me start guessing: If running default 1.0 ss, are your results significantly better? Trying to figurer out if the issue is gpu or cpu related

Share this post


Link to post
Share on other sites

Thanks radek.

 

I'll check the scores a bit later today hopefully with 1.0 SS.

 

I certainly don't run any SS normally as I'm well aware the GPU is a bit below par for the rest of the kit. I've got a 1080Ti on the way soon but I couldn't get the lot together, so grabbed the CPU/Mem/MB combo first.

 

I do think though generally, given that IL-2 is mainly CPU-limited, I should get better performance than I do - would certainly be nice to run in-game settings a bit higher - I'm not that bothered about SS as I haven't seen much benefit in quality when I've tried it.

Share this post


Link to post
Share on other sites

Yes, I set it in the OTT...

 

I'm thinking about rebuilding the OS, etc. anyway at some point, so maybe that will make things better.

It's been many many years without a refresh on various hardware, so I'll maybe bite the bullet soon and clean her up a bit!

Edited by wheeliemonsta

Share this post


Link to post
Share on other sites

Your graphics card is what's killing that result. The GTX 1060 3GB is no where near enough to run VR smoothly even if Oculus claims otherwise. With the Ryzen 5 1600x at 4.0 GHz and an RX 480 8GB, I got a minimum result of 43 fps and an average of 44 fps (max was around 50-ish I think). So your GPU is holding you back badly. You don't need a 1080 TI for BoS in VR, but I strongly suggest a GTX 1070 TI or 1080.

Share this post


Link to post
Share on other sites

Thanks - I thought that might be the case. Yes, the 1080Ti is a bit of overkill but figured it should give me at least a couple of years of good use!

Share this post


Link to post
Share on other sites

That's the choice I had when moving to VR. My RX 480 wasn't enough for BoS or DCS, so I could do a 1070 and get it mostly there but then want to buy a 2070 the next year. Or I could just get a 1080 TI and wait for the GTX 3080 or whatever.

Share this post


Link to post
Share on other sites

Thanks - I thought that might be the case. Yes, the 1080Ti is a bit of overkill but figured it should give me at least a couple of years of good use!

no way overkill, more like perfect!

Share this post


Link to post
Share on other sites

Hi

 

Thank you for answering. I installed a 1080 ti and it took the average fps from 61 to 72. However still not good enough for playing at high settings, as you indicated. I have to look into the whole OC thing again.

 

Question though, with respect to the spread sheet: you report 2 different single thread numbers before you OC, 2153 and 2354 (user vs. public). How are they derived?

 

Hi again dear IL2 players

 

 

So, playing a different kind of game, I found out that my DDR4 was running at the wrong frequency. Stock should have been 2666 however BIOS indicated that it was running at 21xx something. I enabled the correct mem frequency and the benchmark result is the following:

 

Frames: 4774 - Time: 60000ms - Avg: 79.567 - Min: 46 - Max: 90

 

In short, correcting the mem frequency in BIOS yielded a similar boost as compared to swapping the 1070 to a 1080ti...isn't life grand?

 

 

    Cheers / Dr Fortran

Share this post


Link to post
Share on other sites

Indeed - I was surprised by the early results indicating how much of an impact RAM speed had on results, but it was confirmed when I did my new build with higher speed RAM.  This benchmark thread has proven very useful in aiding decisions about upgrading HW for improvement in VR experience with BoX.

Share this post


Link to post
Share on other sites

 

 

Still working on the optimal setting for my cooler. The 7900X is really a piece of work. Once I have that, I will run the VR benchmark to add to Chilis collection

 

And expensive one. It would be certainly good to have the first test results for the 7900X beast and allow other people to understand that if they are only interested in having good VR IL-2 fps there are other cheaper CPUs that achieve good fps.

I understand that you use your CPU for other real multithreads tasks.

Share this post


Link to post
Share on other sites

 

 

Yes, I set it in the OTT...

 

Very interesting to know your test results for two reasons: your CPU and your GPU (1060).

I think that a 1060 should be quite OK to run BOS in VR, but still no evidence since there is not a single test with 1060 in VR.

Samuel did various tests with monitor with different GPUs (1060, 1070, 1080, 1080Ti) with no SS, and the avg results were not too different (look at "Monitor Only" tab in the google sheet).

 

So, let´s see if we all can help you to discover your issue before you replace your GPU by the 1080Ti.

 

Firstly, please report your CPU Passmark and Single-Thread Passmark as indicated in the instructions of first post. With these we will know if your CPU is working as it should.

With Passmark you can also benchmark your CPU, RAM, GPU, etc to diagnose if some of your components is not performing as expected (the expected is thousands of user tests with same components like yours).

 

I also see that you use OTT. So, be sure you don´t use both (OTT and SteamVR) to set SS.

 

For the VR test you should use:

 

OTT --> 1.3

SteamVR --> 1.7 

 

Apart from this I will try to compile a number of checkpoints to verify when your fps are not as expected and post it in the first post. So, it will help you and many other people.

Share this post


Link to post
Share on other sites

 

 

So, playing a different kind of game, I found out that my DDR4 was running at the wrong frequency. Stock should have been 2666 however BIOS indicated that it was running at 21xx something. I enabled the correct mem frequency and the benchmark result is the following:   Frames: 4774 - Time: 60000ms - Avg: 79.567 - Min: 46 - Max: 90

 

Glad to see you increased your performance. Thanks for reporting.

 

In your previous test you reported 61 fps avg  (https://forum.il2sturmovik.com/topic/29322-measuring-rig-performance-common-baseline/?p=544157 ).

 

You didn´t use OC in that test (but according to your STMark of 2480 you were using TurboBoost mode, so your CPU should be 4.2)

Now you achieve 79.5 fps avg. This is +18fps. It seems a lot for a single RAM speed change. I understand that you use also OC. If this is the case, please report your OC and the Passmark (CPU and Singlethread) as indicated in the instructions.

Share this post


Link to post
Share on other sites

13282 overall CPU, single thread: 2743

 

1080GTX FE +10%power, +200 Mhz Gpu and +215 Mem Clock

 

Frames, Time (ms), Min, Max, Avg

5115, 60000, 58, 91, 85.250

 

Frames, Time (ms), Min, Max, Avg

5152, 60000, 62, 91, 85.867

have to scrap this result chili...my ram will not hold 3000Mhz. went down to 2800Mhz and no issues so far. will retest. no probs with Aida64 stress test but game will crash and bsod

Share this post


Link to post
Share on other sites

hmm thats interesting for a "Single Core" Game

IL2 Manual Forced to run single Core only. (Affinity- Core 0)
Took a while to load (100% core usage)
Stuka was all black cockpit,
Canopy was transparent and i could see out the back of plane.
Frames Down to 11fps from 100fps.

1037kv5.jpg



After about 1 minute it loaded in Cockpit Textures all Blurry and fps slowly crept upto 50fps.

2jb0coo.jpg


------------------------------------------------------------------------------------------------------------------------
Retested with 2 cores (Affinity Core 0,2)
Instant Load in No issues Full FPS

Share this post


Link to post
Share on other sites

Nobody said the game was single threaded. The game is multi-threaded, but the load isn't evenly balanced. This causes single-core performance to be the limiting factor.

  • Upvote 1

Share this post


Link to post
Share on other sites

Glad to see you increased your performance. Thanks for reporting.

 

In your previous test you reported 61 fps avg  (https://forum.il2sturmovik.com/topic/29322-measuring-rig-performance-common-baseline/?p=544157 ).

 

You didn´t use OC in that test (but according to your STMark of 2480 you were using TurboBoost mode, so your CPU should be 4.2)

Now you achieve 79.5 fps avg. This is +18fps. It seems a lot for a single RAM speed change. I understand that you use also OC. If this is the case, please report your OC and the Passmark (CPU and Singlethread) as indicated in the instructions.

 

Hi

 

The average of 61 was achieved before I installed the 1080ti. With the 1080ti I got 72 as average. Thus, setting the correct mem freq. only added 8 fps to the average.

 

I ran the passmark again and got the following results:

Passmark CPU Test: 12152, Single Thread 2481

 

I assume that this suggests that I was running @ 4.2GHz for the previous passmark as well? I had no idea that the CPU was running at 4.2 GHz instead of 4.0 GHz.

 

 

    / Fortran 

 

Edit: I'm on an HTC Vive (re-projections turned off during the benchmark)

Edited by Fortran

Share this post


Link to post
Share on other sites

updated and stable @ 4.6Ghz  dram 2800Mhz (no longer 4.7)

1080GTX FE +10%power, +200 Mhz Gpu and +215 Mem Clock

cpu single: 2706

cpu mark: 13236 

 

I believe my tinkering has come to a close. Fun times. Thanks again for the idea

 

Frames, Time (ms), Min, Max, Avg
  4996,     60000,  53,  91, 83.267

 

Frames, Time (ms), Min, Max, Avg
  5005,     60000,  54,  91, 83.417

Edited by katdog5

Share this post


Link to post
Share on other sites

 

 

The average of 61 was achieved before I installed the 1080ti. With the 1080ti I got 72 as average. Thus, setting the correct mem freq. only added 8 fps to the average.

 

OK, that explains better everything.

 

You can verify in the BIOS that you have the TurboBoost mode on. So you can enjoy this 4.2 GHz.

In all your tests, you were most likely running with 4.2. You can verify that with CPU-Z free tool while running test.

 

Now, your next thing could be doing moderate OC. Did you consider it?

Share this post


Link to post
Share on other sites

 

 

I believe my tinkering has come to a close.

 

Thanks for reporting your final tests. Tinkering is fun but more fun is to play BOS. Once you have done everything in your hand to squeze your hardware to max,  then you play play more comfortably, with your conscious in calm. Don´t forget to switch in game fps counter off!  :salute:

Share this post


Link to post
Share on other sites

Very interesting to know your test results for two reasons: your CPU and your GPU (1060).

I think that a 1060 should be quite OK to run BOS in VR, but still no evidence since there is not a single test with 1060 in VR.

Samuel did various tests with monitor with different GPUs (1060, 1070, 1080, 1080Ti) with no SS, and the avg results were not too different (look at "Monitor Only" tab in the google sheet).

 

So, let´s see if we all can help you to discover your issue before you replace your GPU by the 1080Ti.

 

Firstly, please report your CPU Passmark and Single-Thread Passmark as indicated in the instructions of first post. With these we will know if your CPU is working as it should.

With Passmark you can also benchmark your CPU, RAM, GPU, etc to diagnose if some of your components is not performing as expected (the expected is thousands of user tests with same components like yours).

 

I also see that you use OTT. So, be sure you don´t use both (OTT and SteamVR) to set SS.

 

For the VR test you should use:

 

OTT --> 1.3

SteamVR --> 1.7 

 

Apart from this I will try to compile a number of checkpoints to verify when your fps are not as expected and post it in the first post. So, it will help you and many other people.

 

Hi chili,

 

My CPU marks are as follows:

 

CPU: 13644 Single-thread: 2524

 

So this seems to be about right for the i5-8600K at stock clocks (I only have air cooling currently, so don't want to run it hot)

 

I'll see if I can run Passmark on the RAM too (although BIOS reports it at 3200Mhz and I've not seen anything to suggest it's not running at that speed)

 

As BeastyBaiter said, I guess it's likely that the 3GB of RAM on the 1060 I have is simply not enough.

 

Anyway, the 1080Ti is on it's way (whoop whoop!)  :biggrin:  but hopefully I can get some tests in before I install it to help out others...

 

EDIT1: OK, so RAM result at 3388 in Passmark (99th percentile - pretty good I'd say)

 

EDIT2: 3D GPU Mark is 9089 (93rd percentile - again, this is OK I think)

 

EDIT3: Disk  Mark is 3285 (74th percentile - not so great, but shouldn't affect the benchmark too much - I'm running my OS on an M.2 Samsung but IL-2 is installed on a separate SATA SSD so the tests are run against the SATA SSD)

Edited by wheeliemonsta

Share this post


Link to post
Share on other sites

Hi guys

 

A long time flight simmer (20+ years) but haven't had the equipment during the last 5 years to run anything than older games (IL-2 '46 for instance). Finally came around and bought a high end gaming PC with Oculus Rift so I could play this game and DCS (woho!) 

 

Anyway, here's my results:

 

CPU: 13007 Single thread: 2862

 

Frames: 4606 - Avg: 76.767 - Min: 44 - Max: 91

 

In hindsight, maybe I should have chipped in another 100 EUR for faster RAM?  :( Should I try to overclock the RAM?

Edited by frosen

Share this post


Link to post
Share on other sites

RAM speed doesn't impact most games but BoX seems to care about it a fair bit. Either way, nice system.

Share this post


Link to post
Share on other sites

Sorry for long, multiquote post but here it comes!!

 

Hi guys,

 

Here are my results (seem very low but I guess maybe that's the 1060 3GB more than anything else)?

 

CPU: i5-8600K stock clocks

User Passmark: CPU: 13644 Single-thread: 2524

16GB DDR4-3200

GTX1060 3GB

Rift

Frames: 2388

Min: 22

Max: 44

Avg: 39.800

 

MSI Z370 SLI PLUS m/b

 

These scores make mine pretty much the lowest on the board but I feel that shouldn't really be the case with the CPU/Mem combo I've got there? Any advice?

I've pushed the clocks up to 4.7 but it didn't make that much difference. I generally play in VR on Balanced with ASW on (as I need it!) and most everything else (Grass, Terrain, Mirrors, etc. at Low-ish.)

 

 

Hi chili,

 

My CPU marks are as follows:

 

CPU: 13644 Single-thread: 2524

 

So this seems to be about right for the i5-8600K at stock clocks (I only have air cooling currently, so don't want to run it hot)

 

I'll see if I can run Passmark on the RAM too (although BIOS reports it at 3200Mhz and I've not seen anything to suggest it's not running at that speed)

 

As BeastyBaiter said, I guess it's likely that the 3GB of RAM on the 1060 I have is simply not enough.

 

Anyway, the 1080Ti is on it's way (whoop whoop!)  :biggrin:  but hopefully I can get some tests in before I install it to help out others...

 

EDIT1: OK, so RAM result at 3388 in Passmark (99th percentile - pretty good I'd say)

 

EDIT2: 3D GPU Mark is 9089 (93rd percentile - again, this is OK I think)

 

EDIT3: Disk  Mark is 3285 (74th percentile - not so great, but shouldn't affect the benchmark too much - I'm running my OS on an M.2 Samsung but IL-2 is installed on a separate SATA SSD so the tests are run against the SATA SSD)

 

Hi guys - back again!

Don't know what was going on with my system but just ran another couple of times through the test. This time I'm running a mild overclock to 4.3GHz

 

First time I got this:

 

Frames: 2724

Avg: 45.4

Min: 43

Max: 79

 

Then I plotted those out on a graph and it looked like ASW was on, even though in OTT I'd turned it off. So I "re-selected Off" in OTT (and put the Oculus HUD overlay on to check) and this time it actually turned off!!

So I ran again and got these numbers:

 

Frames: 4287

Avg: 71.45

Min: 44

Max: 91

 

I'm still not convinced that SS was at 1.3 as when I put the Pixel Density Oculus overlay on, it suggested it was at 1.0 but hey, I'm much happier with those numbers!!

 

If I could just figure out a way of getting the Ctrl-2 thing to work (lock at 45fps with ASW off) I'd be very happy as I don't have much of an issue with running at 45 rather than 90... especially if I can run at High graphics settings, etc.

This (Ctrl-2) just doesn't seem to work for me. I tried holding it down for some time but nothing and the OTT locks at 45fps with ASW permanently ON so I get the prop artifacts...

I'm running the Oculus Home 2.0 beta - maybe this is interfering with the Ctrl-2 thing (or maybe I've misunderstood??)

 

Edited by wheeliemonsta

Share this post


Link to post
Share on other sites

RAM speed doesn't impact most games but BoX seems to care about it a fair bit. Either way, nice system.

 

Thanks. I bumped up the RAM to 2800 just for the sake of it. Also I noticed that I missed to turn off the target fps setting in the game so when I did these two things my average went up to 80 sharp! I'm quite happy with that.

Share this post


Link to post
Share on other sites

Hi everyone,

 

since VR performance of this game is a bit curious, I am doing my first posting in here. Since OP is unclear on how to set reprojection in SteamVR, I did two runs. Async. and interleaved reprojection set to off, Always-On Reprojection enabled on the first run and disabled on the second run:

 

2.012d 1,7 B1rdy Asrock X99M Killer/3.1 i7-6800K 3,4 4,2     16568 2453 DDR4 32 3200 1080Ti Vive 5279 75 91 87,983 68,796 19,2         VGA: Core 1911-1949Mhz, Mem: 5900Mhz; A-O Repro: On

2.012d 1,7 B1rdy Asrock X99M Killer/3.1 i7-6800K 3,4 4,2     16568 2453 DDR4 32 3200 1080Ti Vive 4869 44 90 81,15 68,796 12,4         VGA: Core 1911-1949Mhz, Mem: 5900Mhz; A-O Repro: Off

 

Always-On Reprojection provides a rather significant boost. I verified it by another bench run.

 

@Memory speeds: The benefits of increased speed are always a bit depended on the application. In general more speed is more beneficial than lower timings for games. However since most games are limited by GPU performance (even more the higher the resolution is), CPU performance, i.e. memory speed, becomes negligible. This is not the case in IL-2, since it places heavy load on one core/ thread. That seems to limit the overall performance. You can check that by decreasing resolution only to 1280x720, IL-2 won't gain much performance (if your GPU isn't a potato). Well multithreaded engines like the one from Elite: Dangerous or the recent Wolfenstein, will reach 300 Fps and more under those conditions.

 

For those of you running Ryzen 7 CPUs, you might want to try disabling SMT. Some games have proven to loose performance due to SMT and more than 12 Threads.

 

What can I do to increase Fps at ground level?

 

 

Edit: The table got nuked. :-( To be clear, I did the benches only with OC.

Edited by B1rdy

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×
×
  • Create New...