TG-55Panthercules Posted January 15, 2023 Posted January 15, 2023 Found myself with a little time on my hands last night (wanted to take a break from creating my personal skins for the FC planes - something I forgot I had never gotten around to doing for some reason), so I decided to pull the trigger and upgrade my new PC to Windows 11. Haven't noticed any super bad results since doing that (everything I've tried to launch so far seems to run OK, although apparently my Saitek throttle and rudder drivers seem to conflict with some sort of memory protection processes which have to be switched off as a result), but I've been rather disappointed in performance on the benchmarks I've run so far. Both of the free 3D Mark benchmarks (Firestrike and Time Spy) showed a small reduction, and the IL-2 SYNVander v6 benchmark for 1080p monitor also saw a bit of a decline (from Avg: 137.717 - Min: 110 - Max: 190 on Win10 to Avg: 126.300 - Min: 105 - Max: 147 on Win11). Unfortunately, when I tried to run this benchmark the game updated itself to version 5.004, so it's a bit apples and oranges since the Win10 run was still on game version 5.003. Haven't had a chance to run the VR benchmarks yet - will probably try that later tonight.
chiliwili69 Posted January 15, 2023 Author Posted January 15, 2023 8 hours ago, TG-55Panthercules said: (from Avg: 137.717 - Min: 110 - Max: 190 on Win10 to Avg: 126.300 - Min: 105 - Max: 147 on Win11). This is a significant drop. In previous post, If I remember correctly, the impact of Windows11 was minimal. Also, There are no major changes from 5.003 to 5.004 for the game engine, so I think results should be comparable. I can run also the test to confirm that.
chiliwili69 Posted January 18, 2023 Author Posted January 18, 2023 On 1/15/2023 at 1:40 AM, TG-55Panthercules said: IL-2 SYNVander v6 benchmark for 1080p monitor also saw a bit of a decline (from Avg: 137.717 - Min: 110 - Max: 190 on Win10 to Avg: 126.300 - Min: 105 - Max: 147 on Win11). Unfortunately, when I tried to run this benchmark the game updated itself to version 5.004 I have been testing the IL-2 version v5.004b and these are the results: CPU test Frames: 6337 - Time: 60000ms - Avg: 105.617 - Min: 87 - Max: 158 Frames: 6305 - Time: 60000ms - Avg: 105.083 - Min: 88 - Max: 137 Frames: 6364 - Time: 60000ms - Avg: 106.067 - Min: 89 - Max: 147 GPU 4K test Frames: 7428 - Time: 60000ms - Avg: 123.800 - Min: 93 - Max: 140 Frames: 7415 - Time: 60000ms - Avg: 123.583 - Min: 100 - Max: 140 Frames: 7403 - Time: 60000ms - Avg: 123.383 - Min: 96 - Max: 139 VR1 test Frames: 2692 - Time: 60000ms - Avg: 44.867 - Min: 42 - Max: 47 Frames: 2693 - Time: 60000ms - Avg: 44.883 - Min: 42 - Max: 46 Frames: 2718 - Time: 60000ms - Avg: 45.300 - Min: 41 - Max: 53 On the CPU test I had 114.5 fps with version 5.002b, so about 9fps less. Exactly what you had! The GPU 4K is not affected The VR1 test I had 61.6 fps with version 5.002b, HERE I HAVE 16fps LESS!!!! MY GOD!!! So YES!, there is big difference on the latest version of IL-2. I have separated in the table all test from version 5.004. What the hell has been introduced in the IL-2 engine to have such a big drop in VR?? is it only me? Please, try to run the VR test as well. 1
Norz Posted January 18, 2023 Posted January 18, 2023 27 minutes ago, chiliwili69 said: So YES!, there is big difference on the latest version of IL-2. I have separated in the table all test from version 5.004. What the hell has been introduced in the IL-2 engine to have such a big drop in VR?? is it only me? Please, try to run the VR test as well. Oh...sad.... I am going to use my Quest2 and 16 frames are a lot:(
dgiatr Posted January 18, 2023 Posted January 18, 2023 (edited) On 9/27/2022 at 5:32 PM, dgiatr said: So here are my vr1 test and vr2 test with steam vr vr1 test 2022-09-27 16:51:04 - Il-2 Frames: 3559 - Time: 60000ms - Avg: 59.317 - Min: 49 - Max: 91 2022-09-27 16:53:17 - Il-2 Frames: 3361 - Time: 60000ms - Avg: 56.017 - Min: 46 - Max: 82 2022-09-27 16:55:47 - Il-2 Frames: 3468 - Time: 60000ms - Avg: 57.800 - Min: 48 - Max: 90 vr2 test 2022-09-27 17:14:00 - Il-2 Frames: 3101 - Time: 60000ms - Avg: 51.683 - Min: 43 - Max: 72 2022-09-27 17:18:27 - Il-2 Frames: 3424 - Time: 60000ms - Avg: 57.067 - Min: 48 - Max: 91 2022-09-27 17:20:22 - Il-2 Frames: 3409 - Time: 60000ms - Avg: 56.817 - Min: 47 - Max: 91 2022-09-27 17:25:50 - Il-2 Frames: 3362 - Time: 60000ms - Avg: 56.033 - Min: 47 - Max: 89 ......i guess i am cpu bottlenecked, is that right? during vr1 and vr2 tests while using FPSVR i see my CPU FRAMETIME MORE THAN 11.1 ms most of the time... vr test 1 2022-12-18 12:52:51 - Il-2 Frames: 2783 - Time: 60000ms - Avg: 46.383 - Min: 40 - Max: 56 2022-12-18 13:27:28 - Il-2 Frames: 2737 - Time: 60000ms - Avg: 45.617 - Min: 39 - Max: 53 .....i thought i have done something wrong during vr test1 that's why i didn't say anything.......YES i have about 57-46=11 fps performance loses , probably i dont have 16 fps loses like you do because in the meantime i overclocked my cpu and ram... Edited January 18, 2023 by dgiatr 1
chiliwili69 Posted January 18, 2023 Author Posted January 18, 2023 1 hour ago, dgiatr said: 57-46=11 fps performance loses , probably i dont have 16 fps loses like you do because in the meantime i overclocked my cpu and ram Thank you for this! It confirms the drop as well. Have a change to run CPU test? (just for table update) The 16fps or 11fps drop in reality is the same. I went from 61 to 45, And you from 57 to 46. We both toauch base at 45fps. The 45fps limit acts as "discountinuity" in the performance. This 45fps discontinuity wall effect is explained here.
102nd-YU-cmirko Posted January 18, 2023 Posted January 18, 2023 (edited) 5800X3D, 32gb DDR4-3800 CL16, FCLK 1900, 6900XT, latest VD and all drivers tested latest update for comparison, I wonder did I make a mistake somewhere in setup for VR1 test original api.dll, VR1 test, pico4 with 2160x2160 (48%) in SteamVR, encoded to 3120x3120 2023-01-18 14:24:10 - il-2 Frames: 3989 - Time: 60000ms - Avg: 66.483 - Min: 58 - Max: 85 opencomposite api.dll, VR1 test, openxrtoolkit at 2160x2160 and no postprocessing of any kind, , encoded to 3120x3120 2023-01-18 14:55:03 - Il-2 Frames: 3984 - Time: 60000ms - Avg: 66.400 - Min: 56 - Max: 85 Edited January 18, 2023 by 102nd-YU-cmirko
chiliwili69 Posted January 18, 2023 Author Posted January 18, 2023 4 hours ago, 102nd-YU-cmirko said: Frames: 3989 - Time: 60000ms - Avg: 66.483 - Min: 58 - Max: 85 Thank you for this. What is your Mobo? do you run Windows11? It is the first Pico4 reported here! If you have time please, run the CPU test in monitor just for comparison. Seems the 5800X3D is performing really well comparing to my 5600X.
TG-55Panthercules Posted January 19, 2023 Posted January 19, 2023 (edited) On 1/15/2023 at 4:25 AM, chiliwili69 said: This is a significant drop. In previous post, If I remember correctly, the impact of Windows11 was minimal. Also, There are no major changes from 5.003 to 5.004 for the game engine, so I think results should be comparable. I can run also the test to confirm that. Finally got my G2 working again (the cable had come loose on the headset, though it didn't seem like that the first few times I checked it), so I was able to run the VR Option 1 test to see if I have the same kind of drop going to Windows 11 and game version 5.004 in VR as I saw in the monitor 1080P test. Sure enough, results went from Avg: 65.417 - Min: 55 - Max: 91 (win 10, ver. 5.003) to Avg: 58.700 - Min: 51 - Max: 76 (Win 11, ver. 5.004) on the VR Option 1 test. Kinda depressing OK - stupid noob Q - I went back in to run the benchmark again just to make sure I had all the settings right and everything, and now I have this dumb overlay that sits on top of IL-2 GB and I cannot remember or find the setting or method to turn the stupid thing off: What am I missing? Edited January 19, 2023 by TG-55Panthercules
dgiatr Posted January 19, 2023 Posted January 19, 2023 (edited) 19 hours ago, chiliwili69 said: Thank you for this! It confirms the drop as well. Have a change to run CPU test? (just for table update) just for the record....cpu 1080p test with previous il2 gb version during 20-9-2022 day... data are without cpu and ram overclocking then... On 9/20/2022 at 11:17 PM, dgiatr said: Hello to all! Just did 3 times the 1080p cpu test and my results are as below 2022-09-20 14:41:26 - Il-2 Frames: 5329 - Time: 60000ms - Avg: 88.817 - Min: 77 - Max: 129 2022-09-20 15:06:14 - Il-2 Frames: 5038 - Time: 60000ms - Avg: 83.967 - Min: 73 - Max: 115 2022-09-20 15:09:10 - Il-2 Frames: 5092 - Time: 60000ms - Avg: 84.867 - Min: 74 - Max: 127 my specs are Motherboard: Z590 AORUS ELITE AX CPU: intel core i5 10600k CPU Freq: 4.5 Ghz L3 cache: 12 MB Cores: 6 Threads: 12 RAM type: DDR4 RAM size: 16 GB NB Freq: 4300 MHz RAM Freq: 3200 MHz RAM TIMING: 14-14-14-34 GPU: MSI SUPRIM X 3090 Ti No overclocking , hyperthreading=on What do you think , Is that ok with that cpu? Is there anything else to do to get a better performance? ( dont want to OC my cpu , i am not still that expert and have enough heat inside my tower due to my GPU...) Setting hyperthreading to OFF will give me better fps? My cpu single threaded passmark score is 71% i will post other tests soon... ...just want to thank chiliwili69 for these tests!!!, today i discovered thanks to cpuid cpu-z that my RAM memory speed was not at 3200 mhz but way lower at about 2200 mhz!! so i had a total average of 86 fps then....and now with cpu and ram overclocked i get 99 fps average during 1080 cpu test, a difference of 99-86=13 fps, which i think it would be probably bigger if il2 was still in the same version. by the way...very informative link there : thank you for the excellent work you have done so far for our il2 gb vr community! Edited January 19, 2023 by dgiatr
chiliwili69 Posted January 19, 2023 Author Posted January 19, 2023 5 hours ago, TG-55Panthercules said: Avg: 58.700 - Min: 51 - Max: 76 (Win 11, ver. 5.004) on the VR Option 1 test. Kinda depressing Thanks for runnning VR test1, it confirms as well the drop of the v5.004 version. Remember that for this is a heavy mission with all IL-2 graphics option maxed-out to bottleneck CPU. In normal play game the shadows/mirror/canopyreflections/distantbuilding/etc can be relaxed to be at the 90fps most of the time. About that overlay it seems related to SteamVR settings. Look this: https://boxthislap.org/disable-home-and-layout-on-steamvr/ 38 minutes ago, dgiatr said: so i had a total average of 86 fps then....and now with cpu and ram overclocked i get 99 fps average during 1080 cpu test, a difference of 99-86=13 fps, which i think it would be probably bigger if il2 was still in the same version. OK, it is clear now. I have updated the table with your VRtest1 result for the CPU/RAM OC tests. Thanks to you!
102nd-YU-cmirko Posted January 19, 2023 Posted January 19, 2023 16 hours ago, chiliwili69 said: Thank you for this. What is your Mobo? do you run Windows11? It is the first Pico4 reported here! If you have time please, run the CPU test in monitor just for comparison. Seems the 5800X3D is performing really well comparing to my 5600X. windows 10 22H2 19045.2486 (latest updates), x570 Aorus Pro, 4 sticks of ram at 1900Mhz and 16/16/16/36/87 1T CPU test 2023-01-19 12:46:59 - Il-2 Frames: 6611 - Time: 60000ms - Avg: 110.183 - Min: 90 - Max: 158 2023-01-19 12:49:20 - Il-2 Frames: 6612 - Time: 60000ms - Avg: 110.200 - Min: 90 - Max: 152 1
TG-55Panthercules Posted January 19, 2023 Posted January 19, 2023 8 hours ago, chiliwili69 said: Thanks for runnning VR test1, it confirms as well the drop of the v5.004 version. Remember that for this is a heavy mission with all IL-2 graphics option maxed-out to bottleneck CPU. In normal play game the shadows/mirror/canopyreflections/distantbuilding/etc can be relaxed to be at the 90fps most of the time. About that overlay it seems related to SteamVR settings. Look this: https://boxthislap.org/disable-home-and-layout-on-steamvr/ Thanks - that seems to have done the trick and gotten rid of the annoying overlay. I had it turned off in one of the obvious places, but there was a second more obscure place that I had missed. I ran the VR Option 1 benchmark several more times, and given the amount of fluctuation between some of those runs I'm not sure how significant the apparent drop in performance I was seeing between Win10/5.003 and Win11/5.004 really is. The average of the 4 runs I did on Win11/5.004 was still lower than my previous run on Win10/5.003, but I didn't do enough runs on Win10/5.003 to get an average, and the highest run of the 4 on Win11/5.004 wasn't very much lower than the one run I did on Win10/5.003, so I'd have to take my results with a fairly large grain of salt at this point. I took another look at trying to OC my CPU in the BIOS for this new PC, but it is so much more extensive and complicated than I remember things being with my old i7-7700K that I decided just to leave everything on the BIOS defaults. IIRC, in addition to the i7-7700K's BIOS being much simpler there was also some sort of simple software app that made overclocking very easy and straightforward, but that was several years ago and I don't remember which app I was using at that time. Is there something like that available now that works with the Intel Core i9 12900K and Windows 11 that would make sense to try for overclocking at this point?
shirazjohn Posted January 21, 2023 Posted January 21, 2023 Hi Guys I've ran the test with the latest game version but I am not seeing the losses that others are noticing. The first results above the line are my current settings with Hud off and -30 all core curve optimiser, power limits ppt 125/ Tdc 75/ Edc 115 (the latest bios has unlocked pbo settings for the 5800x3d). The second test below the line are as my previous test pbo disabled and Hud on. My tests are similar to the previous test except for vr test2 which is lower. Curve all core -30 power limits ppt125/tdc75/edc115 Cpu Test 2023-01-20 15:34:46 - Il-2 Frames: 7943 - Time: 60000ms - Avg: 132.383 - Min: 111 - Max: 178 2023-01-20 15:36:41 - Il-2 Frames: 7953 - Time: 60000ms - Avg: 132.550 - Min: 112 - Max: 177 Vr test 1 50% 2023-01-20 15:51:40 - Il-2 Frames: 4268 - Time: 60000ms - Avg: 71.133 - Min: 58 - Max: 91 Vr test 2 100% 2023-01-20 15:56:47 - Il-2 Frames: 4211 - Time: 60000ms - Avg: 70.183 - Min: 57 - Max: 91 Vr Test 3 2023-01-20 16:06:43 - Il-2 Frames: 2591 - Time: 60000ms - Avg: 43.183 - Min: 36 - Max: 48 ---------------------------------------------------------------------------------------------------------- Cpu test hud on and settings as per previous results 2023-01-20 16:22:13 - Il-2 Frames: 7713 - Time: 60000ms - Avg: 128.550 - Min: 109 - Max: 174 2023-01-20 16:24:36 - Il-2 Frames: 7693 - Time: 60000ms - Avg: 128.217 - Min: 107 - Max: 173 Vr Test 1 50% Hud on and settings as per previous results 2023-01-20 16:59:31 - Il-2 Frames: 4231 - Time: 60000ms - Avg: 70.517 - Min: 58 - Max: 91 Vr test 2 100% hud on and settings as per previous results 2023-01-20 16:42:33 - Il-2 Frames: 3899 - Time: 60000ms - Avg: 64.983 - Min: 53 - Max: 91
RAAF492SQNOz_Steve Posted January 22, 2023 Posted January 22, 2023 (edited) Results for VR2 scenario, Varjo Aero and Monitor 1080 testing CPU = 5800X3d, -30 all core undervolt but with standard power settings GPU= RTX 4090, Gigabyte Gaming OC with 450W power limit (only have three power cables so cannot run 600W even If I wanted to) VR Headset = Varjo Aero with Steam SS settings at 116%. Headset Varjo Base setting 35 ppd RAM = 32Gb G. Skill CL16 @3866 MHz IL2 version = v5.004 O/S = Windows 10 Median values of several test runs.1080P Monitor test Frames, Time (ms), Min, Max, Avg 7743, 60000, 105, 185, 129.050 VR2 Test, Steam SS @ 116% or 3380x2896 Frames, Time (ms), Min, Max, Avg 4873, 60000, 70, 91, 81.217NOTE: I was also running fpsVR at the same time as running these tests and even with the VR2 test saw times when the CPU was showing amber blocks/ extended spikes and very occasional Red spikes. Sometimes the CPU and GPU had amber blocks at the same time and other times only the CPU had them. This suggests, to me, that in the Benchmark fly thru there are times when I am CPU limited. The other unusual and rather surprising result of running fpsVR at the same time was that it gave me something to focus on during the fly thru and I got more consistent/ higher average fps results. Interesting...... Edited January 22, 2023 by RAAF492SQNOz_Steve Added IL2 version details
chiliwili69 Posted January 22, 2023 Author Posted January 22, 2023 21 hours ago, shirazjohn said: I've ran the test with the latest game version but I am not seeing the losses that others are noticing. Thank you for running again the tests. In your case not only don´t see any loss but a bit of a gain perhaps thanks to the new PBO. I happy to see those numbers, so it means one of these two things: - It is just something wrong in my PC (and some others) - The lattest release has something affecting the CPU with 32Mb L3 cache like mine (5600X).
chiliwili69 Posted January 22, 2023 Author Posted January 22, 2023 13 hours ago, RAAF492SQNOz_Steve said: VR2 Test, Steam SS @ 116% or 3380x2896 Frames, Time (ms), Min, Max, Avg 4873, 60000, 70, 91, 81.217 Thank you so much for making the tests. It is the first Aero tested here. And you obtained a very good performance. As you said, you were bottlenecked by the CPU in certain times. In fact I think that mostly is due to CPU. To prove that you only need to run the VRtest1 (where the GPU will be fully relaxed) and you will check that your result will not be very far from 81. You have a pretty good combo (5800X3D, 4090, Aero).
WheelwrightPL Posted January 22, 2023 Posted January 22, 2023 (edited) Motherboard: MSI B550-A PRO (MS-7C56) CPU: Ryzen 5800X with Ryzen Master Default Overclock CPU Freq: 3774 Mhz Cores: 8 Threads: 8 (I turned off the Hyperthreading) RAM: 32 GB RAM Freq: 3600 Mhz NB Freq: 1800 Mhz RAM timings: 18 20 20 42 GPU: 4070 Ti with default clocks (from Gigabyte) 1024px TESTS WITH HYPERTHREADING OFF TEST Run 1 Results: Frames, Time (ms), Min, Max, Avg 6982, 60000, 99, 160, 116.367 TEST Run 2 Results: Frames, Time (ms), Min, Max, Avg 6915, 60000, 97, 164, 115.250 1024px TESTS WITH HYPERTHREADING ON (the Default setting) TEST Run 1 Results: Frames, Time (ms), Min, Max, Avg 7301, 60000, 103, 172, 121.683 TEST Run 2 Results: Frames, Time (ms), Min, Max, Avg 7191, 60000, 99, 165, 119.850 4k TESTS WITH HYPERTHREADING OFF TEST Run 1 Results: Frames, Time (ms), Min, Max, Avg 5273, 60000, 74, 97, 87.883 TEST Run 2 Results: Frames, Time (ms), Min, Max, Avg 5290, 60000, 78, 99, 88.167 4k TESTS WITH HYPERTHREADING ON TEST Run 1 Results: Frames, Time (ms), Min, Max, Avg 5260, 60000, 74, 99, 87.667 TEST Run 2 Results: Frames, Time (ms), Min, Max, Avg 5284, 60000, 75, 99, 88.067 BTW: I ran all the tests with Fullscreen On (the Instructions called for "Fullscreen Off"), because I can't imagine anybody running Il-2 in Windowed (ie. Fullscreen Off) Mode. Not sure if that made any difference. EDIT: I just re-ran 4k test with Fullscreen Off and it made no difference. Edited January 22, 2023 by WheelwrightPL
RAAF492SQNOz_Steve Posted January 22, 2023 Posted January 22, 2023 (edited) 2 hours ago, chiliwili69 said: Thank you so much for making the tests. It is the first Aero tested here. And you obtained a very good performance. As you said, you were bottlenecked by the CPU in certain times. In fact I think that mostly is due to CPU. To prove that you only need to run the VRtest1 (where the GPU will be fully relaxed) and you will check that your result will not be very far from 81. You have a pretty good combo (5800X3D, 4090, Aero). One additional comment I should make and will be interested to see what other Varjo Aero owners results are..... As stated in my previous post, I did the VR2 tests at Steam 116% SS. I subsequently did a test at 100% SS and got a pretty much identical result. So, I am not sure how much attention is being paid to SS settings with the Varjo Aero. Changing Varjo Aero ppd (in Varjo Base) has an immediate and brutal impact on performance so ppd has a clear and obvious impact. May be useful to specify a ppd setting for Varjo Aero owners for VR testing. Perhaps 35 ppd? When running 116% SS and monitoring fpsVR resolution, it did show the higher resolution setting. So, until other Varjo Aero owners come in with some results, the jury is out on SS impacts on the Aero for VR tests. Edited January 22, 2023 by RAAF492SQNOz_Steve Typo's
WheelwrightPL Posted January 23, 2023 Posted January 23, 2023 Anybody has an idea why my results above with 4070 Ti are so poor in 4K ? My G3D PassMark Result is 35355 (the average for 4070 Ti is 30949).
chiliwili69 Posted January 23, 2023 Author Posted January 23, 2023 (edited) 10 hours ago, WheelwrightPL said: 4070 Ti with default clocks (from Gigabyte) Thank you for your tests! It is the first 4070Ti tested here. Regarding Hyperthreading (Ryzen SMT) ON vs OFF I did 10 tests each in the past with about +-5fps variability and I was concluding that there were no difference: https://forum.il2sturmovik.com/topic/66924-syn_vander-benchmark-v6-to-measure-il-2-performance-in-monitor-vr/?do=findComment&comment=1082531 Perhaps the BIOS has been changed since then or something changed in the IL-2 engine. Ideally you could run at least 5 test each to have more datapoints. Your 5800X is giving a good performance in the CPU test. So, it add more data to the idea that v5.004 has not degraded performance. Regarding your 4K tests, this is not aligned with what I expect from a 4070Ti. It should be well above what a 3080 would deliver. It is really weird. Edited January 23, 2023 by chiliwili69
chiliwili69 Posted January 23, 2023 Author Posted January 23, 2023 10 hours ago, RAAF492SQNOz_Steve said: So, I am not sure how much attention is being paid to SS settings with the Varjo Aero. Changing Varjo Aero ppd (in Varjo Base) has an immediate and brutal impact on performance so ppd has a clear and obvious impact. May be useful to specify a ppd setting for Varjo Aero owners for VR testing. Perhaps 35 ppd? Yes indeed. The Aero software settings regarding resolution has six modes (Highest, Very High, High, Medium, Low, very Low) and they change the internal supersampling ratio which every headset in the market is using. In the same spreasheet of the SYN_Vander results you will see another tab where all the internal SS ratios are calculated. In particular to the Aero they are: As you followed well you used well the High Settings mode to run the VRTest2. The first post instructions say: VRTest1: Varjo Aero SS=56% 2350x2015=9.46 Mpixels with High Settings (default) in Varjo base software VRtest2: Varjo Aero SS=116% 3382x2899=19.61 Mpixels with High Settings (default) in Varjo base software Initially, when Aero was launched, the High Settings mode was the 27 ppd, look here, (in fact the webpage is still like that), but you showed me that the High Settings mode referes to 35ppd, look here. So, to avoid confusion I just refer them as High Settings in the instructions. In fact, anyone can check that the number of pixels for 116% is around 3382x2899. Now, regarding the influence of SS over the performance of the Aero (or any other headset) it depends basically from the total number of pixels and the capability of the GPU. For the 4090, it can handle easily 19.5 million pixels at 90fps, so it will not be the bottleneck in VR tests. And any VR test will less than 19.5 will give esentially the same Avg fps (81 in your case, which is determined by the CPU). If you increment the SS gradually from 116% you will see that at some point your are starting to be below 81. This is the point where the 4090 will start to be fully loaded and will start to became the bottleneck. We really don´t know where the 4090 has the limit in SS for the VRtest1&2 settings (apart from SS). We tried to increase SS using other devices but we faced one limit of SteramVR for certain devices. Look here. In the case of the Aero, this limit is quite high, so it is not a problem and you can increase the SS as much as you want using the setting of the VRtest1&2 until you see an Avg fps drop in fps (from 81).
1Sascha Posted January 23, 2023 Posted January 23, 2023 (edited) Just found out that you guys created a spreadsheet with all the results... very nice! I wonder if I should re-run my tests, since I just upgraded my system from 2x8 GB to 4x8 GB of RAM. Still the same RAM speed and CL (3600, 18), so I'm not sure if it would make a huge difference and if I should even bother. Time Spy and Passmark yielded slightly higher overall scores, but, naturally, no spectacular gains... so...? S. Edited May 9, 2023 by 1Sascha
chiliwili69 Posted January 23, 2023 Author Posted January 23, 2023 8 hours ago, 1Sascha said: Just found out that you guy's created a spreadsheet with all the results... very nice! LOL! Yes, that was the purpose of the test to record those tests so we can compare them ? I just put the link of the spreadsheet at the top of the instructions as well. You are always welcome to report here your test (CPU and VR). 1
chiliwili69 Posted January 23, 2023 Author Posted January 23, 2023 19 hours ago, WheelwrightPL said: Anybody has an idea why my results above with 4070 Ti are so poor in 4K ? Could the memory interface width be the explanation for IL2 in 4K? (in other games the 4070Ti is above the 3080) It could be just he way this 4K test is designed, specially tyhe MSAAx8. Have you tried in VR or with other MSAA values? I can run the 4K bench with other settings.(without MSAA).
WheelwrightPL Posted January 23, 2023 Posted January 23, 2023 1 hour ago, chiliwili69 said: Could the memory interface width be the explanation for IL2 in 4K? (in other games the 4070Ti is above the 3080) It could be just he way this 4K test is designed, specially tyhe MSAAx8. Have you tried in VR or with other MSAA values? I can run the 4K bench with other settings.(without MSAA). I had similarly poor results with 2080 Ti, I think my Il-2 config may be screwed up, because other games and benchmarks work fine.
RossMarBow Posted January 24, 2023 Posted January 24, 2023 I expected the 4070 ti to have these issues. The ram bus is lacking severely.
1Sascha Posted January 24, 2023 Posted January 24, 2023 (edited) 4 hours ago, RossMarBow said: The ram bus is lacking severely. Yup... it seems a bit of a joke when even my puny 3070 and the 3060 Ti have wider memory busses (256 bit). Came across this 3090 vs 4070 Ti benchmark comparison and while the 4070 Ti looks very good in most of the games tested, it's still inferior in MSFS - and I'm not sure that's down to the 4070 Ti "only" having 12 GB of VRAM as some have suggested. That game will only use slightly above 12 GB on the 3090. Looking at the CPU load in both configs, it does seem like the 4070 Ti equipped system is GPU bottlenecked quite heavily. https://www.youtube.com/watch?v=3VLsLNkuK0E ^ Video doesn't allow embedding for some reason. S. Edited January 24, 2023 by 1Sascha
chiliwili69 Posted January 24, 2023 Author Posted January 24, 2023 17 hours ago, WheelwrightPL said: I had similarly poor results with 2080 Ti, I think my Il-2 config may be screwed up Yes, that´s true. Your previous test in the table with your previous 2080Ti was also quite below expected. So, the issue is not in the card. Just review GPU load, temps, amperage, power of the GPU while running the bench. Or reset to default the NVIDIA or IL-2 config files
dgiatr Posted January 24, 2023 Posted January 24, 2023 (edited) New cpu 1080p test with a new i9 13900kf cpu stock settings no O.C yet, 5.5 ghz Os win 10 Ram 16 gb ddr4 3200 mhz xmp profile 14 14 14 32 Asus z690 p d4 Msi suprim x 3090 ti 2023-01-24 18:57:22 - Il-2 Frames: 9441 - Time: 60000ms - Avg: 157.350 - Min: 136 - Max: 203 2023-01-24 18:59:34 - Il-2 Frames: 9183 - Time: 60000ms - Avg: 153.050 - Min: 130 - Max: 200 2023-01-24 19:01:35 - Il-2 Frames: 9203 - Time: 60000ms - Avg: 153.383 - Min: 135 - Max: 201 Edited January 24, 2023 by dgiatr
dgiatr Posted January 24, 2023 Posted January 24, 2023 I want to thank DBCOOPER011 for his useful advices about mobo choice and to stay for now with DDR4 memory sticks and mobo, since as 1080 cpu test indicates i can achieve more than 150 fps average with a good and relatively cheap pair of ddr4 memory sticks 3 1
DBCOOPER011 Posted January 25, 2023 Posted January 25, 2023 5 hours ago, dgiatr said: I want to thank DBCOOPER011 for his useful advices about mobo choice and to stay for now with DDR4 memory sticks and mobo, since as 1080 cpu test indicates i can achieve more than 150 fps average with a good and relatively cheap pair of ddr4 memory sticks That's great it worked out well for you! It also shows you can get pretty good performance without the top of line motherboards and/or ram..
WheelwrightPL Posted January 25, 2023 Posted January 25, 2023 10 hours ago, chiliwili69 said: Yes, that´s true. Your previous test in the table with your previous 2080Ti was also quite below expected. So, the issue is not in the card. Just review GPU load, temps, amperage, power of the GPU while running the bench. Or reset to default the NVIDIA or IL-2 config files I took your advice and re-installed the damn thing (there was filesystem blue-screen-of-death error during the uninstall). Here are the updated results in 4k: Motherboard: MSI B550-A PRO (MS-7C56) CPU: Ryzen 5800X with Ryzen Master Default Overclock CPU Freq: 3774 Mhz Cores: 8 Threads: 16 RAM: 32 GB RAM Freq: 3600 Mhz NB Freq: 1800 Mhz RAM timings: 18 20 20 42 GPU: 4070 Ti with default clocks (from Gigabyte) 4k TESTS WITH HYPERTHREADING ON (Default) TEST Run 1 Results: Frames, Time (ms), Min, Max, Avg 7086, 60000, 96, 134, 118.100 TEST Run 2 Results: Frames, Time (ms), Min, Max, Avg 7101, 60000, 103, 133, 118.350 TEST Run 3 Results: Frames, Time (ms), Min, Max, Avg 7118, 60000, 97, 131, 118.633
chiliwili69 Posted January 26, 2023 Author Posted January 26, 2023 On 1/24/2023 at 6:40 PM, dgiatr said: 13900kf Nice! The 13900K equioed with a good DDR4 RAM is as good as DDR5 at higher freqs. Very good upgrade,
dgiatr Posted January 26, 2023 Posted January 26, 2023 7 hours ago, chiliwili69 said: Nice! The 13900K equioed with a good DDR4 RAM is as good as DDR5 at higher freqs. Very good upgrade, Hello chiliwili69, just checked your IL2 VR spreadsheet here : https://docs.google.com/spreadsheets/d/1gJmnz_nVxI6_dG_UYNCCpZVK2-f8NBy-y1gia77Hu_k/edit#gid=1266758920 i have two corrections to tell : 1. i haven't done any CPU or RAM OC yet with my new cpu+mobo combination 2. i haven't done any VRTEST1 yet , so you can leave those cells empty if you want. Thank you,
dgiatr Posted January 26, 2023 Posted January 26, 2023 (edited) On 1/25/2023 at 2:10 AM, DBCOOPER011 said: That's great it worked out well for you! It also shows you can get pretty good performance without the top of line motherboards and/or ram.. Hello to all, trying to figure out what is the best performance tuning for i9 13900 kf considering running only IL2 GB (single core performance) and HP G2 in my rig and nothing else as a heavy program workload. So before cpu O.C. my main concern is to lower cpu temps as much as i can trying to keep the highest cpu clocking speed at the same time. So checking ASUS z690 p d4 BIOS i see that my P and E cores management ( EXTREME TWEAKER) are set to AUTO and while playing IL2 GB with my HP G2 and checking cpu cores activity ( temps and clocks) through MSI AFTERBURNER, i see all of my P cores to work at about 5.5 ghz at the same time and all of my E cores to work at about 4,3 ghz at the same time, like my cpu working on full load all the time. Considering that IL2 GB is a single thread program which loads only one core-thread each time (of course not the same all the time ) i am thinking of using the "BY CORE USAGE" Bios setting in order to relax the cpu cores that wont be loaded with IL2 GB execution and lower cpu temps and activate the following BY CORE USAGE cpu settings : 13900K (Stock):Max turbo - 5.8GHz P-cores - 3.0GHz/5.4GHz E-cores - 2.2GHz/4.3GHzP: 58x2 - 54x8 ( so only two cores will be ready to work on full load and 5,8 ghz each time and not all of them) E: 43x16 Full load @ P-55x/E-43x - Vcore=1.137V which is a stock 13900kf setting by the way.. What do you think? Edited January 26, 2023 by dgiatr
chiliwili69 Posted January 26, 2023 Author Posted January 26, 2023 On 1/25/2023 at 3:44 AM, WheelwrightPL said: GPU: 4070 Ti with default clocks (from Gigabyte Very good then, this is more what I should expect from the 4070Ti. Congrats!
DBCOOPER011 Posted January 27, 2023 Posted January 27, 2023 (edited) 13 hours ago, dgiatr said: Hello to all, trying to figure out what is the best performance tuning for i9 13900 kf considering running only IL2 GB (single core performance) and HP G2 in my rig and nothing else as a heavy program workload. So before cpu O.C. my main concern is to lower cpu temps as much as i can trying to keep the highest cpu clocking speed at the same time. So checking ASUS z690 p d4 BIOS i see that my P and E cores management ( EXTREME TWEAKER) are set to AUTO and while playing IL2 GB with my HP G2 and checking cpu cores activity ( temps and clocks) through MSI AFTERBURNER, i see all of my P cores to work at about 5.5 ghz at the same time and all of my E cores to work at about 4,3 ghz at the same time, like my cpu working on full load all the time. Considering that IL2 GB is a single thread program which loads only one core-thread each time (of course not the same all the time ) i am thinking of using the "BY CORE USAGE" Bios setting in order to relax the cpu cores that wont be loaded with IL2 GB execution and lower cpu temps and activate the following BY CORE USAGE cpu settings : 13900K (Stock):Max turbo - 5.8GHz P-cores - 3.0GHz/5.4GHz E-cores - 2.2GHz/4.3GHzP: 58x2 - 54x8 ( so only two cores will be ready to work on full load and 5,8 ghz each time and not all of them) E: 43x16 Full load @ P-55x/E-43x - Vcore=1.137V which is a stock 13900kf setting by the way.. What do you think? I followed this thread at OC.net to tune my 13700KF. Works good for keeping the temps as low as possible and also get good single core performance. You got an awesome CPU for 1.137V stock at 55x allcore. I think I run 1.25v on mine at that ratio... https://www.overclock.net/threads/asus-maximus-z790-extreme-and-intel-i9-13900k-a-tuning-guide-for-beginners.1801569/ Edited January 27, 2023 by DBCOOPER011 1
RossMarBow Posted January 29, 2023 Posted January 29, 2023 Just a heads up for zen 3 zen 4 owners Power limits won't effect IL-2 performance unless you change to them to well below stock to like sub 50w or something ridiculously low. For x3d CO will give you a slight increase in single thread performance. For tuning CO install core-cycler and change the settings file so stressTestProgram = YCRUNCHER mode = 00-x86
Norz Posted January 29, 2023 Posted January 29, 2023 Could someone make a video and share it (youtube or google drive etc) for a test process with CPU i5 10600k ? I have changed my mobo set but the result is lower as expected (in the shared sheet usual result is avg 90...94 but mine is only 84)
Recommended Posts
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 accountSign in
Already have an account? Sign in here.
Sign In Now