Jump to content
chiliwili69

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

Recommended Posts

You can disable hyperthreading on your 4790k to check this assumption

 

Yep I always have hyperthreading disabled for my I7 cpu.

I5 and I3 do not have hyperthreading.

 

There is no real benefit for hyperthreading in gaming currently. Plus it  runs slightly  higher cpu core temps.

There is benefit for heavy duty processing , like video editing etc.

Edited by dburne

Share this post


Link to post
Share on other sites

Looks like I`m getting somewhere, holds a really steady 90 fps, in low action and free flight on the Moscow ship mission. Heavy combat over and near the ships is between 55 and 70+. Set the gpu to default which should get 1759mhz on 3d/boost, but the monitoring software is recording 1885mhz not sure why that is. CPU is not getting hot at 1.25v and 4.8 sitting at max of around 60 degrees. I did have it 5.0 at one point, might try for that and see what happens, or just leave it as it`s running fine atm and I`ve spent 2 days chasing my tail.

Share this post


Link to post
Share on other sites

Just ran the test again and with the new bios and looking like it`s functioning as it should:

 

Frames, Time (ms), Min,    Max,       Avg
  4557,     60000,      44,       91          76

 

Passmark was 2824 and it was running at 4.8ghz on 1.25v CPU temp was around 56 to 60C

 

So Chiliwili was right looks like the 8350k set up properly is a good buy. I`m happy at that.

 

Just done another test at 4.9ghz on 1.3v. CPU temp was 67C

 

Passmark was 2873

 

Frames, Time (ms), Min,    Max,       Avg
  4771,     60000,      45,       91          79.5

 

​So a 100mhz and 0.05v increase achieved an extra 3.5fps on average, however it also increased cpu temp by 8C, max v demanded by the cpu during that test period was 1.206v so 1.3 maybe overkill.I`ll reduce it back to 1.25 to see if it will accept that.

 

​If I`m derailing this thread mate please tell me to stop, no offence will be taken.

Edited by BFsSmurfy

Share this post


Link to post
Share on other sites

 

 

Just ran the test again and with the new bios and looking like it`s functioning as it should:   Frames, Time (ms), Min,    Max,       Avg   4557,     60000,      44,       91          76   So Chiliwili was right looks like the 8350k set up properly is a good buy. I`m happy at that.

 

Good!!  My honor is saved!  :dance:

 

I am happy to see you find the solution and reach 76 fps which is a good score. And glad to know 8350K is a good processor for IL-2 VR.

 

Please, report the CPU GHz you used for that test and also the Passmark numbers.

 

Cheers  :drinks:

Share this post


Link to post
Share on other sites

You delete all the results chiliwilli? Was just going to check the google doc and it's nearly empty now.

Share this post


Link to post
Share on other sites

 

 

You delete all the results chiliwilli? Was just going to check the google doc and it's nearly empty now.

 

Humm... no. I didn´t. They are there as always. You will see now different tabs called "Main2018" which is almost empty, "Main2017", "Kuban2018", etc.

Let me know if you can see them. 

Share this post


Link to post
Share on other sites

Hello there !

 

Thanks a lot for this helpful thread ! I've been struggling for months with my setup, now I might have found a way to fix it, thanks to you guys.

 

My setup consists of a Gigabyte Z170-HD3P motherboard, fitted with a Core i5-6600 (3.3GHz, turbo up to 3.9), 2x8GB 2133MHz RAM and a 1080Ti, all running at default settings.

I've bought BoK (v. 2.012d) last week and was a bit disappointed with the framerate with my Rift CV1 (I was expecting a drop coming from WarThunder, but not such a big one). I'm basically stuck at 45fps all the time, except over the sea where fps jump up to constant 90fps.

 

Running the test track with recommended settings on page 1 of this thread, here are my results :

Frames: 2691 - Time: 60000ms - Avg: 44.850 - Min: 42 - Max: 46

Looks as if ASW remains always on, even though I tested several times, both trying the ctrl+numpad1 key combination or the Oculus Tray Tool option.

 

Passmark CPU Mark : 6499 Global, and 2005 Single Threaded.

 

I've been suspecting my CPU for a while and am about to replace it with a 7700K (along with proper air cooling), but still am not 100% sure it's the root cause. My RAM is also a bit slow, but I'm not willing to replace it considering the outrageous RAM prices right now. It also seems my Kingston HyperX Fury set could possibly be OC to much higher frequency, but that field reaaaaally is not my cup of tea and I have yet to dig into my mobo settings for proper OC (i.e: I'm totally lost atm).

 

Anyway, thanks again for your help.

Edited by ShugNinx

Share this post


Link to post
Share on other sites
34 minutes ago, ShugNinx said:

I've been suspecting my CPU for a while and am about to replace it with a 7700K

Your suspicion is correct. For sufficient single-thread performance to attempt 90 fps, a CPU with clock speed close to 5 GHz is desirable. The 6600 at 3.9 GHz falls well short. 

 

Ideally, RAM should also be something like 3200 MHz CL15 or better. However, this has a relatively small effect. 

 

If you're not comfortable with basic overclocking, even the most expensive components are unlikely to provide satisfactory performance. I would recommend that you upgrade the CPU and follow an overclocking guide for your motherboard model (it is quite easy to learn if you watch a good video). 

 

That said, you may be aware that a major update for BoX is coming in about two weeks. It is known to include significant rendering quality improvements. It is possible, however unlikely, that this update will substantially change the VR performance situation. Therefore, it may be advisable to postpone any upgrades until the effects of the update are known.

Edited by Mitthrawnuruodo
  • Upvote 1

Share this post


Link to post
Share on other sites
8 hours ago, Mitthrawnuruodo said:

Your suspicion is correct. For sufficient single-thread performance to attempt 90 fps, a CPU with clock speed close to 5 GHz is desirable. The 6600 at 3.9 GHz falls well short. 

 

Ideally, RAM should also be something like 3200 MHz CL15 or better. However, this has a relatively small effect. 

 

If you're not comfortable with basic overclocking, even the most expensive components are unlikely to provide satisfactory performance. I would recommend that you upgrade the CPU and follow an overclocking guide for your motherboard model (it is quite easy to learn if you watch a good video). 

 

That said, you may be aware that a major update for BoX is coming in about two weeks. It is known to include significant rendering quality improvements. It is possible, however unlikely, that this update will substantially change the VR performance situation. Therefore, it may be advisable to postpone any upgrades until the effects of the update are known.

Oh God I hope it doesn't drive performance demands up. That would be extremely disappointing considering that BOS is by far and away the best flight sim performer in VR at the moment. Dev's would be insane to ruin all that!

 

Another thing that is hugely important aside from the GPU and RAM/CPU is an SSD.  To be honest on my CPU the difference between stock and overclocked to 5.0ghz isn't really that noticeable, but yesterday I accidentally installed BOS on my HDD without realizing where steam had put it, and I couldn't figure out what the hell was wrong with it (stuttering)!

 

I'd go as far to say that with this game an SSD is essential!

Share this post


Link to post
Share on other sites
1 hour ago, Wolf8312 said:

Oh God I hope it doesn't drive performance demands up. That would be extremely disappointing considering that BOS is by far and away the best flight sim performer in VR at the moment. Dev's would be insane to ruin all that!

 

Another thing that is hugely important aside from the GPU and RAM/CPU is an SSD.  To be honest on my CPU the difference between stock and overclocked to 5.0ghz isn't really that noticeable, but yesterday I accidentally installed BOS on my HDD without realizing where steam had put it, and I couldn't figure out what the hell was wrong with it (stuttering)!

 

I'd go as far to say that with this game an SSD is essential!

 

 

If you have enough RAM you can load the game on a RAM-disk. (plenty of software that does that for you for free). It is actually the only way to make stuttering tolerable for me while flying (model-load stutter). 

 

I don't recall exactly, but I think its around 8-10GB RAM that you need for the ramdisk for il2-bos.

Share this post


Link to post
Share on other sites
On 07/03/2018 at 12:26 AM, Mitthrawnuruodo said:

Your suspicion is correct. For sufficient single-thread performance to attempt 90 fps, a CPU with clock speed close to 5 GHz is desirable. The 6600 at 3.9 GHz falls well short.

 

What strikes me the most here, is that the task manager CPU usage graphs (set to show individual core activity) doesn't show any core saturated at 100% activity... Is there any application that can provide with more detailed CPU, RAM and GPU usage data ?

Share this post


Link to post
Share on other sites
3 hours ago, ShugNinx said:

 

What strikes me the most here, is that the task manager CPU usage graphs (set to show individual core activity) doesn't show any core saturated at 100% activity... Is there any application that can provide with more detailed CPU, RAM and GPU usage data ?

The CPU usage (even of individual cores) is not an accurate indication of CPU performance for Il-2.

 

Often, no core is close to 100%, even when the CPU is clearly limiting performance. 

 

HWiNFO64 is quite a good application for monitoring, but it might not reveal anything interesting in your case. 

 

Sometimes, the best way to evaluate CPU performance is to look at the GPU usage. If the GPU is below 100%, it’s very likely that the CPU is holding it back  (note that frame rate limits can interfere with this test). 

Edited by Mitthrawnuruodo

Share this post


Link to post
Share on other sites
3 hours ago, Mitthrawnuruodo said:

Sometimes, the best way to evaluate CPU performance is to look at the GPU usage. If the GPU is below 100%, it’s very likely that the CPU is holding it back  (note that frame rate limits can interfere with this test). 

 

Interesting ! I'd never had thought about watching the GPU graph to evaluate the CPU performance. And indeed, both seem to have plenty of headroom while IL2 is running, which puzzled me even more, but your input makes a lot of sense.

 

Thanks for HWiNFO64, it doesn't display graph but can log tons of sensor values, which will do the trick.

Share this post


Link to post
Share on other sites
On 06/03/2018 at 11:57 PM, ShugNinx said:

Running the test track with recommended settings on page 1 of this thread, here are my results :

Frames: 2691 - Time: 60000ms - Avg: 44.850 - Min: 42 - Max: 46

Looks as if ASW remains always on, even though I tested several times, both trying the ctrl+numpad1 key combination or the Oculus Tray Tool option.

 

Passmark CPU Mark : 6499 Global, and 2005 Single Threaded.

Thank you for reporting your test.

I have being quite busy lately with no time to update these test results. 

I have uploaded your test to the online sheet.

Now there are four tabs:

Balapan3.0: For tests with the balapan track with the new version3.0x

Kuban 3.0: For tests with the Kuban track with the new version 3.0x

Balapan2.0: For tests with the old version 2.0x (here your test)

Kuban 2.0: For test with the Kuban track with the old version 2.0x (only a few tests were performed before release of 3.001)

 

Regarding your test:

This 100% as expected.

You 2005 single-thread mark, according to the correlation (column U), should achieve 43.7 fps. You achieved 44.85, so you are a bit lucky.

In your case the CPU is 95% responsible of your low performance with this test. Don´t spend money in RAM upgrade.

You can run with LOW or MEDIUM settings. Or upgrade your rig.

You can try to run again the test with the new 3.001 version of BoX.

 

 

On 07/03/2018 at 9:17 AM, Wolf8312 said:

Another thing that is hugely important aside from the GPU and RAM/CPU is an SSD.  To be honest on my CPU the difference between stock and overclocked to 5.0ghz isn't really that noticeable, but yesterday I accidentally installed BOS on my HDD without realizing where steam had put it, and I couldn't figure out what the hell was wrong with it (stuttering)!

 

I'd go as far to say that with this game an SSD is essential!

 

Hey, the only advantage of a SSD over a HD drive would be in the loading times when you launch the game or launch a mission. Once you do that, the data used by the CPU is loaded into the RAM and no data is used from SSD/HD while you are flying. So, a SSD or HD will not affect to the results of this VR test. You can test it by yourself.

 

Instead, overclocking the CPU has a very significant impact in the achieved fps. The facts are here:

 

On 10/03/2018 at 8:25 PM, ShugNinx said:

What strikes me the most here, is that the task manager CPU usage graphs (set to show individual core activity) doesn't show any core saturated at 100% activity... Is there any application that can provide with more detailed CPU, RAM and GPU usage data ?

 

A heavy thread can saturate a Core and the CPU%load could show just 30%. But the CPU is the bottleneck.

The way the %load of a CPU or Core is measured is misleading. Don´t trust on it.

The threads jumps from Core to Core to distribute heating:

 

Edited by chiliwili69

Share this post


Link to post
Share on other sites

I can only report my own empirical experience.  I loaded the game on HDD by accident without realizing and couldnt understand why the game was suddenly very stuttery. Realized and moved back to SSD and it was fine again.

  • Upvote 2

Share this post


Link to post
Share on other sites

Well guys, the balapan track record flight from first post doesn´t work any more with the new 3.001 version.

I suppose it is due to new graphics options available and other render engine improvements. 

 

We need to create a new track flight with 3.001. I can do that but if there is any volunteer to create it feel free to give me the track and we will the same for all testing.

 

For the setting, I will try to maintain the same graphics setting from previous test, but there is new graphics options in the new version 3.001.

I attach a picture with the new proposed graphics. The differences are:

 

Screen resolution: 1600x900 is better than previous 1024x768 and it doesn´t affect performance. Since both eyes are displayed we need more horizontal pixels than 1024

UI scale: leave it at default 100%

Mirrors: Off . Here the previous Balapan track was using a plane with no mirrors. We could use a plane with mirrors for the new track, but for the purpose of the test the mirrors will be set off. So it allows to test the influence of mirrors, but the baseline is with no mirrors.

Horizon draw distance: This is a new setting. I would propose to set it up to 100Km which is a medium value.

Clouds quality: Here again I propose Medium. We can play later with it to measure the influence on fps.

Use 4K textures: Off. Since it is a new feature. We can play later with it to measure the influence on fps.

 

The length of the track could be around 2 minutes (no more), so we will need to run it multiple times with different settings and study the influence.

 

Regarding the scenario, I don´t know what it would be best. If using Stalingrad map (like Balapan) or using new Kuban map ( with more mountains, sea, etc). But I think it should be just only one. Otherwise it will take too much effort to maintain two spreadsheets along the time. (despite new VR devices will arrive this year).

 

So please, If any of you would want to create that recorded flight I will be please to use it and will update the first post instructions. Remember to include 2-3 bombers in the scene and shoot at them heavily to produce smoke, include some clouds, sea water and some low level flight over a city.

 

 

vrtestsettings.jpg

Share this post


Link to post
Share on other sites

Hello Chiliwili !

 

Thanks for your advice on my results, 7700K is my best option.

 

10 hours ago, chiliwili69 said:

Well guys, the balapan track record flight from first post doesn´t work any more with the new 3.001 version.

I suppose it is due to new graphics options available and other render engine improvements. 

 

We need to create a new track flight with 3.001. I can do that but if there is any volunteer to create it feel free to give me the track and we will the same for all testing.

 

That is a huge setback... Creating a new track will make it harder to compare new results with the old ones right ?

Wouldn't it be better to wait for fix, as I expect a few patches will be issued within the next weeks for various other reasons ?

 

Anyway, ditching the tracks compatibility will likely annoy many people way beyond our testing purposes.

Edited by ShugNinx
typos...

Share this post


Link to post
Share on other sites

Can anyone help me to understand. I have an OC set to 5.0 ghz. But it seems to work alright especially outside of the game recording at about 4.9 when the game is loading, but then when I go into the game and actually fly it just hovers around 4688.5 Mhz and rarely seems to go any higher. Shouldnt it be closer to 5.0 MHz in game, and anyone know why it's not getting there? Is it a case perhaps of only using what it needs at any given moment or someting?

 

Strange thing is Doom seemed to hold the OC.

Edited by Wolf8312

Share this post


Link to post
Share on other sites
11 hours ago, ShugNinx said:

That is a huge setback... Creating a new track will make it harder to compare new results with the old ones right ?

Wouldn't it be better to wait for fix, as I expect a few patches will be issued within the next weeks for various other reasons ?

 

Yes, It is a pity that the track records doesn´t work. But the important thing is not to measure the fps difference (2.0x vs 3.0x) but to have a common measuring tool.

With this tool people can compare their performance with their peers and study what to do. It can be also used to analyse the influence of various settings, since it is something that always throws tons of posts in this forum due to subjectivity.

 

I don´t know if the fix will arrive since this is not an important thing (there are other more important VR issues), but if the fix arrives the Balapan track will be valid again and people will be able to run any of the tracks (Balapan and the new one)

4 hours ago, TUS_Samuel said:

I'd like to make a track. Within 2 days I'll make a custom mission on Kuban and make a record.

 

Thanks Samuel, no rush for that. Try to use a plane with mirrors and available to anyone (not collector planes, I am not sure if a person who don´t have a collector plane can fly a trackrecord of that collector plane)

1 hour ago, Wolf8312 said:

Can anyone help me to understand. I have an OC set to 5.0 ghz. But it seems to work alright especially outside of the game recording at about 4.9 when the game is loading, but then when I go into the game and actually fly it just hovers around 4688.5 Mhz and rarely seems to go any higher. Shouldnt it be closer to 5.0 MHz in game, and anyone know why it's not getting there? Is it a case perhaps of only using what it needs at any given moment or someting?

 

Strange thing is Doom seemed to hold the OC.

 

Just two things to check (without being an expert on the matter):

 

1. AVX offset: IL-2 uses AVX instructions, when they are used the CPU uses the AVX offset to down-clock the CPU to prevent excessive heating. You should put it a zero, but keeping an eye on temps.

 

2. Thermal throttling: There are a number of parameters in the BIOS (Amperage, power, temp) that down-clock your CPU. At the end of the day the important one is Temp. I have my limits on Amp and Power set to max.

Share this post


Link to post
Share on other sites

Yeah interesting you should say that as AVX offset was set to 3 in the bios. I reinstalled Asus suite 3 and it seemed to be working but is now downclocking again.

 

OK think I got it now turns out some of the settings in Asus suite were actually AVX settings that were changing bios setting to 3 AVX. Turned them off and now the OC is holding!

 

Thanks man!

Edited by Wolf8312

Share this post


Link to post
Share on other sites
20 hours ago, chiliwili69 said:

Thanks Samuel, no rush for that. Try to use a plane with mirrors and available to anyone (not collector planes, I am not sure if a person who don´t have a collector plane can fly a trackrecord of that collector plane)

 

I was not able to run Kuban mission because I did not order BoK, only La5-FN. So I made a Stalingrad mission with heavy clouds, 8 bombers, AAA and some ground units.

Today I'll select a plane from standard BoS or BoM with mirrors and make a final record.

Share this post


Link to post
Share on other sites
2 hours ago, TUS_Samuel said:

So I made a Stalingrad mission with heavy clouds, 8 bombers, AAA and some ground units.

that´s quite fine. Stalingrad city and their black smoke is also a demanding scenario, although not as demanding as the mountains of Kuban.

 

I think the planes with mirrors are:

Spitfire, Yak-1b, P-40, P-39L and Hs-129

 

I think none of the BOS planes incorporate mirrors, but that´s ok if there is no mirrors.

With BOM you can use the P-40 but it would be better to use BOS since Stalingrad city is heavier and BOS is own by more people.

 

Share this post


Link to post
Share on other sites

The more tracks you bring in, the more spread will be the comparison, and the harder it will become to assess and evaluate.

;)

 

Share this post


Link to post
Share on other sites

NEW BENCHMARK AVAILABLE AT THE FIRST POST!!

 

Hey guys!  With the kindly help of Samuel, who has created a very demanding track in Stalingrad city, we have updated the procedure of the first post with the links and new settings to be used for version 3.001 and beyond.

 

As you will see this track is quite intrepid (Samuel is an ACE!) and brings the requirements to a new level!

It is a bit longer than before (200 seconds), but now there is a variety of scenes (ground targets, ground explosions, shooting, smoke, multiple planes, water, sun, clouds, low altitude, etc, etc, so mix of everything). Being longer the final results is an average of everything.

 

I have run the test with my 4790K at 4.9GHz, RAM at 2933 and with the 1080Ti with no OC:

 

Frames: 13152 - Time: 200000ms - Avg: 65.760 - Min: 39 - Max: 91

 

So, please, feel free to report your performance of your rigs with this new IL2 v3.001

 

On 21/3/2018 at 10:24 PM, SCG_Fenris_Wolf said:

The more tracks you bring in, the more spread will be the comparison, and the harder it will become to assess and evaluate.

;)

 

yes, the more tracks the worse. My target is to have just one track (right now there is only one working).

There were three problems with the old Balapan track:

- It doesn´t work anymore with the new IL-2 v3.001 (this is a thing of the new v3.001. I don´t know if it can be repair it)

- Most of top rigs were hitting the 90fps, so they can not be used as a measuring tool. (here we can increase settings but then it would be like having a new track)

- The track didn´t include items of the new track (sun, sun shadows, water, low fly, ground targets, intensive damage model, etc). The new track is more complete.

Edited by chiliwili69

Share this post


Link to post
Share on other sites

i5 4690k @ 4.5Ghz, RAM at 2400Mhz, 980ti, cpu passmark single-threaded 2636

 

Frames: 10690 - Time: 200000ms - Avg: 53.450 - Min: 34 - Max: 88

Share this post


Link to post
Share on other sites
22 minutes ago, coconut said:

i5 4690k @ 4.5Ghz

Thank you for your test. This CPU is interesting since in the previous Balapan test (shown in the tab Balapan2.0 and marked in yellow) it was showing an overall below than expected performance (ie. a good singlethread mark and a not good fps)

Let´s see if we have more test to compare with.

Share this post


Link to post
Share on other sites

Okay, just had a little bit of time this afternoon to do a few quick tests. Results below. One comment though, your instructions for setup assume the older version of SteamVR is installed. In the latest Beta, they have once more changed how to set SuperSampling because now they have an option to allow it to be set automatically based on GPU power (and most likely a few other settings). It no longer has a scale from .60 to 5.00 though. Its now a percentage and goes from 20%-500%. There is a manual override setting which I selected as I set my SS using OTT.  I just wanted to point this out for those that might be using the SteamVR Beta. One other comment, I tested with the 3Dmigoto mod installed and also uninstalled.

 

Nvidia Drivers: 391.24 (installed 3/22/2018)

GPU Clock Speed: 2050Mhz (set via profile in ATI Afterburner)

Video Mem Speed: 5961Mhz (set via profile in ATI Afterburner)

I7 7700k OC to 5Ghz although ATI Afterburner only indicated it was running at 4.9Ghz during these tests. 

ACX Offset: None

 

VR Test Results with 3Dmigoto mod installed (all effects disabled)

2018-03-22 14:06:56 - Il-2
Frames: 12826 - Time: 200000ms - Avg: 64.130 - Min: 43 - Max: 91

 

VR Test Results with Game Stock

2018-03-22 14:20:12 - Il-2
Frames: 13959 - Time: 200000ms - Avg: 69.795 - Min: 41 - Max: 91

 

Comments:

Looks like lefuneste has really been tweaking his mod as overhead impact when running it is now is very low. I think that the discrepancy shown between the minimum framerates with it installed and not installed was just an anomaly. If I get more time I will run a few more cycles and see if that changes.

 

I didn't have time to test the desktop in-monitor version yet.

 

Madmatt

 

 

 

 

SteamVR Beta Auto SS.jpg

Edited by Madmatt
Added Screenshot of SteamVR Beta Auto SS Settings

Share this post


Link to post
Share on other sites

Thanks Madmatt for reporting the performance with/without Lefuneste mod. The difference now is lower.

Regarding the SteamVR, yep, I will also update the first post to reflect also in % (I assume 170% is the former 1.7 SS)

You didn´t reported you CPU-Mark or STMark, but I assumed the same than in previous test you reported with the Balapan2.0.

Share this post


Link to post
Share on other sites

Hello there !

 

Here are my results with the new benchmark for 3.001 :

- Frames: 8917 - Time: 200000ms - Avg: 44.585 - Min: 26 - Max: 84 (first post here was definitely with ASW ON...)

- Passmark CPU Mark : 2132 Single Threaded (slightly more than the first I posted here, should run this one more often...)

 

 Gigabyte Z170-HD3P, i5-6600 (3.3GHz, turbo up to 3.9), 2x8GB 2133MHz RAM 1080Ti, all running at default settings.

 

I'm receiving a delidded 7700K tomorrow, I'm just posting this as initial reference before upgrade.

 

Edit : for what it's worth, Pixel Density overlay displays 1.32 with SS 1.7 set in SteamVR settings (SS in Oculus tray tool at 0).

Edited by ShugNinx

Share this post


Link to post
Share on other sites
6 hours ago, chiliwili69 said:

Thanks Madmatt for reporting the performance with/without Lefuneste mod. The difference now is lower.

Regarding the SteamVR, yep, I will also update the first post to reflect also in % (I assume 170% is the former 1.7 SS)

You didn´t reported you CPU-Mark or STMark, but I assumed the same than in previous test you reported with the Balapan2.0.

 

chiliwili69,

 

I didn't have time to run the CPU Mark tests this afternoon but I just ran them. They are actually noticeably higher than last time I ran them. Probably due to some more BIOS tweaking I did recently plus I think the latest patches for Spectre and Meltdown have regained some of performance loss shown in earlier patches.. Here are the new scores:

 

CPU Mark: 14280

CPU Single Threaded: 2877

 

Madmatt

CPU Mark.jpg

Share this post


Link to post
Share on other sites
On 22/3/2018 at 10:55 PM, ShugNinx said:

I'm receiving a delidded 7700K tomorrow, I'm just posting this as initial reference before upgrade

 

Thanks for posting your results just before upgrading your CPU. You will definitely get a fps boost with the 7700K.

Take your time to do a step-by-step approach for your overclocking. I made a thread for mine:

 

Share this post


Link to post
Share on other sites

Just ran a series of benches at various clock speeds, both with and without hyper-threading enabled. Hardware specs in sig, results below:

 

Stock with 3200MHz RAM (note: clock speed is 4.3 to 4.4 GHz in game):

Min: 40

Max: 91

Avg: 57

Frames: 6843

 

stock.jpg.9c6946850d3bf39c00d2f53025b8788b.jpg

 

4.7 GHz with HT:

Min: 42

Max: 91

Avg: 67.8

Frames: 8141

 

47.jpg.4d12644b58798a5327c388e1229a0e61.jpg

 

4.9 GHz without HT:

Min: 41

Max: 90

Avg: 66.6

Frames: 7998

 

49_no_HT.jpg.c241288cba7c351ed97064d7bf031e59.jpg

 

5.0 GHz with HT (note: 4200 MHz cache, default is 4400 MHz):

Min: 41

Max: 91

Avg: 69.2

Frames: 8312

 

50_42_with_HT.jpg.e644af2777648d7bc0719c3e12a9bed6.jpg

 

5.0 GHz without HT:

Min: 41

Max: 91

Avg: 69.4

Frames: 8327

 

50noHT.jpg.ca0308bc2f616750d5499103bf0a2c93.jpg

 

 

 

The interesting thing here to me is that the 300MHz jump from 4.4 to 4.7 GHz results in a massive FPS increase while the 300 MHz jump from 4.7 GHz to 5.0 GHz has only a marginal impact. The 4.9 GHz without HT result doing worse than 4.7 with HT is likely margin of error, I only ran each test once.

 

Edited by BeastyBaiter

Share this post


Link to post
Share on other sites
6 hours ago, BeastyBaiter said:

The interesting thing here to me is that the 300MHz jump from 4.4 to 4.7 GHz results in a massive FPS increase while the 300 MHz jump from 4.7 GHz to 5.0 GHz has only a marginal impact.

Thanks for reporting your tests. But the frames number doesn´t match with the expected values, it should be in the range of 13 thousands...

 

It seems that beyond 4.7 the bottleneck it somewhere else. Perhaps RAM.

The test performed by Madmatt with a 7700K at 5.0Ghz was slightly higher than your test with the 8700K at 5.0Ghz, but he was using 3600Mhz RAM.

Perhaps Madmatt can try to run the test with the RAM at 3200 or other values just to see the influence.

 

Also, this morning I have slightly modified the first post with a new reference view of the cockpit (to avoid using landing gear indicators) which is just what it is seen after loading the track. This might influence the tests performed before this morning.

Share this post


Link to post
Share on other sites

Ah, it's 200 seconds, read it as 2 minutes. BTW, before we go too nuts with this, we should change tracks. That one is really far removed from normal gameplay, especially given that it goes 2 minutes straight of almost non-stop shooting thanks to unlimited ammo and invulnerability. A good test track should also start and end with basically nothing going on, to account for minor differences in when the game may start the track (it loads unpaused, then pauses 2-3 seconds, then user unpauses).

Share this post


Link to post
Share on other sites

I just run the benchmark as advised in the first post. Out of curiosity i run it both in the 1920x1080 resolution and on 2560x1440 that I normally play it on.

Both times all settings in game were maxed out (should I be using different/lower settings ?)

 

Here are the results:

 

1920x1080: Frames: 24685 - Time: 200000ms - Avg: 123.425 - Min: 83 - Max: 145

2560x1440: Frames: 18805 - Time: 200000ms - Avg: 94.025 - Min: 52 - Max: 125

 

As for the hardware:

CPU 8700K (stock clocks)

GPU: 1080Ti (stock clocks)

RAM: 32GB @ 3000Mhz (XMP enabled)

Share this post


Link to post
Share on other sites

Reran test at 4.9 GHz (no HT)

Min: 43

Max: 91

Avg: 72.68

Frames 14536

 

I still think we should change tracks. I also think we should drop to 1.3 SS from 1.7 and bump both clouds and shadows to high. I tried playing the game at the benchmark settings and planes completely vanish against clouds. Though FPS was largely fine, it was basically unplayable in any mission with clouds due to invisible planes. The aforementioned changes make planes reasonably visible (though still a bit garbled).

Share this post


Link to post
Share on other sites

Just ran the new benchmark:   Frames: 14426 - Time: 200000ms - Avg: 72.130 - Min: 42 - Max: 91

 

Also re-ran Passmark, just to see if Windows updates or anything else had changed anything - results were slightly less than a few months ago, but not by much:

 

Old results:  Passmark = 7087.6; CPU Mark 14131.3; CPU single thread = 2913

 

Current results:  Passmark = 7059; CPU Mark 14026; CPU single thread = 2932

 

Same PC specs in each case:

 

Intel Core i7-7700K @ 5.0 GHz (Corsair H75 liquid cooler); 16GB RAM (Corsair Vengeance DDR4 3200MHz)

EVGA GTX 1080 TI SC2, 11 GB; ASUS ROG Maximus IX Code MB; Realtek ROG SupremeFX audio

Windows 10 Pro 64-bit; Oculus Rift (CV1); Logitech Force 3D Pro; Saitek Rudder Pedals and Throttle Quadrants

  

 

[edit] I've been holding off doing any testing/flying until they get around to issuing the hotfix that seems almost inevitable with a patch/update of this size, so I have no idea what settings I'll actually be using for my own flying, but just for grins I ran the new benchmark again using what I would like those eventual settings to be.  Results were somewhat disappointing:  Avg: 53.190 - Min: 37 - Max: 75. And this was without running my map/OVRdrop app, which eats up even more FPS.  So, it looks like I'll be joining what seems like the rest of you in the hunt for good settings that will make the game look decent and still run well in VR after the update.

Edited by TG-55Panthercules

Share this post


Link to post
Share on other sites

Hello again !

 

17 hours ago, chiliwili69 said:

Thanks for posting your results just before upgrading your CPU. You will definitely get a fps boost with the 7700K.

 

Thank for cheering me up ! i7-7700K arrived yesterday, along with a Masterliquid Pro 240 and a Fractal Design Define S for better airflow.

First result with stock settings (no OC) : Frames: 11768 - Time: 200000ms - Avg: 58.840 - Min: 38 - Max: 91

(previous result with i5-6600 : Frames: 8917 - Time: 200000ms - Avg: 44.585 - Min: 26 - Max: 84)

 

Passmark CPU Mark : 2572 Single Threaded (previous 2132)

 

 

These are very promising results. Going to enjoy it a little bit before starting the long road to stable overclocking.

 

Gigabyte Z170-HD3P, i7-7700K, 2x8GB 2133MHz RAM 1080Ti, all running at default settings.

 

Edit : first attempt to RAM OC from 2133 to 3200 MHz (yeah I know, I could have tried this before upgrading the CPU...) :

Frames: 13125 - Time: 200000ms - Avg: 65.625 - Min: 39 - Max: 91

 

Edited by ShugNinx
Adding new results

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...