Jump to content
chiliwili69

Benchmark for CPU/RAM performance: Remagen 4.002 to 4.005

Recommended Posts

2 hours ago, Jaws2002 said:

Even the single thread performance is better with the 3950x. Here are the ST mark difference between this two

 

I was taking a closer look to your tests with PerformanceTest from Passmark.

 

With 3600X:

Running only STmark run: STMark=3115

Running full CPU test: STMark=3052

 

With 3950X (game mode 8 cores):

Running only STmark run: STMark=3264

Running full CPU test: STMark=3117

 

With 3950X (creator mode 16 cores)

Running full CPU test: STMark=3142

 

Firstly with the 3600X it is strange that you get different STmarks if you run the full CPU test or just only the single threaded. I think the explanation is that the CPU or core freq is not the same and it is changing depending on the CPU load. If you only load one core, the system will OC more that single core. 

 

Then, with the 3950X also happen the same.

 

I don´t know if it is possible to do fix OC for all the enabled cores in the 3950X.

 

26 minutes ago, Jaws2002 said:

That should tel you everything about the impact the CPU has on Game performance

 

Yeah, I saw you other post and saw the cores utilization. But it doesn´t mean that CPU is not important for IL-2 VR. What it is not really important is going above 6-8 cores.

Edited by chiliwili69

Share this post


Link to post
Share on other sites

It's still important, but as graphics cards gained more performance than CPUs and the resolution increased, more and more of the work load was moved to the GPU. The tasks done on the GPU are also significantly better optimized for multi threading. So while CPU is still important, i'd say that anything  with six or eight cores from Intel or AMD will do about the same in games.

 In this one six cores is all you need. It doesn't even need to have two threads per core. While i thought that boost clock was very important in game and the high boost gave intel the edge, now i know that it's the higher FSB speed and memory controller that still gives Intel the edge for gaming, but in the same time the difference is so small, that it's not worth losing the more modern platform and better work load performance, for two three frames in some games.

Edited by Jaws2002

Share this post


Link to post
Share on other sites
1 hour ago, Loki_1982b_rock1080 said:

I've ordered some faster RAM (3600mhz CL16), should be here next week then I will re-run the tests so we can see how that influences things.

 

Thank you for running the test, it is the first 2080-Super in the table.

 

Let´s see how your new memory goes.

31 minutes ago, Jaws2002 said:

In this one six cores is all you need. It doesn't even need to have two threads per core.

 

I was thinking about this. In the intel side, the top performers (9900K) uses 8 cores, whereas all the others uses 6 or 4.and also others 8 cores, 6 cores and less

I have added another two columns to the sheet:

- Number of active cores

- HT On or Off

 

Perhaps having 2 extra cores make the difference on the Intel side (or perhaps is the Cache, or the NB freq) or maybe the GPU influence in this test.

Edited by chiliwili69

Share this post


Link to post
Share on other sites
29 minutes ago, chiliwili69 said:

Perhaps having 2 extra cores make the difference on the Intel side

 

That´s not true. I just realized that the 9700K has also 8 cores. And the Alonzo test with 6 cores (2 are disabled) confirmed that having 8 cores doesn´t give and advantage.

Share this post


Link to post
Share on other sites
35 minutes ago, chiliwili69 said:

I was thinking about this. In the intel side, the top performers (9900K) uses 8 cores, whereas all the others uses 6 or 4.

I have added another two columns to the sheet:

- Number of active cores

- HT On or Off

 

Perhaps having 2 extra cores make the difference on the Intel side (or perhaps is the Cache, or the NB freq) or maybe the GPU influence in this test.

 

 

That's true, but are both on custom loop and  running overclocked 2080ti. 

 

Edit:

 

Hey chiliwilli, we could get a lot more people to run the benchmark if you start a thread in the General discussion forum and link it to this thread. Very few people have VR and visit this thread. It could benefit everyone. I bet we'll get a lot more people to test their game.

Edited by Jaws2002

Share this post


Link to post
Share on other sites
16 minutes ago, Jaws2002 said:

Hey chiliwilli, we could get a lot more people to run the benchmark if you start a thread in the General discussion forum and link it to this thread. Very few people have VR and visit this thread. It could benefit everyone. I bet we'll get a lot more people to test their game

 

Yeah, but I already put it in the Hardware section to get more people. 

 

In any case, this thread is not "mine". I am just the recorder. You are welcome to create a link wherever you want.

Share this post


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

 

I was taking a closer look to your tests with PerformanceTest from Passmark.

 

With 3600X:

Running only STmark run: STMark=3115

Running full CPU test: STMark=3052

 

With 3950X (game mode 8 cores):

Running only STmark run: STMark=3264

Running full CPU test: STMark=3117

 

With 3950X (creator mode 16 cores)

Running full CPU test: STMark=3142

 

Firstly with the 3600X it is strange that you get different STmarks if you run the full CPU test or just only the single threaded. I think the explanation is that the CPU or core freq is not the same and it is changing depending on the CPU load. If you only load one core, the system will OC more that single core. 

 

Then, with the 3950X also happen the same.

 

I don´t know if it is possible to do fix OC for all the enabled cores in the 3950X.

 

About the STmark scores, it's normal to get lower single core score when you run the full CPU test, because the CPU is already heated up after running the multicore tests. The single core test is ran last.

I did overclock it manually a bit with all cores locked in, but that way it can only boost to 4.25Ghz. Remember that means 16 cores, 32 threads blasting at 4.25Ghz. In game that's a waste of resources, energy and you loose the much higher opportunistic boost on the cores the game actually uses. I got lower game performance with all games locked in, but in heavy threaded tasks that's the way to go. I got the highest cinebench score with manual all core overclock.  Gaming, not so much, because the game engines are so poorly optimized for multithreading.

 

Think of Ryzen 3000 like a video card. It will boost based on a complex algorithm, that takes into account heat, power, fan speed, work load and a bunch of other data points on all cores to decide how to use the CPU. You'll never get the same score twice in a test.

Edited by Jaws2002

Share this post


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

 

That´s not true. I just realized that the 9700K has also 8 cores. And the Alonzo test with 6 cores (2 are disabled) confirmed that having 8 cores doesn´t give and advantage.

 

Not sure that would be an accurate statement, maybe in one benchmark but not in the overall picture.

Especially considering all the other stuff/processes that are running in addition to the game.

 

Edited by dburne

Share this post


Link to post
Share on other sites
6 hours ago, Loki_1982b_rock1080 said:

For what it's worth I ran a GPU passmark for my 2080 Super. Not sure if you are incorporating this too, but curious how it stacks up against the 2080 and 2080ti

2080 super.png

 

Does anyone know how accurate this is for benchmarking GPUs? I'm getting crazy high numbers for my 2080 Super. I've ran the test ten times including downclocking my ram and cpu and I'm constantly getting between 19000 and 21000 for my GPU benchmark which is 30% higher than average for this card. Maybe I won the silicon lottery?

 

Sorry to sidetrack the thread, just a bit baffled.

Share this post


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

 

But the fps reported by fpsVR are not exactly the same than the one reported by FRAPS. FRAPs just only counts for the 3 minutes of the flight record, whereas fpsVR start to count fps from the moment I launch the game. So fpsVR counts the launching game (enter password), going to track record, launch track record, run track record and exit game. 4 minutes is total.

 

I don´t know how to synchronize fpsVR to start counting just when I press the P to run the test. And stop after 3 minutes.

 

 

I have asked the fpsVR developer to implement frametimes logging and also some kind of time snapshot for histograms.

He responded positively.

Feel free to encourage him at the link below. 😉

https://steamcommunity.com/app/908520/discussions/0/1750142176477291142/

Edited by wju
  • Thanks 1

Share this post


Link to post
Share on other sites
9 hours ago, Loki_1982b_rock1080 said:

Not sure if you are incorporating this too

 

In principal, I didn´t want to include the GPUs into the performance equation. I just wanted to know what CPU-RAM is optimum for IL-2 and consequently for IL-2 VR.

GPUs are important for IL-2 at 4K and IL-2 VR, specially with latest devices and depending on how much SS you apply.

 

But in the table, based on Grief comments, we included a column to indicate the "power" of each GPU (Passmark GPU public average). Just in case there is some hidden correlation with GPUs, but I wanted to create a benchmark where the GPU is not the bottleneck (that´s why I reduced clouds from high to low).

We use the public average of all the GPUs of that model (this is the easiest approach). I don´t know if we should complicate more the test by including individual "3Dgraphic mark".

 

The performance you achieve with your 2080S is well above the average of your model. 

9 hours ago, Jaws2002 said:

I did overclock it manually a bit with all cores locked in, but that way it can only boost to 4.25Ghz

 

Just only as an experiment, have you tried to enable only 4 or 6 cores in BIOS and see how much fixed OC you can reach. 

Then you can run the IL-2 test, maybe you will surpass the 9900K.

8 hours ago, dburne said:

Not sure that would be an accurate statement, maybe in one benchmark but not in the overall picture.

Especially considering all the other stuff/processes that are running in addition to the game.

 

Yes, you could be also right, specially with all the VR software running.

Questions are:

 

Does 8 cores perform better than 6 cores for IL-2 in 1080p monitor???

and

Does 8 cores perform better than 6 cores for IL-2 VR???

 

Does 6 cores perform better than 4 cores for IL-2 in 1080p monitor???

and

Does 6 cores perform better than 4 cores for IL-2 VR???

 

I only have 4 cores ( 😞  )  so no possible experiments for me, but for all of you with 8 cores or 6 cores CPU, could you run the benchmark in monitor and VR with only 6 and 4 cores enabled? (running test at identical CPU freq and with the same version of IL-2)

 

PD: Disabling/Enabling cores is easy. Just go to BIOS CPU section.

Share this post


Link to post
Share on other sites
3 hours ago, Loki_1982b_rock1080 said:

Does anyone know how accurate this is for benchmarking GPUs?

 

This is a good question. I think the info it delivers is good to know where you are. In this case you are at the top of the 2080 Supers out there.

The performance depends on the cooling capability of the cards and the PC case, the OC capability of the card, (perhaps also influenced by MoBo), etc.

In the PerformanceTest you can see this charts for each GPU (or CPUs) by clicking in the red box and selecting the GPU model:

2080S.thumb.png.eabee6578e3153d9297e9eb030c38efb.png

 

Share this post


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

 

The avg fps reported by the fpsVR stats are 56 fps for the run with same benchmark settings and 66.5 for the one with the modified settings.

But the fps reported by fpsVR are not exactly the same than the one reported by FRAPS. FRAPs just only counts for the 3 minutes of the flight record, whereas fpsVR start to count fps from the moment I launch the game. So fpsVR counts the launching game (enter password), going to track record, launch track record, run track record and exit game. 4 minutes is total.

 

I don´t know how to synchronize fpsVR to start counting just when I press the P to run the test. And stop after 3 minutes.

 

In any case, It is strange that having your GPUs frametimes from 16 to 22 ms your reported fps by fraps is 52. Did you disabled Motion reprojection in WMR? (you have to do it to properly run the test)

 

Yes. I didn´t describe that in the post but basically is how you described. I run the test on my head (without moving it) and pressed the recenter button.

 

 

Chili - Using fpsVR, I do match it up with your track.  When starting up fpsVR I'm not recording the history, after everything is set up an ready to go,  I go to fpsVR and hit the "record" button, then go back into game and hit "P" to start the track.  It takes a couple seconds for that transition but that is minor. 

 

My fpsVR history was for the full length of the track (3.6 min), while Fraps stopped at 3 min, which now occurs to me while my fps is different between Fraps  and fpsVR.  I could have stopped it at 3 min too but forgot.

 

Maybe you could try it that way.  Don't know how Goblin did his timing with fpsVR.

 

------------------------

fpsVR Report:
App: Il-2 HMD: HP Reverb VR Headset VR1000-2xxx (90.0009994506836 Hz, IPD 61.0)
GPU: NVIDIA GeForce GTX 1080 (26.21.14.3630) CPU: Intel(R) Core(TM) i5-9600K CPU @ 3.70GHz
Delivered fps: 35.4  Duration: 3.6min.

Edited by 71st_AH_statuskuo

Share this post


Link to post
Share on other sites

The timing with fpsVR was a bit off, about 30 sec, and thus not completely accurate.

Share this post


Link to post
Share on other sites
18 hours ago, Loki_1982b_rock1080 said:

I've ordered some faster RAM (3600mhz CL16), should be here next week then I will re-run the tests so we can see how that influences things.

 

 Motherboard: MSI Z390-A Pro
 CPU:                 9700K
 CPU Freq:        5.0 Ghz
 L3 cache:        12 MB
 RAM type:        DDR4
 RAM size:        16 GB

 NB Freq:         4700 MHz
 RAM Freq:        3000 MHz 
 RAM Latency:     16  
 GPU:             2080 Super
 STMark:          3028


I ran 3 passes all with the same settings:

Frames: 25498 - Time: 180000ms - Avg: 141.656 - Min: 95 - Max: 238

Frames: 25340 - Time: 180000ms - Avg: 140.778 - Min: 95 - Max: 238

Frames: 25940 - Time: 180000ms - Avg: 144.111 - Min: 100 - Max: 239

 

Loki this is really interesting. You and I have almost identical setups. Mine is an 8086K at 5.0ghz, and a 2080 SC (not Super, the older 2080 but with EVGA's beefy air cooler on it). But I have my RAM at 3333 CL16 and I'm getting about 146-148 FPS. I did the run on December 12th, is that pre-4.003 ?

 

Anyhow, I'm really interested to see how your 3600mhz ram works out. I have been eyeing some very fast RAM, 4000 or more, but it's only CL18 I think.

Share this post


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

Just only as an experiment, have you tried to enable only 4 or 6 cores in BIOS and see how much fixed OC you can reach. 

Then you can run the IL-2 test, maybe you will surpass the 9900K.

 

 

I did....How do you think i burned my motherboard.:blush: :lol: I did try 8 cores without hyperthreading, but i got worse performance than 8x16. In the end i'll stick with 16x32. It's not worth going back and forth for three frames.

 

And i can't beat a 9900k running 2080Ti with a two and a half year old 1080TI. If you look at the top performers in that list, as you go higher up you get just 2000 series cards and three out of top four performers are 2080TIs.

 

That's what makes the most difference.

 

I'd like to know how the a 9900k/1080Ti compares with my setup. that would be more apples to apples. I think there shouldn't be more than five frames difference. I intend to keep my 1080ti until the RTX 3000 series comes out, or the new AMD card if it's good.

Edited by Jaws2002

Share this post


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

 

Loki this is really interesting. You and I have almost identical setups. Mine is an 8086K at 5.0ghz, and a 2080 SC (not Super, the older 2080 but with EVGA's beefy air cooler on it). But I have my RAM at 3333 CL16 and I'm getting about 146-148 FPS. I did the run on December 12th, is that pre-4.003 ?

 

Anyhow, I'm really interested to see how your 3600mhz ram works out. I have been eyeing some very fast RAM, 4000 or more, but it's only CL18 I think.

Just get some b die 3600CL16 and overclock it. 4000mhz is overpriced for the performance gain you get. You can get 16gb of Gskill TridentZ Neo for under $200 US shipped. I think their ripsjaw 3600CL16 is on sale at Newegg for $75 right now too. It's b die. I got the NEO because I'm a sucker for RGB.

  • Upvote 1

Share this post


Link to post
Share on other sites
On 12/30/2019 at 5:50 PM, Jaws2002 said:

 

I'd like to know how the a 9900k/1080Ti compares with my setup.

Yes, It would be interesting to see the benchmark with people with 9900K or 9900KS at 5.2 or 5.3 and 1080Ti. Who in the forum has that?

Edited by chiliwili69

Share this post


Link to post
Share on other sites

In case anyone's interested, I tried some VR benchmarks tonight on my rig, seeing what different graphics settings do. I have the 3D Migoto mod installed because personally I find it's almost 100% necessary for VR flying. Some more experienced players can ID in VR based on the anemic in-game zoom, I can't. So I need my Migoto mod.

 

Here's the spreadsheet, you can see my specs and the kind of stuff I'm changing between each run. It's Chili's Remagen benchmark track, with a Rift S. https://docs.google.com/spreadsheets/d/1DrO3AiC_jWdfmoJLaNxoEEktTrgYPvTTO1xEAuB-CZU/edit?usp=sharing

 

Possibly interesting datapoints:

  • It's impossible to hit a 100% constant 80 FPS, even with my 8086K @ 5.0ghz and an RTX 2080.
  • Clouds and Shadows both have a fair impact on FPS.
  • HDR appears to have negligible impact, and I like it on to help spotting on my Rift S.
  • Grass distance has no impact.
  • Fullscreen mode is better than Windowed.
  • With Balanced settings where I'm seeking reasonable visual quality and good FPS, my RTX 2080 can bump the supersample from 1.1 to 1.2 pixel density with apparently no loss of performance.

I have some 3600-CAS16 memory on the way, and it's Samsung "b-die" (whatever that means) so I hope to be able to improve memory timings from my current setup of 3333-CAS16. From other folks' testing it might help. Once I get the new memory I'm also going to push the OC on my ring ratio and see the highest stable setting I can get. So basically I have a memory subsystem upgrade planned over the next couple of weeks.

Share this post


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

In case anyone's interested, I tried some VR benchmarks tonight on my rig, seeing what different graphics settings do. I have the 3D Migoto mod installed because personally I find it's almost 100% necessary for VR flying. Some more experienced players can ID in VR based on the anemic in-game zoom, I can't. So I need my Migoto mod.

 

Here's the spreadsheet, you can see my specs and the kind of stuff I'm changing between each run. It's Chili's Remagen benchmark track, with a Rift S. https://docs.google.com/spreadsheets/d/1DrO3AiC_jWdfmoJLaNxoEEktTrgYPvTTO1xEAuB-CZU/edit?usp=sharing

 

Possibly interesting datapoints:

  • It's impossible to hit a 100% constant 80 FPS, even with my 8086K @ 5.0ghz and an RTX 2080.
  • Clouds and Shadows both have a fair impact on FPS.
  • HDR appears to have negligible impact, and I like it on to help spotting on my Rift S.
  • Grass distance has no impact.
  • Fullscreen mode is better than Windowed.
  • With Balanced settings where I'm seeking reasonable visual quality and good FPS, my RTX 2080 can bump the supersample from 1.1 to 1.2 pixel density with apparently no loss of performance.

I have some 3600-CAS16 memory on the way, and it's Samsung "b-die" (whatever that means) so I hope to be able to improve memory timings from my current setup of 3333-CAS16. From other folks' testing it might help. Once I get the new memory I'm also going to push the OC on my ring ratio and see the highest stable setting I can get. So basically I have a memory subsystem upgrade planned over the next couple of weeks.

B die refers to memory that uses Samsung chips. Its, or it was (others are making ram chips equal to it now apparently but at a higher price) the best for quality and overclocking tolerance. 

 

As much as I appreciate this thread and the work going on, the only actual improvement we see in VR is going to come from the devs updating a decade old game engine to work efficiently with current hardware. Who knows if that ever happens. I've yet to even see them address it or offer advice. Until then we are going to be left struggling with FPS and spending hundreds of dollars trying to squeeze out 5 more FPS with high end machines at low/medium settings.

 

You'd think that they'd at least chime in rather than letting their customers do all the work for them trying to make the game tolerable in VR. If I had my customers running tests and problem solving for my business with zero input from myself I would be embarrassed. We may be a minority but we are still a dedicated group who pay for and support their product despite the fact we are offered no help. If anything, with the growing popularity of VR, they will benefit from the work you guys are doing with these benchmarks. We appreciate you.

Edited by Loki_1982b_rock1080
  • Upvote 2

Share this post


Link to post
Share on other sites

I have done some upgrades to my PC (GPU and SSD) and run the test before and after. It appears neither hardware have any measurable effect in this test as expected.

 

IL-2 version:    4.003
Motherboard:    Asus Z97-A
Hard Drive:    Samsung SSD 840 EVO 250GB (regular SSD)
CPU:        4790K
CPU Freq:    4.7 Ghz
L3 cache:    8 MB

NB Freq:    4000 MHz
RAM type:    DDR3
RAM size:    16 GB
RAM Freq:    2666 MHz
RAM Latency:    11
GPU:        GTX 1070
STMark:        2829


RESULTS:
2019-12-21 10:39:36 - Il-2
Frames: 22381 - Time: 180000ms - Avg: 124.339 - Min: 86 - Max: 226

 

IL-2 version:    4.003
Motherboard:    Asus Z97-A
Hard Drive:    Samsung SSD 840 EVO 250GB (regular SSD) >> Adata SX8200 Pro (M.2 NVME SSD)
CPU:        4790K
CPU Freq:    4.7 Ghz
L3 cache:    8 MB

NB Freq:    4000 MHz
RAM type:    DDR3
RAM size:    16 GB
RAM Freq:    2666 MHz
RAM Latency:    11
GPU:        GTX 1070 >> GTX 1080 TI
STMark:        2829


RESULTS:
2019-12-26 14:21:59 - Il-2
Frames: 22535 - Time: 180000ms - Avg: 125.194 - Min: 80 - Max: 221

Share this post


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

I have done some upgrades to my PC (GPU and SSD) and run the test before and after. It appears neither hardware have any measurable effect in this test as expected.

 

Hmm. The NVMe drive won't help your FPS, as I've said in other threads (I'm not sure why some people keep recommending them -- they do nothing for gaming performance unless it's Star Citizen or something with horrible disk access patterns). But the GPU upgrade should help you in VR, for sure. This test deliberately has low settings so that the CPU is more of a bottleneck than the GPU.

 

Sadly I think the 4.7ghz CPU with the 2666 DDR3 is likely what's holding you back the most at this point, and fixing that would require new motherboard, CPU and RAM, so it's a pricey upgrade.

Share this post


Link to post
Share on other sites
5 hours ago, Alonzo said:

 

Hmm. The NVMe drive won't help your FPS, as I've said in other threads (I'm not sure why some people keep recommending them -- they do nothing for gaming performance unless it's Star Citizen or something with horrible disk access patterns). But the GPU upgrade should help you in VR, for sure. This test deliberately has low settings so that the CPU is more of a bottleneck than the GPU.

 

Sadly I think the 4.7ghz CPU with the 2666 DDR3 is likely what's holding you back the most at this point, and fixing that would require new motherboard, CPU and RAM, so it's a pricey upgrade.

 

Yes i would have been surprised to see any improvement with the new SSD. My intention was bigger space not more FPS and wanted to try this tech so went with it. And I know Chilli's intention was to get the GPU out of the equation for the benchmark. It appears to be working well.

 

Next thing is CPU-MB-RAM update indeed, and this benchmark thread is perfect to see the best options.

Share this post


Link to post
Share on other sites
On 12/31/2019 at 7:20 AM, Alonzo said:
  • Clouds and Shadows both have a fair impact on FPS.
  • HDR appears to have negligible impact, and I like it on to help spotting on my Rift S.
  • Grass distance has no impact.
  • Fullscreen mode is better than Windowed.

 

Thanks for sharing and documenting it so well.

HDR and Grass is aligned with previous tests. Negligible impact.

The Fullscreen is a nice discovery, I have to try it as well.

The 3Dmigoto mod has a small impact 3-5 fps.

 

I think any of the test correspond to the settings of the OP, so I can not upload your VR test to the table.

 

Let´s see how the RAM goes.

 

 

 

On 12/31/2019 at 8:31 AM, Loki_1982b_rock1080 said:

As much as I appreciate this thread and the work going on, the only actual improvement we see in VR is going to come from the devs updating a decade old game engine to work efficiently with current hardware. Who knows if that ever happens. I've yet to even see them address it or offer advice. Until then we are going to be left struggling with FPS and spending hundreds of dollars trying to squeeze out 5 more FPS with high end machines at low/medium settings.

 

You'd think that they'd at least chime in rather than letting their customers do all the work for them trying to make the game tolerable in VR. If I had my customers running tests and problem solving for my business with zero input from myself I would be embarrassed. We may be a minority but we are still a dedicated group who pay for and support their product despite the fact we are offered no help. If anything, with the growing popularity of VR, they will benefit from the work you guys are doing with these benchmarks. We appreciate you

 

Honestly I think I will be always in debt with the 1C/777 for bringing me to the flight sim again thanks to VR, otherwise I would not play anything else routinely.

I think they know well how important is (and will be) VR in the flight sims (Look at the first edition of Flight Sim Expo where they brighted thanks to VR) and they are perfectly conscious about the limitations of the hardware and the game code.

It is a small company with limited resources and many items to address/develop. It is impressive what they have developed along the years and how this great community has helped them in reaching their goals.

I know that the Dev team got an HP Reverb for Christmas, so they are on the same boat than us, trying to squeeze every single fps out of IL-2. ;)

  • Like 2
  • Upvote 2

Share this post


Link to post
Share on other sites
On 12/31/2019 at 10:15 PM, HunDread said:

Next thing is CPU-MB-RAM update indeed, and this benchmark thread is perfect to see the best options.

 

Thank you for your test. It confirms that even with a 1070 it is almost not bottlenecked by GPU.

I also plan to do a CPU-MB-RAM+Windows upgrade this year (so this thread is not purely altruistic... :blush:)

Share this post


Link to post
Share on other sites
Posted (edited)

S!

 

Adding my results, hope helps.

 

 Motherboard: Asus ROG Crosshair VIII HERO
 CPU:                 3900X
 CPU Freq:        3.7 Ghz
 L3 cache:         4 x 16 MB

 Cores:              12 (number of active physical Cores)

 HT:                    Off
 RAM type:        DDR4
 RAM size:        32 GB

 NB Freq:           1862.9 MHz
 RAM Freq:        3733 MHz (this is 2xDRAM freq if Dual channel)
 RAM Latency: 18  (CAS Latency CL)
 GPU:                 2080Ti
 STMark:           3076

 

 

IL-2 results(NVCPL settings on default, game settings as in OP)

 

2020-01-03 12:25:25 - Il-2
Frames: 27812 - Time: 199031ms - Avg: 139.737 - Min: 90 - Max: 238

Edited by LLv34_Flanker
  • Upvote 1

Share this post


Link to post
Share on other sites
Posted (edited)

Hi all.

 

Here are my results to compare 3900X

 

IL-2 version:      4.003

Motherboard:    Gigabyte X570 Aorus Pro

CPU:                    Ryzen 3900X

CPU Freq:           4.4 Ghz

L3 cache:           4x16 MB

Cores:                 12

HT:                       On

RAM type:          DDR4

RAM size:           32 GB

NB Freq:             1863 MHz

RAM Freq:          3733 MHz

RAM Latency:    16 (CAS Latency CL)

GPU:                    RTX 2080

STMark:              3063

 

I ran 3 passes at same settings to ensure no anomalies.

 

Run 1
Frames: 25003 - Time: 180000ms - Avg: 138.906 - Min: 95 - Max: 238

Frames: 25514 - Time: 180000ms - Avg: 141.744 - Min: 95 - Max: 238

Frames: 25271 - Time: 180000ms - Avg: 140.394 - Min: 95 - Max: 238

Edited by ICDP
Added Il-2 version
  • Upvote 1

Share this post


Link to post
Share on other sites
Posted (edited)

That's interesting.:blink: I guess Intel CPUs do make a difference. All our Ryzen 3000 CPUs are withing a frame from each other  with three different graphics cards.

I was expecting the GPU to have much higher influence.

Edited by Jaws2002

Share this post


Link to post
Share on other sites
3 minutes ago, Jaws2002 said:

That's interesting.:blink: I guess Intel CPUs do make a difference. All our Ryzen 3000 CPUs are withing a frame from each other  with three different graphics cards.

I was expecting the GPU to have much higher influence.

 

Only the Intel i9 9900K clocked at 5.2- 5.3 GHz is markedly superior by ~25% but they are similarly faster compared to 9700K or 8700K for example.  Also the i7s show a similar score with 2080Ti compared to 2080 S or even 2070 for example.

 

I'm going to try with SMT (hyperthreading off)

Share this post


Link to post
Share on other sites
4 minutes ago, ICDP said:

 

Only the Intel i9 9900K clocked at 5.2- 5.3 GHz is markedly superior by ~25% but they are similarly faster compared to 9700K or 8700K for example.  Also the i7s show a similar score with 2080Ti compared to 2080 S or even 2070 for example.

 

I'm going to try with SMT (hyperthreading off)

 

I tried SMT off, and it made no difference. I got worse performance running 16 cores/16 threads than 16 cores 32 threads. I got better performance disabling one chiplet and running the CPU at 8 cores and 16 threads. 

 

Share this post


Link to post
Share on other sites
7 minutes ago, Jaws2002 said:

 

I tried SMT off, and it made no difference. I got worse performance running 16 cores/16 threads than 16 cores 32 threads. I got better performance disabling one chiplet and running the CPU at 8 cores and 16 threads. 

 

 

Too late, I tried and as you stated... no difference :)

Share this post


Link to post
Share on other sites
9 hours ago, LLv34_Flanker said:

CPU:                 3900X
 CPU Freq:        3.7 Ghz

 

Thank you Flanker for your test. It is great to test a 3900X paired with a 2080Ti.

 

I would like to know if your 3.7GHz is a fixed frequency or it is set by the system depending on load/temperature/etc.

Have you tried to enable only 6 cores and run it at a higher fixed frequency? (assuming your CPU cooler can keep it below 70 celsius)

 

It is also good to know the NB frequency of the AMD CPUs, it is quite low comparing to the Intel. We still don´t know the effect over IL-2.

Share this post


Link to post
Share on other sites
Posted (edited)
2 minutes ago, chiliwili69 said:

 

 

 

It is also good to know the NB frequency of the AMD CPUs, it is quite low comparing to the Intel. We still don´t know the effect over IL-2.

 

You can add the NB frequency for mine as well. It's running at 1800Mhz.

 

And the 3.7 GHz is the base clock speed for 3900x. It will boost to 4.6 i think.

Edited by Jaws2002

Share this post


Link to post
Share on other sites
56 minutes ago, ICDP said:

CPU:                    Ryzen 3900X

CPU Freq:           4.4 Ghz

Many Thanks.

All the three 39X0X chips deliver same fps regardless of the CPU frequency. As I said, I don´t know if it is a fixed freq or not.

 

The GPUs has no net effect (1080Ti, 2080 and 2080Ti) deliver almost same results. So, it is good to see that this benchmark really took the GPU out.

55 minutes ago, Jaws2002 said:

I was expecting the GPU to have much higher influence

 

This benchmark was intended to not be influenced by GPU. So we can concentrate analysis just in CPU/RAM.

It would be nice if someone with a 9900K at 5.2GHz and a 1080 or 1080Ti or 2070 could run this test.

  • Upvote 1

Share this post


Link to post
Share on other sites

Lot of variables across GPU's that has to be taken into account.

GPU clock, Boost Clock which can vary of course on temp along with overclocking. Memory speed and whether it is overclocked

 

PC's - just about every machine is different and has it's own uniqueness.

 

Share this post


Link to post
Share on other sites
2 minutes ago, dburne said:

GPU clock, Boost Clock which can vary of course on temp along with overclocking. Memory speed and whether it is overclocked

 

Yes, all the GPU variables you mention affects the IL-2 VR performance, but in this benchmark test for monitor it should not influence.

 

You can try all kind of variables for your GPU, but with this test in monitor, since your GPU is not bottlenecked you will get always the same results.

 

So, that´s why we don´t log GPU variables.

Share this post


Link to post
Share on other sites
7 minutes ago, chiliwili69 said:

 

Yes, all the GPU variables you mention affects the IL-2 VR performance, but in this benchmark test for monitor it should not influence.

 

You can try all kind of variables for your GPU, but with this test in monitor, since your GPU is not bottlenecked you will get always the same results.

 

So, that´s why we don´t log GPU variables.

 

I would certainly think it would effect in monitor also. 

 

But nonetheless speaking of VR there are really only two things that come into play. Are you going to be running at full capability, ie 90 hz, or 80 hz, whatever hz your particular device runs at. Or are you going to be below that with some form of reprojection/ASW ( 40 or 45 hz interpolated). 

Whether one gets 60 fps, 65 fps, 70 fps, etc has no effect The fps will be 40/45 interpolated to 80/90.  Obviously the goal would be able to run with no reprojection/ASW at any time.

But that can not be done with today's technology with this game.  We can get closer now, really close but there will be times it drops down into

interpolating those frames, that can not be helped with a game as demanding as this one.

Of course one can disable ASW or Reprojection, and see their fps run in between what ASW would be. But then may induce some stuttering if much below the 80/90 fps range.

 

Best thing one can do, get the most rig they can for the money they can spend. Lots of info out there on the more popular boards, CPU, Ram, and GPU. I researched heavily when I built this rig almost a year ago, and have been very pleased. I got the best I felt I could get at the time in CPU, GPU, Ram, Motherboard, and even Power Supply. And it is kicking some pretty good butt for me.

 

Shoot for smoothness in VR play, turn the FPS counter off. It means nothing. As long as you can run between min (40/45) and max (80/90) you are good.  Focus on smooth.  Adjust graphic settings if it is not as smooth as one would like, or if you would like to be closer to no ASW/Rreprojection.

 

Don't sweat it too much, get it where it looks nice then forget it. Load up the game or games, and have a blast!!

 

 

Share this post


Link to post
Share on other sites

I just overclocked my 2080 and in GPU benchmarks it was ~8% faster but in this test it was margin of error.  So Chili is correct, the GPU has marginal (if any) effect.

Share this post


Link to post
Share on other sites

S! 

 

I have no overclocking or tweaking done on my system, not CPU or GPU. Just set the OCDP profile for memory to run at rated 3733MHz and the Fabric in coupled mode aka 1800MHz. Disabled PBO as well. 

 

The GHz is not fixed, just took it from the CPU-Z. It varies between 3.7GHz to 4.6GHz. I have a cooler comparable to Noctua NH15, so temps are no issue. 

 

I wonder why Grass is set to maximum value in the test? Does not even draw until you fly so low you risk bending the propellor. I run it at Normal while gaming. 

 

What I noted from comparing the test and gaming, I get better FPS?! Never drops below 100fps, hovers more closer to 140-200fps. I wish I knew more how to tweak my system, the motherboard for sure has a plenty of options to do so.. 

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...