What do you think can cause the "semi-freezing/jumpy" display

Paco572 said:
I know SLI does work for rF2 in DX11 for me because my frame rates when up about 40%

PACO
Click to expand...

Well, you have claimed that SLI does work before but when I asked you to help others and post your settings in this post: https://forum.studio-397.com/index.php?threads/gtx-1080-ti-struggles.56872/page-5#post-905095 you have been silent. I know that SLI is working but I doubt that you have micro-stutter, stutter free game experience. To proof me wrong all you have to do is post your settings and we see if SLI is working as it should or not. If you don`t just zip it please. o_O
 
Ari Antero said:
When I read Stonec's Observation 1 I was forced to wipe off the dust from my Z77 & I5 3770k@3,5 and run few test just to look how high BUS Usage is in rF2. OS in all rigs is Win 10 Pro.

ASRock Z77 Extreme 6, CPU: I7 3770k @3,5 Ghz : Bus usage: 40-50%, CPU usage: 20-25%
ASRock Z87 OC Formula, CPU: I7 4770k @3,5 Ghz: Bus usage: 20%-25%, CPU usage 15-20%
Asus X99 WS-E/10G, CPU: i7-5960X @3,5 Ghz: Bus usage: 10-15%, CPU usage 10-15 %
Click to expand...

Are those chipsets all using PCIe 3.0 @ x16? Interesting to see there is so much variation in CPU and bus usage. The 40-50% is a lot, I had my PCIe in wrong mode and at 50-60% it already started bottlenecking. When I tried to OC my GPU with the PCIe bottleneck, it wouldn't add any more FPS to the game (with 20 cars on grid, the more opponents, the more PCIe bottlenecks).
 
stonec said:
Are those chipsets all using PCIe 3.0 @ x16? Interesting to see there is so much variation in CPU and bus usage. The 40-50% is a lot, I had my PCIe in wrong mode and at 50-60% it already started bottlenecking. When I tried to OC my GPU with the PCIe bottleneck, it wouldn't add any more FPS to the game (with 20 cars on grid, the more opponents, the more PCIe bottlenecks).
Click to expand...

Yes all chips are using PCI-e @x16 Test is done with in-game max settings, PP medium and 22 AI. I was using single GFX card, Z77>780Ti, Z87>980Ti and X99>1080Ti.
 
Ari,
My GPU bus is the same as your, never exceeding 20%.
I did remove the PCIE>>USB card for a while to test, but I can't yet get the problem if Bodnar Config Software is not running.
So need more test.
 
Ari Antero said:
Yes all chips are using PCI-e @x16 Test is done with in-game max settings, PP medium and 22 AI. I was using single GFX card, Z77>780Ti, Z87>980Ti and X99>1080Ti.
Click to expand...

Hi Ari,

What frame rates were you getting with these 2 systems. An obvious observation is that PCIE BUS goes up quiet a bit when the system is struggling with FPS.

A brief test i did, was stressing with a RACE start 24 cars at Nola with some 4k skins. when riding with car at front of grid PCIE bus would drop down to 20-30% region with frames 70-100fps. riding at the back of the field PCIE usage would shoot up to 50% and fps down to 47-58fps

as @stonec pointed out there seems to be a large communication with system RAM, but also with Virtual Ram. my usage figures were around 3-4gb VRAM, System ram 7.6gb and a Wopping 11gb of Virtual RAM (paging)

I do wonder if this is all a limitation of the game engine and how it handles loading in textures, shaders etc

I will post a more depth reply after doing more testing.
 
Daniel Surace said:
Hi Ari,

What frame rates were you getting with these 2 systems. An obvious observation is that PCIE BUS goes up quiet a bit when the system is struggling with FPS.

A brief test i did, was stressing with a RACE start 24 cars at Nola with some 4k skins. when riding with car at front of grid PCIE bus would drop down to 20-30% region with frames 70-100fps. riding at the back of the field PCIE usage would shoot up to 50% and fps down to 47-58fps

as @stonec pointed out there seems to be a large communication with system RAM, but also with Virtual Ram. my usage figures were around 3-4gb VRAM, System ram 7.6gb and a Wopping 11gb of Virtual RAM (paging)

I do wonder if this is all a limitation of the game engine and how it handles loading in textures, shaders etc

I will post a more depth reply after doing more testing.
Click to expand...

My test was done In Silverstone with GT500, max in-game settings, weather raining, 22 AI. Video settings, AA 5 and PP medium. Monitor resolution 3440x1440. Reason I was using PP at medium is that 780Ti has 3Gb of memory and High and Ultra is simply too high to that size of memory when all other settings are at the maximum.

ASRock Z77 Extreme 6, CPU: I7 3770k @3,5 Ghz, GTX 780Ti > 40-70 fps
ASRock Z87 OC Formula, CPU: I7 4770k @3,5Ghz, GTX 980Ti > 60-90 fps
Asus X99 WS-E/10G, CPU: i7-5960X @3,5 Ghz, GTX 1080Ti > 160-190 fps

Z77 and Z87, PC ram usage is close and it is almost similar ~ 5,2 GB, Virtual ram ~ 8,1 GB.
Z77, 780Ti >VRAM usage 2,8-2,9 GB
Z87, 980Ti > VRAM usage, 3,4-3,5 GB
It is bit unfair to compare my Asus X99 WS-E/10G, CPU: i7-5960X @3,5 Ghz, GTX 1080Ti with my two other PC`s becourse it has
2x Vapochill Lightspeed - Vapor Compression Cooling , not cooling only CPU but also north and south bridge, 3x 1080Ti and temps are all the time round -30C. Test was made with single 1080Ti and to remove two 1080Ti was a huge task which I am not going to do again in near future and the same applies to my Z77 and Z87 which has also two-way SLI and are water cooled.:oops:
X99, PC ram usage ~4,4 GB, Virtual ram 6,8 GB.
VRAM usage, 4,2-4,4 GB

With single screen I can not see that there is a problem with performance if S397/ISI content and reasonable in game and video settings are used with ordinary PC. Z77 and Z87 GPU loads are 90-100% and X99 98-100% with S397/ISI content. However if triple screens are used there is not so much headroom to wet weather racing no mater how powerful PC is used. Would be nice to have SMP support to 10xx cards which could give 15-30% better performance to triple screen users but it seems that S397 is ignoring larger group off users and giving all the attention to VR users and does not realize that not everybody can use VR becourse of health issues or simply becourse VR is too expensive to many who already has triple screens and 10xx gfx card. I doubt Monitoring Steam stats gives the right picture how man of rFactor 2 users has 10xx card.
The biggest problem is rF2 3p content which is not optimized to DX11. When fps drops GPU load does not increase it drops below monitor hz to ~40% and GPU load goes up first when fps does increase above monitor hz but it don`t never go higher then ~70%. I can not understand why GPU load is dropping when hardware does not bottleneck.

Any toughs?
 
Last edited:
Daniel Surace said:
as @stonec pointed out there seems to be a large communication with system RAM, but also with Virtual Ram. my usage figures were around 3-4gb VRAM, System ram 7.6gb and a Wopping 11gb of Virtual RAM (paging)

I do wonder if this is all a limitation of the game engine and how it handles loading in textures, shaders etc

I will post a more depth reply after doing more testing.
Click to expand...

Interesting. I wonder if it's still duplicating GPU VRAM in main RAM (+virtual memory), as that would explain the huge system RAM requirements and the bus load. DX9 has mandatory duplication of everything, but it becomes very impractical these days when we have GPU's with 11 GB VRAM to move anything to system RAM. When I run other sims like Assetto Corsa on my 4 GB GPU, there is hardly any system RAM usage going on, and anyway system RAM is always order of magnitude slower than VRAM.

All this might explain why frame rate drops so much when there is lots of content on the screen (worst case scenario start, lots of AI cars and rain spray). If the game is dependent on the BUS and CPU speed to move things, then not even the strongest GPU will help much.
 
stonec said:
Are those chipsets all using PCIe 3.0 @ x16? Interesting to see there is so much variation in CPU and bus usage. The 40-50% is a lot, I had my PCIe in wrong mode and at 50-60% it already started bottlenecking. When I tried to OC my GPU with the PCIe bottleneck, it wouldn't add any more FPS to the game (with 20 cars on grid, the more opponents, the more PCIe bottlenecks).
Click to expand...
What were those settings and what did you change them to?
 
Look here guys this is with GT500 at Silverstone v1.6
X99, PC ram usage ~4,4 GB, Virtual ram 6,8 GB.
VRAM usage, 4,2-4,4 GB

This is with new MClaren at Silverstone v1.6
X99, PC ram usage ~6,1 GB, Virtual ram 10,2 GB.
VRAM usage, 3,7 GB-3,8 GB

Lower VRAM usage but so much higher PC-and Virtual ram usage.
How can it be like this?
 
suryaprihadi said:
This troubleshooting is really fun ;)
Thanks to all who gave input.

Too bad I am not IT smart, but troubleshooting I do enjoy.

I was observing what SKAN wrote :
https://forum.studio-397.com/index....emi-freezing-jumpy-display.57025/#post-905877
- Was windows downloading one of its updates during this time?
- That same question can apply to much of the rest of your software.
I don't allow auto update on OS or any other software. I dont know if this is true, but it seems RF2 when running, will freeze any data download on the internet, it seems to monopolize internet bandwith.

- Drive speed and fragmentation could also be a factor.
I use M2 SSD for OS and SATA-SSD for steam library. So these two drives cant even rest, its working all the time. I know HDD sometime can hibernate/sleep. I restarted my Steam and PC every time I start to play on this PC, it is not on sleep mode.

- There is also the Microsoft Indexing Service which keeps track of every file and other things. This process alone can cause these problems if you have a lot of files or you move/copy/delete files often. This process requires the drive or drives be accessed even if you are doing something important as rF2.
This one got me thinking ..Thank you SKAN :) .. not due to SSD data transfer disturbance wise, but CPU wise , because Window Search can sometime take a lot of CPU ( 1 thread only maybe ? ).

Look at how taxing Window Search can be :

Below two screen captures are :
1 - RF2 at 10% overall CPU load, while Window Search is 6%.
2 - RF2 at 11% overall CPU load, while Window Search is 8% ...dang !!
NOTE : Look at the internet clock I place there to match system time with Afterburner data logging time and camera video logging time, that clock stop the moment RF2 started. Hence I assumed RF2 block other internet activity when it starts.
View attachment 9413
.
View attachment 9414
Too bad I do not know which CPU core/thread is being used by Window Search.

I also do not know if RF2 trigger Window Search to look for files ? ...or Window Search is doing its
regular thing. I did reset the RF2 ( no PC restart ) and play again, the Window Search stop using CPU resources... LOL.

Window Search consuming CPU power is a common problem :
https://www.sevenforums.com/performance-maintenance/168177-sudden-high-cpu-usage.html


ANYWAY.........
With other background services/application which could run without me knowing, its time to take a look at how RF2 appoint CPU resources , I must learn some. Been testing a few times. RF2 always appoints my CPU core no 1 thread 1 ( Afterburner call it CPU1, Resource Monitor call it CPU Zero ) as main game task, and CPU core 2 thread 1 for other resources (Afterburner call it CPU 3, Resource Monitor call it CPU 2 ).

Below graph is not of the same time line as screen recording of Resource Monitor, but it represent RF2 being played for 635 seconds.
View attachment 9415
.
View attachment 9416
.
View attachment 9417

This is the highest average CPU load , 25% and greater. But none last more than 2 seconds.
The data logging is 2 data points per 1 second.
View attachment 9418

I logged for 635 seconds, if I sort out lowest to highest average CPU use, discounting the time it happened, it looks like this, to get easy visual look-see of the average CPU use.
View attachment 9419

Having experience such a simple Configuration Software from Bodnar can cause issue when RF2 is running due to probably some glitch somewhere.... it makes me wonder.

AA. Can Windows7 ( my OS ) while doing its own application/services/yada-yada ... once in a while, make a mistake appointing CPU resources which may clash with RF2 choosen CPU thread ? Who has priority here, RF2 or Windows7 ?

BB. How about other parasitic load of extra hardware : steering, pedals, gearbox and whatever extra hardware for RF2 game which needs 2 way communication with RF2 and need some CPU resources.
Who is the CPU management boss here, RF2 or Windows7 ?

Its kinda sad and fun at the same time. I wanted simply to play RF2, but I ended up needing to learn more of IT stuff, whereby I am an IT dummy....LOL.

I realized now the complication of needing to do real time process for a game with high graphic needs and having multiple extra hardware it has to communicate with, steering , pedals and all those things.
When I edit/render 4K video, my GPU and CPU will throttle up very high, but its not real time, the hardware can take its own sweet time doing the work, more power is only faster process.
A game cant have delay in its data processing, it has to be real time, if not we then see stutter ...LOL.
Learning...learning....never ends.

Again thank you very much to everyone....
Click to expand...
I have exactly the same issue as the OP, but on my end it is clearly a problem with the CPU utilization, same as shown in our posted graphs btw. It is quite easy to see that CPU 1 reaches 100% quite a few times and that's the point when the game is running out of real time and starts warping.
 
You must log in or register to reply here.
Back
Top