Whitestar
Premium
- 2,103
- Norway
- Whitestar127
Yes!
PC Hardware | video settings | benchmarksPC
- Thread starter Whitestar
- 385 comments
- 42,198 views
- 328
- Lisbon
- DrummerManiac
- 2,428
- England
- Lewis_Hamilton_
@drummermaniac
You'll have no problem with that setup and those graphics settings, you can push the resolution even higher if you're using a single screen. Like @Harry6784 said, for triple screens, you'll have to turn post processing off or turn down the reflection and shadow settings to medium, the higher settings for all of these are Demanding with a capital D.
You'll have no problem with that setup and those graphics settings, you can push the resolution even higher if you're using a single screen. Like @Harry6784 said, for triple screens, you'll have to turn post processing off or turn down the reflection and shadow settings to medium, the higher settings for all of these are Demanding with a capital D.
- 4,022
- Cincinnati, OH
Will run triples no problem. If you overclock the i5 you could probably run the max number of AI as well without issue. I just got my i5 to 4.8Ghz only using 1.300v. With stock Voltage you can overclock it to 4.5 easy.
- 206
- United States
- boxster233
I have 5400 rpm. What impact would an ssd have for assetto?
- 28,470
- Windsor, Ontario, Canada
- Johnnypenso
Significantly reduced loading times.
Xiamoto
Premium
- 673
- United States
Thanks for your reply. So after doing some testing with MSI afterburner, I'm even more confused..
If you look at my graphic below, my GPU (GTX 1070) seems to be bottlenecking when running 15 Ai cars (70fps) at Road America, but neither the GPU or the CPU are bottlenecking when running 48 cars (40fps) at Road America. Any thoughts on why my GPU and CPU won't work harder to get back to a minumum of 60fps? 
What is going on here
, Cheers!
Last edited:
- 716
It is easy, the game does not load your cpu more than that. Ie it is not that demanding.
Like you see at 15 cars the gpu is working nicely and it makes the cpu working hard to because it need to keep up with the gpu. higher fps=higher cpu load in AC
And when you have more cars on track the gpu struggles so much more that the fps goes down and so will the workload of the cpu go down. Like I said, in AC high fps=high cpu load. That is if you are not running an old slow cpu like a q6600 or similar that will show high % even at 30fps
Like you see at 15 cars the gpu is working nicely and it makes the cpu working hard to because it need to keep up with the gpu. higher fps=higher cpu load in AC
And when you have more cars on track the gpu struggles so much more that the fps goes down and so will the workload of the cpu go down. Like I said, in AC high fps=high cpu load. That is if you are not running an old slow cpu like a q6600 or similar that will show high % even at 30fps
Last edited:
Xiamoto
Premium
- 673
- United States
It is easy, the game does not load your cpu more than that. Ie it is not that demanding.
Like you see at 15 cars the gpu is working nicely and it makes the cpu working hard to because it need to keep up with the gpu. higher fps=higher cpu load in AC
And when you have more cars on track the gpu struggles so much more that the fps goes down and so will the workload of the cpu go down. Like I said, in AC high fps=high cpu load. That is if you are not running an old slow cpu like a q6600 or similar that will show high % even at 30fps
The AC in-game FPS monitoring displays CPU usage at 90% percent during the 48 car test, with an occasionally 99% CPU warning. What would account for the difference between in-game 90% CPU usage vs. MSI Afterburner 49% CPU usage
.......and do you have an opinion on what hardware upgrade I should consider to attain 60 FPS at Road America with 48 cars? Maybe (2)-GTX1070 SLI's? Thanks and Cheers 👍.
- 716
hmm strange, do you have the latest msi afteburner? It shows exactly the same % figures as the AC app for me.
Is it with stuff on in the background? because with Firefox on in the background and listening to web radio the cpu load also goes up to 90% with my skylake and a gtx 970@over 80-90fps. Road America is a modtrack I guess, try running on nurburing.
No do not do anything rash. If you do get 99% load warnings you need to first see if there is a performance hog program running in the background so there is nothing going on that steals your cpu performance. If nothing strange is going on in the background then you actually need to overclocking your cpu, that is if you are still on the 2600k, But was this with a triple screen setup?
But give me the link on how to archive the 48cars on track so I cant test it out myself, cant seem to find anything about that myself.
Is it with stuff on in the background? because with Firefox on in the background and listening to web radio the cpu load also goes up to 90% with my skylake and a gtx 970@over 80-90fps. Road America is a modtrack I guess, try running on nurburing.
No do not do anything rash. If you do get 99% load warnings you need to first see if there is a performance hog program running in the background so there is nothing going on that steals your cpu performance. If nothing strange is going on in the background then you actually need to overclocking your cpu, that is if you are still on the 2600k, But was this with a triple screen setup?
But give me the link on how to archive the 48cars on track so I cant test it out myself, cant seem to find anything about that myself.
Last edited:
Xiamoto
Premium
- 673
- United States
Yes, I'm on the latest MSI Afterburner, still on the i7-2600k and have done some overclocking up to 4.8Ghz. Now OC'd @ 4.4 Ghz to regain some system stability. These tests were conducted w/ 60hz/2ms triple screens @ 5760x1080. Road America track mod w/48 pit stalls here: http://www.racedepartment.com/downloads/road-america.7095/. I'll check again on background processes, thanks!
- 716
haha nah man, it is to much for my skylake too, the ingame cpu load is to high and I get stuttering, even got a high cpu warning. But it seems the overall cpu load in win and the msi afterburner is waaay lower... Maybe the game only uses max 4 threads? But in windows cpu task manager says all threads are in use but only to about 50% on each thread, hmm. Gonna ask my friend with a older 8c/16t cpu how this track with all ai cars is performing, who knows maybe it really needs more cores.
So if you get stuttering then then the cpu really needs some more ocing. try 1.35-1.4 vcore to see if 4.8 will be more stable if you have not already tested it at that. Try lowering the graphical settings. 48cars@a triple monitor setup even with the gtx1070 is what gets you under 60fps.
My game stuttered even though the fps was 60-70 fps avg and 55 minimum on a single monitor.
edit: I have been talking with my friend that has an older i7 8core cpu. The 48 cars seems not to use his cpu quite heavy too, 45-48%load in win cpu taskmanager. He is using a VR headset so he did only see about 70fps and said that he did not feel any stutters but that he felt that the fps was low. But to me it sounds like the same problem here although with out the VR heatset
Baah, hard to know when he uses VR....
Last edited:
Xiamoto
Premium
- 673
- United States
Thanks for your research! It interesting that your Skylake and your friends 8c/16t is behaving just like my Sandy Bridge with CPU loads at about only 50%. Here's two baseline tests I just completed:
TEST 1
1) Track - Road America
2) # of cars - 48
3) AC Settings - ALL set to 100% maximum
4) Avg FPS - 38
5) Avg GPU load - 84%
6) Avg CPU load - 46%
7) # CPU core/threads being utilized - 4/8
TEST 2
1) Track - Road America
2) # of cars - 6
3) AC Settings - ALL set to 100% maximum
4) Avg FPS - 51
5) Avg GPU load - 100%
6) Avg CPU load - 34%
7) # CPU core/threads being utilized - 4/8
Again, thanks for all of your efforts to help! 👍
EDIT: On another note, I just checked CPU load for each individual core/thread. One of my 8 threads is averaging 85% load while the rest are averaging well below 50% during the above two tests.
Last edited:
- 2,428
- England
- Lewis_Hamilton_
I see my post at the bottom of the last page was completely glossed over then 
It answered all your questions. Assetto Corsa will use more than 4 cores but it will not take advantage of hyperthreading (and this is probably why you're not seeing 100% usage in windows). AC pummels one core more than the others too, as is often typical in many games.
It answered all your questions. Assetto Corsa will use more than 4 cores but it will not take advantage of hyperthreading (and this is probably why you're not seeing 100% usage in windows). AC pummels one core more than the others too, as is often typical in many games.
Xiamoto
Premium
- 673
- United States
I see my post at the bottom of the last page was completely glossed over then
It answered all your questions. Assetto Corsa will use more than 4 cores but it will not take advantage of hyperthreading (and this is probably why you're not seeing 100% usage in windows). AC pummels one core more than the others too, as is often typical in many games.
Ah, OK, just went back and read it......
My System Bios will let me change the multiplier of each core. Do you have any knowledge of if changing each multiplier might improve performance, or which CPU (sandy, haswell, broadwell, xeon, etc.) tends to perform best when just one... core is getting hammered with the game? Cheers 👍
Xiamoto
Premium
- 673
- United States
Well I headed over to the AC and Steam Forums, got some input and came up with some assetto_corsa.ini file edits that have added 4-5 FPS and re-allocated my Sandy Bridge CPU core/thread loads. Here's the changes I made to the assetto_corsa.ini file:
[THREADING]
USE_TIMER_PROCESS=0
SET_THREAD_AFFINITY_MASK=1
[PHYSICS_THREADING]
THREADS=3
MSI Afterburner (Road America with 48 AI cars) now shows Core #1 of my CPU is under 100% load (none of my cores were above 55% load prior to these .ini file changes) and my GPU is at 80-85%. So I think I can definitely say that in this test scenario my CPU is bottlenecking the GPU.
- 716
Well I headed over to the AC and Steam Forums, got some input and came up with some assetto_corsa.ini file edits that have added 4-5 FPS and re-allocated my Sandy Bridge CPU core/thread loads. Here's the changes I made to the assetto_corsa.ini file:
[THREADING]
USE_TIMER_PROCESS=0
SET_THREAD_AFFINITY_MASK=1
[PHYSICS_THREADING]
THREADS=3
MSI Afterburner (Road America with 48 AI cars) now shows Core #1 of my CPU is under 100% load (none of my cores were above 55% load prior to these .ini file changes) and my GPU is at 80-85%. So I think I can definitely say that in this test scenario my CPU is bottlenecking the GPU.
Yes, but it is not the cpu that is bottlenecking the gpu per say but that the cpu is not up to the task with 48cars on track. This happens with my skylake too. I have over 60fps and the game sttuters anyway. The computation that is needed for a smooth gameplay with all those cars is not there.
My fiend tested again RA @48cars without his VR headset and got smooth gameplay and got about 120fps, That is with a x79 sandybrinde 8core cpu at 3.7ghz max boost with a gtx1070.
If you think the gpu is heavily bottlenecked by the cpu then run the game with 1 monitor at the same graphical settings as now, note not on RA with 48cars. If it is bottlenecked then you will see low fps and high cpu usage.
With a gtx1070 and my skylake @1920x1200 and every setting set to ultra I got about 140fps, but it was at nurburing and only with 32 cars online.
My friend with a 8c sandy, gtx 1070 and with a 2560x1080 21:9 monitor got 120fps at RA and 48 cars on track.
So the game seems to scale with cores but maybe it is like Lewis_Hamilton_ said that it does not utilize HT(logical cores) properly only real physical cores.
So before you throw a bunch of money on hardware because of one single game and with one single track/condition, I advice you test your system with 1 monitor to see if it really bottlenecks the gpu to much. You will not get the same fps as a skylake but in higher resolution a cpu that can output 120fps compared to 140 at low resolution will not be that bothered when the gpu cant maintain fps close to cpu limit like the 120fps or even dream about 150 fps. The difference in performance between the two cpus will be close to none @ high resolution.
Last edited:
Xiamoto
Premium
- 673
- United States
Thank you. 👍 I'll run some tests with some of the scenarios you mention here for sure. I've already made the decision to build a gaming computer and switch over this i7-2600k/z68 to power an nVidia Quadro graphics card for my CAD work. And I am considering the i7-7700k/Z170/DDR4-3200Mhz/GTX1070 combo for gaming......
Cheers!
- 716
I would say wait to upgrade until AMD Ryzen is out at least, what if it means affordable 8core cpus. I would not get a Kabylake compared to sandybridge, is just faster mhz and ipc, ie more of the same so to say. I did not think my 6700@4.8 gave me better performance than stock 6700 in AC and bf1.Thank you. 👍 I'll run some tests with some of the scenarios you mention here for sure. I've already made the decision to build a gaming computer and switch over this i7-2600k/z68 to power an nVidia Quadro graphics card for my CAD work. And I am considering the i7-7700k/Z170/DDR4-3200Mhz/GTX1070 combo for gaming......
Xiamoto
Premium
- 673
- United States
I saw what appears to be legit graphics and physics benchmarks showing 35% FPS gains for sandybridge@2133Mhz vs. kabylake@3000Mhz. 35% is a massive gain and that's with only 3000Mhz....
. So I suppose it really comes down risk tolerance and whether Ryzen makes our Intel CPU's plummet in value overnight .....
- 716
I saw what appears to be legit graphics and physics benchmarks showing 35% FPS gains for sandybridge@2133Mhz vs. kabylake@3000Mhz. 35% is a massive gain and that's with only 3000Mhz....
Only at low res when the cpu is the bottleneck and the memory which is the weakest link in that case, will show any difference. When gpu is the bottleneck it will do nothing.
Last edited:
Whitestar
Premium
- 2,103
- Norway
- Whitestar127
True, so if you run games at 1080p (which is sort of low res nowadays) which many of us still do, then you will actually see those benefits in some games when using higher CPU and DDR4 frequencies and a powerful GPU, such as in this video: .
In that video they (Digital Foundry) are running the games with maxed out Titan X Pascal settings at 1080p.
@kapnk006: Notice the bold part above. It varies wildly from game to game, so it might not be worth it. And since Pillo-san hints that AC is not such a game then probably not worth it if you mostly play that. But in any case I agree with him: Wait and see what Ryzen brings to the table.
Whitestar
Premium
- 2,103
- Norway
- Whitestar127
Yes, well then you've also seen that it's not that simple unfortunately.
As you can see in the video yourself this varies from game to game due to the engine. And from Pillo-san's findings I'm assuming you won't find this "benefit" (i.e. that you'll get better performance if you up the memory freq) in ACs engine.
