AC: Stuttering reduction

[Rick Malm]

What can a user do to reduce or eliminate stuttering even in practice runs where their are no other cars on the track?

With Lotus F1 on Monza track I sometimes get as many as 6 stutters per lap, sometimes its 0 which is great.

What is the key cause of stuttering when running AC local vs MP which involves network response time as well as AC MP implementation?

Background:

My system: CPU: Intel i7-4770K Haswell 3.5GHz (some small amount overclocking was installed from Main Performance)
Video: NVIDIA | EVGA GeForce GTX 780Ti to 3 ASUS VE278Q 27” monitors 85hz refresh

I use a SSD memory for the AC install (The SSD is my C: drive)
My cpu usage is about 26% when AC is running. Processor Affinity shows all checked (0 to 8), Priority=Normal,
window resource monitor shows the cpus are getting used and none are any where near 100%, more like 40-50% and there are graphics showing PARKED cpus.

I turned off the car interior controllable items: no wheel, no driver arms.
I set Frame Limit to match my monitor max at 85.
FRAPS shows FPS will dip to 67 in some corners.
I tried reducing smoke generation and mirror resolution, I turned blur off completely.
I have not turn off HDR yet.
I have not tried to us msconfig and kill all background processes. I typically have firefox running in the background
which is the larger memory user (then Spyhunter4, explorer, LCore, office.bin, Steam....)
I have not tried to ALT-TAB out of AC when I see stuttering to see what is running with the resource monitor.
After I ran the resource monitor and looked at the CPU tab and saw a number of PARKED cpus I downloaded and ran a disable parking utility (edits the registry). This seems to have helped. Any problem with doing this?

http://coderbag.com/Programming-C/CPU-core-parking-manager

http://www.coderbag.com/Programming-C/Disable-CPU-Core-Parking-Utility

AC settings: Fullscreen Rendering ON | Triple Screen ON (6065x1080) | Vsync - OFF | Frame Limit - 85 | Aniso x 8 | AA x 4 | FXAA x 3 | Shadows -Low | World Detail - Maximum | HDR Color Sat On @ 92%| Smoke Generation – Low||Smoke in Mirrors OFF| Motion Blur - OFF | Mirror Resolution High| CubeMaps at Low | Faces per Frame - 4

 

[Blkout]

Hey Rick, I'm going to need your results at 1920x1080. I don't have triples so that's the one common resolution we share. I have a 30" 2560x1600 monitor but you won't be able to run that resolution so, it has to be 1920x1080. Try running the benchmark at max settings at 1920x1080, full screen.

Also, I noticed you said you ran it in windowed mode, not full screen, that should disable any antialising.

 

[Rick Malm]

Ok. Thought windowed might be harder.
FPS:
49.9
Score:
1257
Min FPS:
24.1
Max FPS:
60.6
System
Platform:
Windows 7 (build 7601, Service Pack 1) 64bit
CPU model:
Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz (3500MHz) x4
GPU model:
NVIDIA GeForce GTX 780 Ti 9.18.13.3788 (3072MB) x1
Settings
Render:
Direct3D11
Mode:
1920x1064 8xAA windowed
Preset
Custom
Quality
Ultra
Tessellation:
Extreme

Full screen was smoother but I did see a few large stutters, lots of little jerk but not as bad as windowed.
FPS:
57.5
Score:
1450
Min FPS:
24.3
Max FPS:
121.1







Settings
Render:
Direct3D11
Mode:
1920x1080 8xAA fullscreen
Preset
Custom
Quality
Ultra
Tessellation:
Extreme

 

[Blkout]

Rick, your scores look fine which makes me think your CPU and GPU are fine. Starting to think you might have a hard drive issue.

These are mine using a single overclocked 780, CPU at 4.4GHz.

Unigine Heaven Benchmark 4.0
FPS:
58.4
Score:
1470
Min FPS:
29.9
Max FPS:
121.4

System
Platform:
Windows 7 (build 7601, Service Pack 1) 64bit
CPU model:
Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz (3292MHz) x4
GPU model:
NVIDIA GeForce GTX 780 9.18.13.3788 (3072MB) x2

Settings
Render:
Direct3D11
Mode:
1920x1080 8xAA fullscreen
Preset
Custom
Quality
Ultra
Tessellation:
Extreme

 

[Blkout]

Same benchmark with two 780's in SLI.

Unigine Heaven Benchmark 4.0
FPS:
111.1
Score:
2798
Min FPS:
31.9
Max FPS:
229.1

System
Platform:
Windows 7 (build 7601, Service Pack 1) 64bit
CPU model:
Intel(R) Core(TM) i5-2500K CPU @ 3.30GHz (3292MHz) x4
GPU model:
NVIDIA GeForce GTX 780 9.18.13.3788 (3072MB) x2

Settings
Render:
Direct3D11
Mode:
1920x1080 8xAA fullscreen
Preset
Custom
Quality
Ultra
Tessellation:
Extreme

 

[Rick Malm]

Thanks so much for this check of Heaven of your computer as well as all the time you have spent thinking about this subject. I owe you one....

As for Disk performance.

Remember we did see some improvement moving to D: but not a total fix and the stutters were longer in duration. (almost too long to be tolerated). To me it does not make any sense to be a little better with D: that should be ALOT slower.
It might have been a fluke since so many changes had unexpected results and then were not repeatable which I think is due to the random nature of AI races.

Did you see a few jerks in Heaven or not bother to watch it as it did the benchmark?

The SSD is a dual dual for Raid0 that should improve performance, that was the selling feature. It sure help boot time alot and program load time.

I downloaded this benchmark for disk performance.

http://www.techspot.com/downloads/6014-as-ssd-benchmark.html

Test the sequential or random read/write performance without using the cache. AS SSD Benchmark reads/writes a 1 GByte file as well as randomly chosen 4K blocks. Additionally, it performs the tests using 1 or 64 threads and it determines the SSD's access time.

Two extra benchmark tests examine the drive's behaviour when (1) copying a few big files, a lot of small files and a mixture of file sizes by using cached copy functions of your OS as well as (2) reading/writing data depending on the data's compressibility.

I ran it on C: (fast) and D: (which took hours, like 30min for just the 4k read test)
I clearly dont know what the expected SSD time should be but I can compare to my standard D: drive.

Sequential C: about is 5x faster on reads, and 3x on writes
4k test C: is 36x faster and, almost 94x on writes
4k-64 Thrd C: is 200x faster on reads, and 284x on writes
Acc time C: is 152x faster on reads, and 76x on writes
Score C: SSD = 1240 vs 45 for D

SSD looks ok to me.

So wonder if I can stop replay saving completely. I will try it.

 

[Blkout]

Rick, I don't think the issues are with the drives showing a speed issue. It might be that one is flaky and just causing sporadic behavior. I'm not even sure its the SSD's that's causing it but you've practically eliminated everything CPU or GPU related. What I would like to test but this may be more than you're willing to do, is to install windows to a different drive, leave the SSD's completely out and do a normal installation without RAID. Or for that matter, I wouldn't mind seeing Windows just installed to one SSD either without RAID. Either way, I would be curious if the RAID controller could be causing the issue. I'm trying to find commonality between my setup and yours which isn't radically different honestly other than you're using a RAID setup and I'm not. Win 7 is installed on an SSD for me too, a Samsung 840 128GB, but I only use one. My games are actually installed on a secondary 2TB Seagate Hybrid drive. That's why I wanted you to check your game performance on both drives. I'm scratching my head at this point and while RAID shouldn't be a problem, there's just not much else to try to eliminate. Solving PC issues is always about careful methodology, making one change at a time until you find the culprit. You've been very patient and documenting your changes so that helps.

 

[Rick Malm]

You have in interesting idea re Raid0 and SSD.
If I had a 3rd SSD drive I would install it (lets call it W and install windows on it and then the game could be put on my D: or on W:

I really dont want to change the C: SSD right now.
I am working with Main Performance on another issue (HANG/CRASH) that might take some time and they might want the PC back to run their confidential diagnostics.
I tracking and waiting for a fail and minidump for them. The goal of the rebuild of w7
is to see if the issue is HW or SW and it is an intermittent. Might have been a sw corruption from a collision with RegHunter Malwarebytes and Norton.

I do have another option 2 at some point in the future if Main Performance and I come up dry. I have an older HP computer as my backup.
It has 2 drives in it. One is a clone of the HP C: drive. Call it HP D:
I could get it installed in my gaming computer (GC W as and put windows on it
and just count on my Acronis backups instead of the Acronis clone HP D:
Would would be apples and oranges since its not a SSD but would remove the Raid0 issue.

Option 3 I could buy a 3rd small SSD and put windows on it sort of like my HP D: clone.
Both 2 and 3 I would have to learn where to plug them into the motherboard
and powersupply which should not be and issue but its new for me since I am a
PC buyer not a builder. I am a scientist and EE so I could learn what to do.

Other background: to do a rebuild its not just OS and 3 games to install.
My system supports my elite golf analysis work (video analysis) and support
of my home theater (Crestron, and a bunch of electronic devices), this all is 30-40 programs to reinstall and that takes me a few days to a week to do. I am about 1/2 done with it after 3 days of work.

So I need to put this idea of your on the table for awhile and let the other HANG/CRASH issue plays out.

Bye the way, I tried setting AC's Replay save amount to 0. Did not help.
I erased my documents/assetto corsa, set all the options up again, no help. 15 stutters 15 AI.

Then with 15 AI cars with AF4 and AA2 all of a sudden 4 stutters.

Then 25 AI and got and clear worse with more AI at 19 stutters. I was also tailgating this last race so the load was higher. But the graphic settings were lower (AF4, AA2) Attached. I also changed from looking with Afterburner at the cpu's % in race 27 to look at Bus usage, Memory usage, and RAM usage. Nothing popped out as related to the stutters.

I think I will go play golf tomorrow and give this a rest.

Thanks again and again for your input.

Rick

 

[Stelios]

Here are my results, sadly i had a popup during the benchmark and it dropped my fps momentarily to single digit numbers, but otherwise it run smoothly with no stutter or hickups.

Unigine Heaven Benchmark 4.0
FPS:
55.1
Score:
1387
Min FPS:
8.4
Max FPS:
119.2
System
Platform:
Windows 7 (build 7601, Service Pack 1) 64bit
CPU model:
Intel(R) Core(TM) i7 CPU 950 @ 3.07GHz (3064MHz) x4
GPU model:
NVIDIA GeForce GTX 780 9.18.13.3788 (3072MB) x1
Settings
Render:
Direct3D11
Mode:
1920x1080 8xAA fullscreen
Preset
Custom
Quality
Ultra
Tessellation:
Extreme


While i was watching this rather uninspiring benchmark, I had an idea that may help you @Rick Malm , if you have an onboard sound card go to your bios and disable it. If you have a pci sound card, disable it in device manager. If you have 2 sounds cards.... well you get the idea, disable ALL sound cards on your pc and try playing AC again.

 

[Blkout]

I'm betting on the RAID controller or SSD's right now.

 

[Rick Malm]

Stelios, thank so much for taking the time to run the benchmark and post the results.
RE sound device suggestion: I just spent the last 2 hours trying the suggestion re disable the sound devices.

1st using Heaven benchmark. from all sound devices on to all off only changed my Heaven score from 750 to 760 this was not full screen but windowed setting in the benchark..
I could not see a difference, lots of stutters. I think this is the best place to test since Stelios looked at my videos
and said "I watched your video file and this looks like a big problem, there is no way i will be stuttering like that." And his computer is like mine but his Heaven score is almost 40% better and x of mine.
1387 vs 1002. My te videos were using triples not just a single screen that Stelios benchmarked like Mike did as well.

I also did a a number or AC races with different sound devices disabled. This is back to the variations of AI position on the track but this is what I found. with AF=4x and AA=2x and 9 cars. I was getting 5 strong stutters in 5 laps.
Will all sounds devices off ac crashes (BUG). So with only headphones on I get 3 stutters and turn all sound devices back on and I got 8 which does indicate and effect but does not account for all stutters given the Heaven benchmark results.

My sound devices list:
ASUS Xonar DGX device for SimVibe
HD Wdbcam C310
Logitech G930 headset
Nvidia High Definition Audio
Nvidia Virtual Audio Device (Wave Extensibile WDM)
Realtech High Definition Audio

Mike, I am considering working on this issue with Main Performance and will pass your idea on to them especially if I have to send it back I could have them add a 3rd SSD card. Can always use it as a C: clone and it would be none RAID.

 

[Stelios]

Just throw a simple HDD in there and install windows 7 and ac only for testing.
If you have one that is...

 

[Rick Malm]

I dont have a spare. I could pull on from my old HP. I will think about it.
I noticed you ran full screen in Heaven so I reran in 1920x1080 so
apples to apples I am 38% lower than you. Me 1002 and You are at 1387 and Mikes 1470.
Unigine Heaven Benchmark 4.0
FPS:
39.8
Score:
1002
Min FPS:
19.2
Max FPS:
60.2
System
Platform:
Windows 7 (build 7601, Service Pack 1) 64bit
CPU model:
Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz (3500MHz) x4
GPU model:
NVIDIA GeForce GTX 780 Ti 9.18.13.3788 (3072MB) x1
Settings
Render:
Direct3D11
Mode:
1920x1080 8xAA fullscreen
Preset
Custom
Quality
Ultra
Tessellation:
Extreme

I reran with
I select SYSTEM at the bottom which fills all 3 screens.
Unigine Heaven Benchmark 4.0
FPS:
19.2
Score:
485
Min FPS:
12.0
Max FPS:
30.1
System
Platform:
Windows 7 (build 7601, Service Pack 1) 64bit
CPU model:
Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz (3500MHz) x4
GPU model:
NVIDIA GeForce GTX 780 Ti 9.18.13.3788 (3072MB) x1
Settings
Render:
Direct3D11
Mode:
5760x1080 8xAA fullscreen
Preset
Custom
Quality
Ultra
Tessellation:
Extreme

 

[Rick Malm]

Here are my results 1920x1080, .

Could you rerun Heaven with the bottom setting to system so I can compare
with mine at 5760x1080 8xAA fullscreen (my Score 485) and let me know if you see any
jerking. I get lots of small jerks, not as large as in windowed and I get the low score. I will video them later.

 

[Blkout]

Mike, I am considering working on this issue with Main Performance and will pass your idea on to them especially if I have to send it back I could have them add a 3rd SSD card. Can always use it as a C: clone and it would be none RAID.

Rick, I'll be honest RAID 0 is pretty pointless though with an SSD for your OS drive. SSD's are so fast already that I just can't see needing RAID 0 to speed up storage performance. While I know it will help performance if you benchmarked the drives, I doubt the real world difference is all that noticeable since SSD's are so fast to begin with. In my opinion, RAID 0 was more of a performance option before SSD's came along because mechanical hard drives were so much slower. Now I could understand using RAID 1 for parity, but I've never been much on RAID and I regularly back up my drives.

I'm just at the point where I feel like you've eliminated the CPU and GPU issues, and I don't believe the system memory would have anything to do with this issue, the only thing left to troubleshoot is the storage. What I'm thinking but certainly can't guarantee is that the RAID controller is possibly eating up CPU cycles trying and causing some jerking. I have seen a flaky hard drive can causing hitching and jerking too, but I don't know the reasons why. I had an SSD that was defective about 2 years ago, and it worked fine for about 3 months, then I started getting random hangs, Windows would just lock up, sometimes it would last for 10-20 seconds and sometimes it would feeze completely requiring a reboot and sometimes it would just cause my PC to randomly reboot about once per week. I was really pulling my hair out to find out what was causing it, never occured to me that it was my C drive SSD causing it, but after eliminating everything else, I finally installed WIndows to a different hard drive and the problem went away. So I ordered another SSD, installed Windows and its been perfect ever since. That SSD clearly had a problem, maybe a bad block of memory or something that it was trying to write to causing system instability. To me, it really sounds like you have to same issue. That SSD performed just fine in benchmarks, but still exhibited a lot of the problems similar to what you're reporting.

Which SSD's did you say you have? Brand and model?

 

[Rick Malm]

• 2 Solid State Drives Raid0 config:Intel 520 Series 240 GB each
  I contacted Main Performancelast night and they think the motherboard has a problem and want me to sent it back for repair. They think the motherboard more likely the performance problem than RAID. I think I will have them remove RAID anyway. Seems risky vs small performance improvement.
  

 

[Stelios]

• 2 Solid State Drives Raid0 config:Intel 520 Series 240 GB each
  I contacted Main Performancelast night and they think the motherboard has a problem and want me to sent it back for repair. They think the motherboard more likely the performance problem than RAID. I think I will have them remove RAID anyway. Seems risky vs small performance improvement.

Damn! I have the same ssd disk just not in raid. Lose the raid anyways as u said.

 

[Rick Malm]

ok. thanks guys.

 

[Blkout]

• 2 Solid State Drives Raid0 config:Intel 520 Series 240 GB each
  I contacted Main Performancelast night and they think the motherboard has a problem and want me to sent it back for repair. They think the motherboard more likely the performance problem than RAID. I think I will have them remove RAID anyway. Seems risky vs small performance improvement.

Rick, Intel generally builds outstanding SSD drives when it comes to reliability due to their very strict testing, hardware screening and their custom firmware for their drives, however those drives do use a Sandforce controller and those have been known to cause stuttering issues for some people. If you Google Sandforce SSD stuttering, you will probably find quite a few people with those issues. Being completely honest, I can't keep up with all of the available Sandforce SSD drives on the market and if they've finally got those bugs ironed out. What I can say is that Samsung and Crucial SSD's are considered extremely reliable and Samsung is probably the market leader now and they build outstanding SSD's that don't use a Sandforce controller.

http://www.anandtech.com/show/5508/intel-ssd-520-review-cherryville-brings-reliability-to-sandforce

While RAID may or may not be your issue, I am curious if one of your SSD's may be be flaky. The good news is that you have two of them so its unlikely that its affecting both, so if you choose to break the RAID array, you can simply install Windows on a single SSD and try it again, if the problem remains, try the other SSD and see if it still does the same. Of course this is going to be a little time consuming, but something has to be causing this issue and you've already come so far, it would seem almost wasteful to not try this and see what happens.

 

[Rick Malm]

Well Main Perf. said they want it back for diagnostic work without any changes (else could void warentee) so I will ship it in the morning and have asked them to remove the RAID0 config. Will test when it returns, so no SimRacing for awhile. Thanks for all the advice.

 

[Blkout]

Well Main Perf. said they want it back for diagnostic work without any changes (else could void warentee) so I will ship it in the morning and have asked them to remove the RAID0 config. Will test when it returns, so no SimRacing for awhile. Thanks for all the advice.

Well, that's a bummer Rick but I'm sure if you documented the issues and they're repeatable they shouldn't have any problems fixing it and I can certainly understand wanting to keep your warranty. In the end, this had to happen I guess because the performance you were getting simply wasn't acceptable or normal. Hopefully they won't have it too long and you'll be racing again soon.