So frustrated! Panning/turning choppy, craphic spikes, lags, stutters

[SubZer0]

yep, 182.50

will edit the typo

Does the job... got rid of 95% of my stutters, even in the forested areas that I complained about recently

[ReVolut1on]

[quote]




Yes, now that we have been expertly warned, back to the regularly scheduled program, already in progress.....







=====================================


I've got
i7 920@4.2ghz


==========================================


I have to question the stability and the temps of that clock. It is VERY rare to see a i7920 hit 4.2GHz without problems which is one of the reasons why they are much cheaper than the 950/975

Just becuse you dont see crashes and the load temps are showing just within range does not mean you are not overruning that CPU with heat and forcing other problems.. and that includes liquid cooling too.

The symptoms are very similar to a overrun CPU, or, PICe system and of course drivers can also dictate success which is one of the reasons I run the 182.50 driver that has been shown to be the best driver for FSX time and time again...

[NickN]

Alright .. here is what I see

First of all, do not assume load/stress tests are a 100% indication of stability. I dont care how long a test was run there is in fact a plateau you can exceed that will SLOW that processor down without spitting out errors. I also dont care if you use dry ice or LN2 either, there is such a thing as thermal inversion at the silicon level

When you are clocking that high there is far more to stability than benchmark stress tests may show and you can see the result of such issues on screen under the right conditions.


What I will do at this point is ASSUME that CPU is in fact stable in every sense of the word but I am telling you 4.2GHz on air, regardless of its stepping type, is pushing that proc to its absolute limit.

There is a reason Intel sells them cheaper and I dont care what is posted on the net about how they are the same as a 975, which is in every sense of the word,..  manure



First

8-8-8 timing @ 1600 is DOG slow for i7. You may as well be on 5-5-5 1066 memory


Second, RAID is a NO-NO unless you are running that RAID off of a decent PCIe RAID card on a minimum x4/x8 port. You are EATING MORE CPU that the RAID array is giving back in file reads.

Third, .. and this is something I do not know if your BIOS has in it or not, but this setting needs to be DISABLED when clocking like you are:

CPU TM Function: DISABLED

[Ang2dogs]

Yes, now that we have been expertly warned, back to the regularly scheduled program, already in progress.....







=====================================

[jwenham]

Seeing your specs and seeing the performance I get with my system even using 8-8-8 1600 OCZ you really should not be seeing what you are so I too would be inclined to also say that maybe the OC is too much for that CPU. I am running very well with 50/50 AI (UT2), 5% cars with UTX, Very dense scenery and Dense autogen. I can run very dense autogen in alot of areas but hate to stop and notch it down when I hit the pain in the ass pinetree problem. The two areas that smoothed out the stutters for me were Affinity mask of 14 and 100000000 bufferpools. Bufferpools helped the most and the Affinity mask made the stutters few and far between. I dont see any eye opening results playing with the TBM but I have it set to 80. Defragging properly (I use O&O) is also very important for FSX performance. The only other thing I can think of to look at would be any background processes you have running or your AV software in use. Good luck and keep us posted.

[ReVolut1on]

First of all, do not assume load/stress tests are a 100% indication of stability. I dont care how long a test was run there is in fact a plateau you can exceed that will SLOW that processor down without spitting out errors. I also dont care if you use dry ice or LN2 either, there is such a thing as thermal inversion at the silicon level

Of course. There is not such thing as "100% stable". I also agree that there can be other complications. During the stress testing the CPU didn't, supposedly atleast, slow down (LinX measures GFlops calculated during every run). But again, this doesn't actually prove anything. You're obviously right about thermal intervension. I just wanted to point the cooling methods I've tried to run FSX with.

When you are clocking that high there is far more to stability than benchmark stress tests may show and you can see the result of such issues on screen under the right conditions.

I agree.

First

8-8-8 timing @ 1600 is DOG slow for i7. You may as well be on 5-5-5 1066 memory

If I was running on a old LGA775-board, I'd agree. However, isn't that just too, how would I say it, harsh statement? 8-8-8-24 is almost the standard for these systems. It's correct that lower latencies are better than higher frequencies. Still, 5-5-5 1066mhz? 1066mhz itself limits overclocking alot. I'd probably run out of DRAM ratio settings at an OC this high. Personally I don't believe that my RAM, while not being the highest end, are the root cause of my stutters.

Second, RAID is a NO-NO unless you are running that RAID off of a decent PCIe RAID card

I've got a RAID card. Forgot to list that in specs.

[quote]CPU TM Function: DISABLED

[NickN]

The timing of that memory does place it in the same perf range as DDR3 1066 5-5-5 and you can believe what you want, but there have been a lot of users who started out on i7 with CAS8 and 9 memory and discovered when they replaced it the visual result smoothed right out

If you were to place 3x2 DDR3 1066 5-5-5 memory in that system you would experience the same result in FSX as 1600 CAS 8

I do agree that issues around memory timing with respect to FSX can be tuned out so I am not telling you the memory alone is the source of the problem however it can very well compound the problem especially of you are assuming a performance level and pushing settings too high for the scenery, or, something in the system has created a situation causing stutters where the primary source of your stutter problem is being multiplied by the higher memory timing.


Are you using a 256K STRIPE or BLOCK on that RAID array?

I assume so, but if not, that card is not doing you as much good as you may think but will still perform better than motherboard RAID assuming the card is in fact not a cheap 150 dollar RAID solution.

I do not think it is wise to include the OS and FSX on one drive or array if RAID is used.


I assume you have installed FSX to a folder OUTSIDE of the Program Files folder in Vista? If not, that can also be a source of issues as well.


As mentioned, assuming the CPU is stable then if all other areas mentioned check then I would pull the video cards and test one at a time in the primary slot on driver 182.50


....and one other item

If you are using a PCIe RAID card what slot is that in?

Because if memory serves on that chipset/board revision, if it is in slot 4 then PCIe 3 becomes x8 instead of x16

[Tech Diver]

I have been following this thread closely, as I have been struggling with trying to elimiate micro-stutters for the week that I have owned FSX. I have an i7 950 and changing the AffinityMask from 15 to 14 did the trick! Thanks.

[zmark]

Hi Revolution,

I experienced the same issues with spiking, stutters, etc. and after a weeks worth of tweaking, things actually got worse before they got better. I was feeling pretty frustrated over the whole thing, and found myself wishing I had a more powerful system. My persistence finally paid off, I was successful in getting it to run very smoothly AND without much graphics compromise. Fortunately for you, my system is only above average: Pentium D 3.2 GHz, 2 GB of DDR2 memory (PC2-5300), GTX 260 graphics card, SATA hard drive with Vista SP2.

Of course, everyone's situation is different, but in my case, in addition to the above mentioned recommendations for FSX, I achieved the best performance gains by modifying the profile settings within the NVIDIA control panel (version 2.7.170.15 or later).

Try going into the NVIDIA control panel and modifying the 3D graphics settings for the FX10 profile as follows:

Anisotropic Filtering: 16x
Antialiasing mode: Enhance the application setting
Antialiasing setting: 16x
Antialiasing transparency: Supersampling
Texture filtering - Anisotropic sample optimization: ON
Texture Filtering - Negative LOD bias: Clamp
Vertical Sync: Force On

For all other settings, leave them set to: Use Global Setting--the default setting in other words.

As for FSX custom graphics settings, make sure they are set as follows:

Target Frame Rate: Unlimited (go figure!)
Full Screen Resolution: 1920x1080x32
Filtering: Anisotropic
Antialiasing: checked (enabled)
Global texture resolution: Very High
Preview DX10: OFF (unchecked)
Lens Flare: checked
Light Bloom: unchecked

Aircraft settings: Ultra High (everything checked and at maximum)

Scenery - Terrain & Water:
- Level of detail radius: large
- Mesh complexity: 100
- Mesh resolution: 10 m
- Texture resolution: 1 m
- Water effects: Mid 2.x (small compromise, but still it has very cool effects)
- Land detail textures: checked

Scenery Objects:
- Scenery complexity: Very Dense (again, small compromise)
- Autogen Density: dense
- Ground Scenery Shadows: Unchecked (I tried it once, and it resulted in ground texture "blurries")
Special Effects Detail: high

Weather settings:
- Global settings: medium high
- Cloud draw distance: 60mi (minimum)
- Detailed clouds with cloud coverage density set to medium
- Rate of weather change: medium

Traffic:
- Airline traffic density: 15%
- General aviation density: 8%
- Airport vehicle density: minimum
- Road vehicles: 5%
- Ships and ferries: 5%
- Leisure boats: 5%

Finally, I did one tweak to the fsx.cfg file, adding the following lines:
[BufferPools]
PoolSize=5000000

I read a post that said this setting helps with stutters, but I'm not sure of its significance. For one thing, I found out my computer was overheating and the insufficient cooling factor was definitely contributing to stuttering, and spikes, etc. just as the other posts above mention. If you system starts off smooth and then gradually gets worse, you will definitely want to consider this. I added 2 extra cooling fans in the back of the case. It currently has 6 cooling fans and I'm considering adding a 7th one in the front of the case to create greater airflow.

Good luck!

[olderndirt]

Not that it means much, but my card is presently benchmarking 2D/3D graphics -118%/-489% (minus values!) - thus far an unknown problem but it's still running default FSX smoothly at 20 fps.

[Flight Ace]

[quote][quote]




Yes, now that we have been expertly warned, back to the regularly scheduled program, already in progress.....







=====================================


I've got
i7 920@4.2ghz


==========================================


I have to question the stability and the temps of that clock. It is VERY rare to see a i7920 hit 4.2GHz without problems which is one of the reasons why they are much cheaper than the 950/975

Just becuse you dont see crashes and the load temps are showing just within range does not mean you are not overruning that CPU with heat and forcing other problems.. and that includes liquid cooling too.

The symptoms are very similar to a overrun CPU, or, PICe system and of course drivers can also dictate success which is one of the reasons I run the 182.50 driver that has been shown to be the best driver for FSX time and time again...

[Tomaz]

For me, the max frequency for achieving realistic and smooth flight with the i7 920 is 3.78 GHz for CL9 Ram and 3.82 GHz for CL7 RAM. Anything above that for me will cause problems like you were experiencing.

I don't understand that. Higher frequencies are not good and present problems? I don't get this.

[Flight Ace]

For me, the max frequency for achieving realistic and smooth flight with the i7 920 is 3.78 GHz for CL9 Ram and 3.82 GHz for CL7 RAM. Anything above that for me will cause problems like you were experiencing.

I don't understand that. Higher frequencies are not good and present problems? I don't get this.

For  my i7 920 CPU, if I over-clock it over 3.82 GHz, it starts getting unstable. At 4.2 GHz (again for my i7 920 CPU) with a little time I would get the blue screen. And you are right, at stable higher frequencies, a Sim will run better but there is a limit to how high you can push a CPU. Running stable at 3.82 GHZ is better than the 2.67 GHz that the i7 920 is designed for. However if I raise the 3.82  to 4.2 GHZ, then my PC will cease to run and in this case higher frequencies are not good and they do present problems.