Audio test

[Chalenge]

For the last few days I have been testing audio configurations. I started out testing by shutting off vsync and turning on all the new graphic details. I am using AlacrityPC to turn off all non-essential processes and to maintain a consistent system state. The result tells me that pushing through a denser audio stream improves over all system performance. If I do the same test with all of the new graphic details off I get the result I would expect. A lighter audio stream improves over all system performance in that case.

Would I be correct in concluding that either the Nvidia drivers are causing issue with the latest changes or that the changes themselves might need some tuning?

[kamori]

Is it onboard sound?

Kam

[Chalenge]

No and I guess I wasnt clear. I am running the AHfilm viewer and taking benchmarks with Fraps. Im just trying to narrow down what might be causing results to go one way wihtout the graphic enhancements and another without. Since the sounds remain the same per film and the films are a fixed length and the same events occur per repetition you would think the frame rates would be consistent.

Say I take sample A and run the test without the new enhancements and get 118 fps and then run sample B under the same conditions and with the same film and get 110 fps. Then I would expect sample A to give better results than sample B with the new enhancements as well. Thats not happening.

So I am asking if it might be a technical issue I have missed somehow or Nvidia drivers somehow messing things up or something else altogether?

[Skuzzy]

I still do not get what you are doing, or what the conclusions are.

First off, Fraps is going to impact game performance.  The results are always going to be +/- some percentage due to everything else that could and/or would run in a Windows computer.

[Chalenge]

Fraps has the ability to record frame rates on a second by second basis without recording a movie. What I am doing is changing the custom sounds that AH is using. I have two samples A and B. A is very similar to the default sounds that AH uses and the difference in frame rates from default is minimal when I benchmark with default or sample A. Sample B is twice the size and should slow performance and does when I leave the video settings as they were before the graphics change. Sample B drops out performance under those circumstances but when I turn the graphics additions on sample B runs better than A. It doesnt make sense to me why that is.

[Skuzzy]

You can run into so many caching issues which will impact those results it is hard to describe.

What are the exact details about the saound files?  Sample rate, encoding, bit length...

[Chalenge]

I tried to adjust for caching issues by running through the films three times consecutively with the third being the benchmark run.

Sample A details:

22kHz PCM signed 16 bit 352 kbps mono

Sample B details:

22kHz PCM signed and unsigned 8 and 16 bit 352 and 705 kbps stereo

Do you think the Sample B varieties in signed and unsigned bit-depth and bit-rate could be the issue?

[Skuzzy]

Are those wave files?  Typically there is no "stream rate" with a PCM encoded wave file as they are not compressed, like MP3's are.

[Chalenge]

Yes they are wave files. The stream rate is the reported property Goldwave and Windows Explorer give for it. Nothing like that comes up on Audition.

[Skuzzy]

8 bit is always going to process faster than 16 bit will.

[Chalenge]

I would still expect the results to be consistent. I did test several 8-bit sounds and found that several come across as total silence with the Miles system whereas before they were working fine.

[Skuzzy]

I would still expect the results to be consistent. I did test several 8-bit sounds and found that several come across as total silence with the Miles system whereas before they were working fine.

I can see where it would be inconsistent.  The test configuration is wrought with so many variables it makes it difficult to realize anything consistent.  Happens when you have to go through a dozen layers of software.