Video is scrambled for fast playback speeds more than 10x

[hayder78]

I tried the same set of videos on 2 computers. PC1 has almost twice the CPU power of the 2nd. Nevertheless, I cannot see any difference in the quality of playing high playback speeds than the computer with the slower CPU.

Both of them are starting to show a bit of scrambling the video on speeds around 10x and when it reaches 30x, it will be heavily scrambled and the playback will be choppy. I have tried the GPU accelaration, but I found that the CPU is better.

If buying a higher end pc helps, please advise me. Does the high speed playback quality depend on the memory and CPU speed? I have a RAM disk (where the files are), so the hard disk does not impose a limit on that. My pc spec (corei7 3770K @3.5GHz , 32GB 1600MHz RAM , Windows 8.1 64bit , SSD intel 520 series).

CPU usage was around 20-25% when the playback was 10x and the video was a bit scrambled. The scrambling increased with higher speeds and eventually reached with CPU usage that did not exceed ~40% (@64x). Why VLC does not fully utilize the CPU usage?

However, all 8 CPU cores have been utilized by VLC which is good, and apparently means VLC is using the power of multicore processors. So, can I buy a processor with more CPU cores or even a multiporcessor system to make VLC able to play such high speed rates like 64x without scrambling the video and with smooth playback?

I searched a lot in the forum and in google to give me a hint about this issue, but with no luck. Any help is highly appreciated!

[Rémi Denis-Courmont]

Increase single core frequency will improve decoder speed. Increasing the number of cores may improve speed, but only up to a certain point, depending on the codec. If you see that VLC cannot occupy 100% CPU, then you have probably already reached the limit. Increasing RAM will have no effects.

GPU decoding hardware is not meant to work faster than real-time (at least not at the specified maximum supported resolution). Adding GPU cores is useless.

[hayder78]

Thanks a lot Rémi for your informative input!
Your opinion is very logical, and that what I have expected myself.

However, what made me confused is my weird test results that I had performed on 2 different CPU power computers and got exactly the same playback speed results with VLC.

Let me give you a brief glimpse on my test:
VLC ver. 2.1.5
Test video Codec: H264 – MPEG – 4 AVC (part 10) (avc1)
Resolution: 854*480
Frame rate: 29.970029
Decoded format: Planar 4:2:0 YVU



Test video file :
Lecture 04 - Error and Noise.flv (resolution 480p)
From 1:02:00 to 1:05:45 (Replay from a to b)
The video file can be downloaded from:
https://drive.google.com/file/d/0B7pomV ... sp=sharing
(the original video was downloaded from: https://www.youtube.com/watch?v=L_0efNkdGMc )

This is an example of a video frame scrambled when the playback speed was 10.5x (anything more than 10x will cause this and will be more severe with even higher speeds):


VLC -> settings -> codecs -> hardware accelerated decoding : DirectX video acceleration 2.0 : disabled
The test was carried out by increasing the playback speed by 0.1x until reaching the highest speed that was scramble free.

Test on Surface pro3 (intel core i5-4300U CPU @2.5 GHZ (2 physical cores = 4 virtual cores using hyperthreading), 8GB RAM 1600 MHz:
Maximum non-scrambled playback speed : 10x



Test on 40% faster frequency CPU and double number of CPU cores PC (intel core i7-3770K @3.5GHz (4 physical cores = 8 virtual cores using hyperthreading), 32GB RAM 1600MHz):
Maximum non-scrambled playback speed : 10x



Now if the number of CPU cores has a little or no effect on the fast playback decoding which I can understand, why is that 40% increase of CPU frequency has no effect?

Aren’t these tests suggestive that there is some other software limiting factor that made the playback speed capability the same on whatever CPU frequency?

And does this mean that VLC will never be capable of smooth higher speed playback even in the distant future with faster frequency processors, if this software limiting factor will not be solved?

I am asking because there are duties and professions where this will save a lot of time. Double the playback speed and they will spare 6 months of work every year. Skimming several hundreds of hours of monitoring video files is one example. Broadcasting companies perhaps need it too.

Thanks again Rémi for your time and effort.

[Rémi Denis-Courmont]

I don't think anybody will form a satisfactory answer to such a precise performance-related question. You will probably have to figure it out yourself or hire a consultant.

[hayder78]

I have reread my post, and now I am suspicious about the RAM speed. This is the only related hardware specification common between the two systems (1600MHz). Perhaps the speed of the RAM is the bottleneck. I will try to find out how to measure the RAM bandwidth utilization and measure them on both systems.