madoka
3rd August 2005, 22:01
This is a follow-up to this thread (http://forum.doom9.org/showthread.php?t=97536).
What I mean is given the same input and settings, does x264 guarantee the same output every time (assume raw stream output)? Empirical evidence suggests that it is the case with --no-asm, but what about with assembly optimizations?
By the end of the before-mentioned thread, I was convinced that I had faulty hardware. So I ran the standard suite of stress test tools--memtest86, Prime95, SuperPI--but they all gave negative results. In contrast, I couldn't even get x264 to generate matching output for a 20-frame input!
Of course, just because my machine passed the diagnostics doesn't imply that it's not faulty. However, assuming that the answer to my question is the affirmative, then this makes x264 the most sensitive stress test tool I know. Maybe the devs should consider writing another stress tester based on x264. ;)
What I mean is given the same input and settings, does x264 guarantee the same output every time (assume raw stream output)? Empirical evidence suggests that it is the case with --no-asm, but what about with assembly optimizations?
By the end of the before-mentioned thread, I was convinced that I had faulty hardware. So I ran the standard suite of stress test tools--memtest86, Prime95, SuperPI--but they all gave negative results. In contrast, I couldn't even get x264 to generate matching output for a 20-frame input!
Of course, just because my machine passed the diagnostics doesn't imply that it's not faulty. However, assuming that the answer to my question is the affirmative, then this makes x264 the most sensitive stress test tool I know. Maybe the devs should consider writing another stress tester based on x264. ;)