|
From: Richard H. <rj...@za...> - 2022-10-31 11:38:16
|
On 28/10/2022 16:18, Thorsten Renk wrote: > >> The next big problem is GC stuttering. It is mainly a problem of Nasal > While that particular claim has come up for a decade, it is actually > pretty hard to solidify - it is quite possible to switch the Nasal GC > off (then memory usage just increases till it hits the limit) - yet > the issues persisted when this test had been done. > > So frankly I think it's unlikely that Nasal is the issue that's > claimed (and by and large I do not have any such issues myself, so... > go figure). > The frame stuttering definitely happens for me, a pause of anything between 20 to 50ms every time the current GC is invoked. I've timed this in the C++ and the values (for the F-15 are in the table below). The main loop has to waits for this so it definitely happens. The time taken will vary with CPU, the measurements below were taken with i7-4790k, 24gb RAM, R9-290 average msec 24.65 max 45.349 min 1.488 The claim isn't hard to solidify - I've measured it using high resolution timers; however I accept that it might be hard to perceive; and doesn't affect Nasal that is coded to avoid GC (either using a lot of globals or being careful about local variables) Fly low and fast in the F-15 and look out of the side window (high zoom level makes it more apparent). |