Scali wrote:I vaguely recall that the AMD InterWave chip used by the PnP has slightly different timing than the GF1 chip. In most cases you will not notice, but there are corner-cases, with very long looping samples, where the InterWave can get out-of-sync a bit.
From my reading of the Programmer's Guide, there are timing differences. I wouldn't expect problems in "enhanced" mode, but its GF1 compatibility mode seems to be more of an approximation, especially beyond 14 channels. They call it frame expansion, and it's essentially the addition of delays to approximate the playback timing of the GF1 while still running at a constant 44.1 kHz.
The interesting question, however, is whether the problems with looping are inherent in the InterWave, or whether they can be worked around by tweaking the layout of the samples in memory.
FT2 seems to use the chip in enhanced mode, being able to use all of the memory I have installed on my card beyond the basic 1 MiB that is available in compatibility mode.
Some software will need to be patched to work on the InterWave, because the timing was somewhat borderline on some systems, even with a real GF1 card. Also, if the programmers took shortcuts when writing their GF1 code, it may not be enough to trigger a switch to compatibility mode, providing incorrect results or crashing outright.
Personally, I quite like the InterWave. The added CODEC functionality adds flexibility, IWSBOS is a little more functional and less temperamental than the older tools, and the additional memory capacity and ability to use customised ROM banks are considerable benefits.
