subtexload and friends

New Message Reply Date view Thread view Subject view Author view

Jean-Francois Panisset (panisset++at++cim.mcgill.ca)
Thu, 05 Jan 95 22:33:46 -0500

I'm trying to get the highest (and most predictable) performance from
the texture subsystem on an Onyx with RM4 boards, which is leading me
to us TX_FAST_DEFINE and (fb)subtexload(). I thus have a few questions
for the RE gurus which haunt this list:

- Is *every last byte* of the 4Mb of texture memory usable when
loading non-mipmapped textures, or is there any hidden overhead
(barring texture memory fragmentation: I'm making sure to always load
textures of the same size). For instance, do I have enough
memory for a 1024x1024 RGBAx8bit texture? How about a 1024x512
RGBAx12bit?

- To load a texture into the RMs from an array in main memory, is it
faster to first lrectwrite() it into an ilbuffer and then use
fbsubtexload(), or should I use subtexload() directly?

- Is there any information available concerning the mysterious last
parameter of subtexload()/pfsubtexload()? Recently posted code for
feeding live video from a Sirius directly into texture memory used a
number of apparently Sirius-related defines (SIR_VEN_16BIT_TEXEL and
other such symbols). Barring that, what default value should be used?

Thanks in advance,
Jean-Francois

New Message Reply Date view Thread view Subject view Author view

This archive was generated by hypermail 2.0b2 on Mon Aug 10 1998 - 17:50:50 PDT

This message has been cleansed for anti-spam protection. Replace '++at++' in any mail addresses with the '@' symbol.