Messages in this thread | | | Subject | Floating point exceptions: how to mask & set to zero w/ no error indication? | Date | Thu, 11 Apr 1996 14:30:29 -0400 | From | ulmo@q ... |
| |
Excuse my Unix & floating point ignorance please, but I'd like to generate some form of compatability with Speak Free (see ftp://ftp.fourmilab.ch/pub/kelvin/speakfree/unix/prior-releases/p/sf61a2.tar.gz ), the main author John Walker <kelvin@fourmilab.ch> said:
> If the original PARC LPC code (which remains in the program, > as it is a standard VAT and RTP compression mode) is generating > floating point errors, this indicates that your system is not > masking floating point exceptions and handling them by default. > The LPC code routinely generates floating underflows, which the > FPU should set to zero with no error indication.
I take it this can be made a default by the kernel, or the user program can ask for this somehow.
> If you're getting interrupts for this, your coprocessor support is > enabling interrupts most systems mask.
The english got jumbled at the end, but I fear what he means is that it's enabling uninterruptable (or unmaskable) interrupts -- which may or may not kill performance of that driver (I don't know how many floating point errors it generates per second -- I suppose this could be counted). What I think he means is that there's some bit that needs setting :).
How do I do that? If there is an interface that I need to interact with in user programs, then I'll do that in Speak Free (& send patch); else, can the kernel handle this by default?
Thanks a lot for assistance, Bradley Allen <Ulmo@Q.Net>
P.S. That latest alpha of Speak Free is pretty good even at 14400bps modem PPP connections, with the -lpc10 option on sfmike (remember not to run everything with -d (debugging)). I just wish the full duplex drivers for the SoundBlaster AWE32 on Linux/*nix were reality (there's an experimental one for Windows already).
------- Forwarded Message
| |