View Single Post
  #11  
Old August 23rd, 2001, 10:38 AM
admin admin is offline
Administrator
 
Join Date: Oct 2000
Location: Raleigh, NC, USA
Posts: 10,515
Memory suggestion...

64MB may be too small. 128MB would be sufficient. We still don't think it will solve the problem. The K6/K6-2 is just not capable. As to your computer friend, business SW does not use Floating-Point math. He should know this. The rounding errors would kill any accounting accuracy. The floating point calculations we are doing in Vogone are designed to max out a 350MHz processor with (it turns out) a 4-pipeline floating point processor.

As to hard drive space, if the input file is on the hard drive, there is clearly sufficient space. Since Vogone does not use hard drive space to compute its sound live, the HD makes no difference for the computations. When you write the processed file back to disk, you either have the space or you don't. I.E. the hard drive does not affect the program operation at all.

It's actually quite easy to understand. The number of "taps" in the filters determine the quality, rolloff sharpness, passband ripple, etc. are limited by the processing power available. I gave our programmer instructions to use all the available power (i.e. machine microcycles) on a 350MHz. That means probably no other SW will run at the same time with Vogone 2 on a 350Mhz.

If we increase the number of filter taps (i.e. require an 800MHz 4-pipeline processor or higher), we might get better sound/vocal reduction. However, that will restrict who can run Vogone 2+, and still won't reduce the reverb-echoes that everyone is wanting to go away.
Reply With Quote