DSS/96 5/6/96, Bill R.
Present DSP systems available are PC32 and PC44. Both are 32 -bit floating-point systems.
The PC32 is an economical board that has already been programmed by Mike Rosing to produce timing pulses, perform rate generation and generate a sinusoid. It can behave as if it were 8 timers with 1ms timebase and 32-bit length. It also can react to logic input signals on 8 lines.
A UET equivalent is also implemented.
At the same time several rate generators can by synthesized for controlling sample rates for A/Ds.
This is DSP system that has more onboard capability than the PC32 but also costs more (basic board is $2000). It has on board two A/Ds and D/As that are relatively high performance. That is, they are better than the standard audiophile systems. They are 16-bit and can run at 200kHz. In addition two additional TMS320C40 processors can be added to this board ($1000+ per board) to provide additional capability. We can put up to 16Mword of memory on this system. It presently has 4Mword (actually 2 Mword).
Mike R. Is presently learning the capabilities of this system using a development system that we have.
Another consideration is A/D capability of these systems and how many processes can go on simultaneously. This will be worked out over the next few months. The optimum assignment of function between these two processors will be left for them future.
We tried to obtain PCI based systems since this would have offered higher bandwidth between the PC and/or Alpha and the DSP cards. This didn't work out as we were too early in the game. The present processors connect to the ISA bus. We are also keeping an eye on fast serial busses such as the 1394 which is an IEEE standard that is slated to achieve 400Mbit/sec in the next year. This could be a low cost method of connecting all devices in a lab. However, we don’t know what the tradeoffs might be at this point.
The DSP approach is quite open-ended. There are no inherent restrictions on the type and or number of signals that can be generated given the customary limitations of the available processor speed, memory and bandwidth. What we would like to do is be sure that there aren't some immediate needs that we have ignored.
Some of what we are concerned with is the human interface. All too often these systems are not very user friendly. We have been exploring some graphical interfaces but really don't know what the optimum path is.