KOSH [Kommunity Orientated Software Hardware]

Weekly Summary

Week:  24th May, 1999

Number:  015

Mailing List:  kosh-hardware-o

In the mailing list this week, the following items were discussed.
Please do not email the scribe regarding any of these topics, it is 
not his job to answer these questions, but merely to report the topics
of conversation. If you have any queries about this summary, please 
email summaries@kosh.convergence.org, stating the Summary Number, and
Mailing List Name and he will try to answer your queries.

a)

Subject: Windows 9x loses mulitasking when using floppy drive?

Summary of Debate:   Windows 9x has been observed to behave very
		     slugishly when using the floppy drive.  This
		     may be because the drivers use the PC BIOS,
		     which is 16 bit code and requires windows
		     to turn of multitasking while it's in use.

b)

Subject:  What determines the bit size of a system (16, 32, 64..)

Summary of Debate:  There are several possible ways to measure this.
		    One is to measure the actual size of the
		    registers in the CPU, though this is not
		    always uniform.  Another is to measure the
		    width of the data bus.  The most common is
		    to measure the software programming model
		    used with the system.

c)

Subject:  Loading drivers from the network.

Summary of Debate:  The fact that USB devices can load their drivers
		    dynamically fits well with the Kosh plan of
		    making device drivers available on servers on
		    either an intranet or the internet.  For devices
		    without drivers, requests could be sent with
		    the devices built-in device id and vendor id
		    requesting the appropriate driver.  It was 
		    mentioned that these requests would be a
		    good way to keep track of what devices were
		    popular but didn't have Kosh drivers yet.  Of
		    course this would be all configurable, for folks
		    that don't want Kosh knowing anything about
		    their hardware, etc...

d)

Subject:  Field Programmable Gate Arrays assisting CPU's

Summary of Debate:  It was suggested that it might be interesting
		    to put one or more FPGA's onto the bus, with
		    the intention of loading them with specific
		    routines that could be used to assist the
		    CPU, such as rendering, encryption, decoding.
		    These routines would be designed by a team
		    of engineers, possibly using software to 
		    synthesize a Hardware Description Language
		    design into a FPGA.  
		    Xilinx were mentioned because they employ
		    SRAM rather than flash memory, as flash would
		    wear out as new routines were continually
		    loaded onto the FPGA.  The Altera Flex
		    FPGA's were also mentioned as they are used to
		    develop the ARC and TDMI's ARM CPUs, and can
		    be found in a Nintendo 64 add on.  
		    Major concerns were that these FPGA's would
		    run too slowly to be useful, and that they
		    would be too expensive.  It was mentioned that
		    the FPGA market is very competetive and the
		    FPGA makers are giving away more development
		    modules.