Hi,
On Thursday, August 25, 2016 at 2:06:53 PM UTC+1, John wrote:
Does anyone know of a transputer style chip currently in production?
STMicroelectronic's ST20 contains what is essentially a Transputer core (historically, STMicroelectronics inherited/acquired Inmos and the Transputer technology/IP). Whether it is still being manufactured I don't know (datasheet I found dated 2004, but
there's probably a _lot_ of them in set-top boxes and similar devices around the world!). The closest descendant, I guess, is XMOS's XCore. Same CSP ideas underlying it, but somewhat different in mechanics (I don't think the XCore was intended to be
general-purpose in the microprocessor sense, but could be wrong!).
Also could anyone indicate the raw processing power of a single T800
for example in relation to an intel family chip?
Googling around suggests about 10 MIPS for a 20 MHz T800. Hard to compare directly, since the feature set is completely different, but looking at a table on Wikipedia suggests AMD's am386 and Intel's 486DX @25MHz are comparable.
(Tricky I know - I'm just looking for an idea - I understand the difference in
purpose and design)
There are also a lot of claimed parallel FPGA chips arround - how do they compare in terms of processing power to the intel family?
[snip]
Whole subject probably. FPGAs are mainly limited by how fast they can go, but with the plug-and-play soft CPU cores (and hard CPU cores on various), you can probably get decently high performance for specific applications. The parallelism, of course,
is inherent in the FPGA, but likely difficult to exploit well, correctly. (Handel-C was an excellent thing, VHDL I'm less convinced).
There have been some papers in CPA (www.wotug.org) in recent years covering Transputers in FPGAs, which look very promising. Perhaps for the hobbyist more than industry currently, but still attractive.
-- Fred
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)