[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Occam compiler using gcc
I don't have any plans do do other back-ends for SPoC than the existing
C, occam and an updated PIC. If any of SPoC is useful to you, feel free to
use it or ask for advice. We're particularly proud of the vector usage
checker---much faster than INMOS's.
On Fri, 31 Aug 2001 oyvind.teig@xxxxxxx wrote:
> Ruth asks for comments:
> 1. Your initiative is just great!
> 2. We are still using occam in products, as a matter of fact
> I have even today compiled C freshly generated from occam
> 3. Anything which makes occam more "runnable" is great!
> 4. I did a "search" in you mail and could find no
> mention of "Linux". Good;-) (I might one day soon thank
> UKC that they did the Linux port! But I presently so miss
> KRoC for Windows..)
> 5. Wish list:
> SPoC has direct C insertion into occam:
> INT var, reply:
> chan ? var
> #C $reply = c_func (&$var);
> will call c_func with occam int "var" expanded.
> There's also a "#H" C direct code insertion.
> It will move the inserted code in scope before
> all generated C, to use for #include's. SPoC
> ensures that #C may be placed everywhere but #H
> only at top occam level. (Why haven't the world
> discovered SPoC? What's wrong with my evangelizing
> abilities? (Don't tell me!))
> This would make ALL my windows based code (with
> sockets and all) directly portable to your
> new ??oC.
> I would also like flags being set by other
> windows threads ("asynchronous events") to be
> connected to occam channels. I've also done
> this for SPoC (End of DMA interrupt routine
> in C signals the end-of-dma channel (DSP)).
> On the PC I've connected socket input thread
> to a channel, and another C thread's callback
> is connected to an occam channel.
> "It's gotta be simple. It's gotta be tasty."
> It is!
> But it takes one or two context switches too much..
> 6. I'd be glad to help!
> Not with compiler writing but with ideas, criticisms,
> evaluation etc.
> 7. By the way, isn't SPoC's scheduler really scheduling
> Green Threads?
> \ Øyvind /
> / Oyvind Teig - oyvind.teig@xxxxxxx
> \ Kongsberg Maritime Ship Systems, Ship Control (KMSS-SC)
> / 7005 Trondheim Norway
> \ 47 73581268 - http://www.kongsberg.com/eng/KM/
> / http://home.no.net/oyvteig/ - Home page
Denis A Nicole WWW: http://www.hpcc.ecs.soton.ac.uk/~dan
High Performance Computing Email: dan@xxxxxxxxxxxxxxx
Department of Electronics Phone: +44 23 8059 2703
& Computer Science Fax: +44 23 8059 3903
University of Southampton