Hi, I aquired this beast a while back, but didnt really have any time to play with this. This is a monster when it comes to TRAMs. A big size-8 packed with a Motorola DSP96002 with 4Mb RAM, T800 with 4Mb RAM, 4Mb shared RAM between the DSP96002 and T800, 2x1Mb VRAM for video out. Does anyone have any documentation/software for this? I would really like to get this thing running.... Thanks, Ram [Comp.parallel moderator: email the user.]
NEL 96KD DSP Transputer TRAM
Started by ●February 19, 2004
Reply by ●February 20, 20042004-02-20
Ram Meenakshisundaram wrote:> I aquired this beast a while back, but didnt really have any time to > play with this. This is a monster when it comes to TRAMs. A big > size-8 packed with a Motorola DSP96002 with 4Mb RAM, T800 with 4Mb > RAM, 4Mb shared RAM between the DSP96002 and T800, 2x1Mb VRAM for > video out. Does anyone have any documentation/software for this? > I would really like to get this thing running....I had some involvement with the transputer many years ago. If all you have is the TRAM itself, you will need to find a B008 or equivalent interface card to interface it to your PC. You will then need the TDS (Transputer Development System) software to write some test code so that you can check that the unit is still functional. You should be able to get the transputer working, but accessing the DSP might prove difficult without any documentation. Personally, I wouldn't bother; the T800 was a very nice chip in its time but is very slow by today's standards. Leon -- Leon Heller, G1HSM Email: aqzf13@dsl.pipex.com My low-cost Philips LPC210x ARM development system: http://www.geocities.com/leon_heller/lpc2104.html
Reply by ●February 24, 20042004-02-24
Leon Heller wrote:> I had some involvement with the transputer many years ago. > > If all you have is the TRAM itself, you will need to find a B008 or > equivalent interface card to interface it to your PC. You will then need > the TDS (Transputer Development System) software to write some test code > so that you can check that the unit is still functional. You should be > able to get the transputer working, but accessing the DSP might prove > difficult without any documentation. Personally, I wouldn't bother; the > T800 was a very nice chip in its time but is very slow by today's standards.But a truly brilliant architecture for distributed control systems from it's elegant implementation and externalization of the CSP model to it's "huffman encoded" instruction set. It was once my job to study and understand its every nuance. That took a while but was one of the more interesting tasks I was ever given. Sad that it didn't remain in development, IMHO. I was asked to design extensions to C to naturally encompass the CSP process constructs it implemented (with blazing speed) and felt the result to be a masterpiece. :-) I think this architecture should be resurrected and fleshed out with vector DSP operations (if that didn't happen after I had to move on from it.) Occam on the other hand... Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein
Reply by ●February 24, 20042004-02-24
Leon Heller <aqzf13@dsl.pipex.com> wrote in message news:<4035ce8f$0$21302$cc9e4d1f@news.dial.pipex.com>...> I had some involvement with the transputer many years ago. > > If all you have is the TRAM itself, you will need to find a B008 or > equivalent interface card to interface it to your PC. You will then need > the TDS (Transputer Development System) software to write some test code > so that you can check that the unit is still functional. You should be > able to get the transputer working, but accessing the DSP might prove > difficult without any documentation. Personally, I wouldn't bother; the > T800 was a very nice chip in its time but is very slow by today's standards.Actually, I have all that and I can load programs on the transputer, but I want to access the DSP and the video out which seems impossible without the documentation. Yes, it is slow, but the technology is solid even for todays standards. There are a lot more processors that are slower than the T8 that are still in use today. Even the 386 is still used for embedded processings and the T8 is much faster than that! Cheers, Ram PS: I maintain the transputer homepage at http://www.classiccmp.org/transputer :-)
Reply by ●February 24, 20042004-02-24
In article <4035ce8f$0$21302$cc9e4d1f@news.dial.pipex.com> on Fri, 20 Feb 2004, Leon Heller <aqzf13@dsl.pipex.com> writes>Ram Meenakshisundaram wrote: >> I would really like to get this thing running.... >I had some involvement with the transputer many years ago.As did many of us who still frequent comp.sys.transputer.>If all you have is the TRAM itself, you will need to find a B008 or >equivalent interface card to interface it to your PC.I guess Leon probably saw Rams post in comp.dsp, though it was cross posted to comp.sys.transputer and comp.parallel as well. Since it may not be immediately apparent, Ram is not in any way a Transputer newbie - he runs one of the most comprehensive Transputer websites out there and has done a lot of work with a wide variety of Transputer based systems. More often than not he is answering questions on c.s.t rather than asking them. http://www.classiccmp.org/transputer/ Ram Meenakshisundaram's Transputer Home Page Ram, sorry I can't help with this board, but I notice that you have copies of the famous Inmos flight simulator on your website now. I'm very tempted to see if I have all the hardware needed to get it up and running at home. My biggest stumbling block is probably the graphics card. Take care, Mark.......... -- Mark Booth
Reply by ●February 24, 20042004-02-24
rmeenaks@olf.com (Ram Meenakshisundaram) wrote in message news:<1e2af2.0402190604.184a9009@posting.google.com>...> I aquired this beast a while back, but didnt really have any time to > play with this. This is a monster when it comes to TRAMs. A big > size-8 packed with a Motorola DSP96002 with 4Mb RAM, T800 with 4Mb > RAM, 4Mb shared RAM between the DSP96002 and T800, 2x1Mb VRAM for > video out. Does anyone have any documentation/software for this? > I would really like to get this thing running....Part of me says that trying to get this old stuff running is a total waste of effort. There's also the part of me that thought it was really neat when I was at college and they got one of Lord Thompson's original prototype triodes working with modern passive components. They made a TRF radio, picking up medium wave AM channels. I can't remember the exact occasion that prompted this to be done. It would be been about 70 years after the triode was made, but that doesn't seem like a very significant number. It think it gives some justification for saying I studied at the birthplace of the electronics industry, though (UCL) :-) Regards, Steve ------------ And now a word from our sponsor --------------------- For a secure high performance FTP using SSL/TLS encryption upgrade to SurgeFTP ---- See http://netwinsite.com/sponsor/sponsor_surgeftp.htm ----
Reply by ●February 25, 20042004-02-25
[C.p. moderator: diverging a bit from ||ism and transputers, so followup to DSP. You guys might want to start editing the Subject line.] Steve Underwood wrote:> rmeenaks@olf.com (Ram Meenakshisundaram) wrote in message news:<1e2af2.0402190604.184a9009@posting.google.com>... > >>I aquired this beast a while back, but didnt really have any time to >>play with this. This is a monster when it comes to TRAMs. A big >>size-8 packed with a Motorola DSP96002 with 4Mb RAM, T800 with 4Mb >>RAM, 4Mb shared RAM between the DSP96002 and T800, 2x1Mb VRAM for >>video out. Does anyone have any documentation/software for this? >>I would really like to get this thing running.... > > > Part of me says that trying to get this old stuff running is a total > waste of effort. There's also the part of me that thought it was > really neat when I was at college and they got one of Lord Thompson's > original prototype triodes working with modern passive components. > They made a TRF radio, picking up medium wave AM channels. I can't > remember the exact occasion that prompted this to be done. It would be > been about 70 years after the triode was made, but that doesn't seem > like a very significant number. It think it gives some justification > for saying I studied at the birthplace of the electronics industry, > though (UCL) :-)Somewhere in my attic I have a U.S. Army Signal Corps VT-1 that my father bought war surplus in 1919. He used it as a diode to replace the crystal in his radio. Later I used it as a regenerative receiver driving headphones, and later still I added a reflex circuit to drive a loudspeaker. It was originally out of a two-man portable field radio. The receiver used four of the tubes, two TRFs and push-pull audio which also was the modulator, and the transmitter used another two: an oscillator and a class-C final. I really ought to get it to a museum, along with a few others nearly as old. Years ago, I threw out all the antiques with no emission. Another old gizmo up there is a General Radio 1 KHz standard on an oaken box that consists of a tuning fork, carbon microphone, and drive coil. Last time I tried it, there was enough gain to sustain oscillation. Jerry -- Engineering is the art of making what you want from things you can get. ����������������������������������������������������������������������� ------------ And now a word from our sponsor ---------------------- For a quality mail server, try SurgeMail, easy to install, fast, efficient and reliable. Run a million users on a standard PC running NT or Unix without running out of power, use the best! ---- See http://netwinsite.com/sponsor/sponsor_surgemail.htm ----
Reply by ●February 25, 20042004-02-25
Mark Booth <markbFrom@anang.com> wrote in message> Since it may not be immediately apparent, Ram is not in any way a > Transputer newbie - he runs one of the most comprehensive Transputer > websites out there and has done a lot of work with a wide variety of > Transputer based systems. More often than not he is answering questions > on c.s.t rather than asking them.Hi Mark, Thanks for the complement. After dealing with transputers for about 14-years, and still going, I just cant let this beautiful architecture just fade away.> Ram, sorry I can't help with this board, but I notice that you have > copies of the famous Inmos flight simulator on your website now. I'm > very tempted to see if I have all the hardware needed to get it up and > running at home. My biggest stumbling block is probably the graphics > card.My transputer system is currently offline at the moment. I am in the process of writing drivers for XP. I finished the standard B4/B8 stuff, but I am working on the TMB17 PCI stuff and its just taking a bit longer. Once I am done, I am back to business. I am planning on re-writing the flight simulator a bit using a more tram-based approach. Here is my current hardware lineup: 1) A graphics TRAM 2) A serial TRAM for hooking up a serial joystick (using a Gravis Stinger Laptop gamepad) 3) A bunch of compute trams I'll post the B4/B8 drivers this weekend (standard Windows DLL with INMOS style OpenLink, CloseLink, etc) and probably finish the TMB17 in another week or so. Once I am done, I'll work on the flight sim. I'll probably do the transputer only version 1st, and maybe add another version using the PC's video card instead of a graphics TRAM... Cheers, Ram
Reply by ●February 25, 20042004-02-25
"Bob Cain" <arcane@arcanemethods.com> wrote in message news:c15vv3019v7@enews3.newsguy.com...> Leon Heller wrote: > > > I had some involvement with the transputer many years ago. > > > > If all you have is the TRAM itself, you will need to find a B008 or > > equivalent interface card to interface it to your PC. You will then need > > the TDS (Transputer Development System) software to write some test code > > so that you can check that the unit is still functional. You should be > > able to get the transputer working, but accessing the DSP might prove > > difficult without any documentation. Personally, I wouldn't bother; the > > T800 was a very nice chip in its time but is very slow by today's standards. > > But a truly brilliant architecture for distributed control > systems from it's elegant implementation and externalization > of the CSP model to it's "huffman encoded" instruction set.The instruction set seemed to me to be truly inventive: I'm not aware of anything similar that preceded it. Despite Inmos' desire to keep people away from the assembler level so they could revise the code it was a delight to program (though I only ever wrote some experimental code).> It was once my job to study and understand its every > nuance. That took a while but was one of the more > interesting tasks I was ever given. Sad that it didn't > remain in development, IMHO. I was asked to design > extensions to C to naturally encompass the CSP process > constructs it implemented (with blazing speed) and felt the > result to be a masterpiece. :-)I'm interested in how these extensions to C looked in the C source. Were they coded as normal C library calls or did they have a different syntax from ANSI C?> I think this architecture should be resurrected and fleshed > out with vector DSP operations (if that didn't happen after > I had to move on from it.)One more thing on the wish lists of the world! I don't think there would be the commercial drive to reimplement. What's more I can't see how a new system could compete, unless it could run existing app code, and there are probably faster alternatives. ------------ And now a word from our sponsor ---------------------- For a quality mail server, try SurgeMail, easy to install, fast, efficient and reliable. Run a million users on a standard PC running NT or Unix without running out of power, use the best! ---- See http://netwinsite.com/sponsor/sponsor_surgemail.htm ----
Reply by ●February 26, 20042004-02-26
Bob Cain <arcane@arcanemethods.com> wrote in message news:<c15vv3019v7@enews3.newsguy.com>...> Leon Heller wrote: > > > But a truly brilliant architecture for distributed control > systems from it's elegant implementation and externalization > of the CSP model to it's "huffman encoded" instruction set. > It was once my job to study and understand its every > nuance. That took a while but was one of the more > interesting tasks I was ever given. Sad that it didn't > remain in development, IMHO. I was asked to design > extensions to C to naturally encompass the CSP process > constructs it implemented (with blazing speed) and felt the > result to be a masterpiece. :-) > > I think this architecture should be resurrected and fleshed > out with vector DSP operations (if that didn't happen after > I had to move on from it.) > > Occam on the other hand...Don't worry Bob An FPGA Transputer 32b well may run just over 300MHz for about $2 on the upcoming spartan3 family. Prices are for 250K units so scale that for more reasonable vols same way older spartan2. Surprisingly a 64b version may be just as straight forward using 2x the slices and rams and same speed. This speed is only possible because of multithreading 4 or 8 way resp. HT is a natural for Transputing so thats part of the model. Like the T9000, workspaces are cached, and some ops like branches can be multiply grouped with non branches so throughput maybe 400 or more mips for 1 in 4 branches. Branches actually help rather than hinder. FP though will have to be in SW but some FP codes may well be added too. Oh yes Occam is still in the prefered language or rather par, alt, seq, ?! chans etc but they are embedded in a mixed Cxx Verilog language and compiler based on lcc. Thats where most of thre real work is now, the HW may be ready long before the compiler is. There will be more news from time to time sometimes here or in FPGA group. johnjakson_usa_com
