12

I am interested to learn about the Texas Instruments TMS34010, a 32-bit fully functional CPU with built-in graphics manipulation capabilities, which appeared in Atari's Hard Drivin' arcade boards amongst others. I was wondering if there were any freely available tools such as compilers to develop software for this processor? So far my searches have been in vain.

  • 1
    I used this chip on one short-lived project years ago. The only tools for it - as far as we could tell at the time - were from TI itself. Someone else may have better knowledge. – Chenmunka Jul 13 '17 at 7:34
  • 1
    Don't know about freely available tools, but the datasheet including opcodes is on archive, and it looks regular enough that writing a backend e.g. for LLVM shouldn't be too difficult, and a fun project. – dirkt Jul 13 '17 at 9:49
  • 1
    @dirkt You'd also need to port a toolchain with an assembler and linker. A C runtime library would also be handy. – Ross Ridge Jul 13 '17 at 15:09
  • I also had one years ago I was playing with. I seem to recall that all the tools were downloadable from TI... – Brian Knoblauch Jul 21 '17 at 16:16
  • 34010.endlessskye.com - author of this page apparently has the compiler on floppies, unclear if he ever got to imaging them... – sendmoreinfo Oct 2 '17 at 17:55
6

There is a port of an old gcc version (2.5.8) and binutils, see this archive.org link

5

I worked for a company that created a TMS34010 TIGA product in the before times. It was a pretty awful chip, all things considered. Very expensive, relative to competing products, made worse by requiring lots of expensive VRAM memory (for both code and data memory) for decent performance, even though they advertised you could use cheap DRAM to build cheap graphics systems (true in that DRAM == cheap and very slow). Memory bus was 16-bit, so if you were trying to do, say, 8-bit graphics, you could only grab 2 pixels per clock (~150ns-200ns depending on speed step). Math was integer only until the follow-on tms34020. 'IBM Standards Compliant' (e.g. SVGA, 8514/A, XGA) video cards offered more of what customers actually wanted at generally higher performance for far less money. Our product wasn't a market success.

The TI 34010 TIGA C compiler was pretty buggy as I recall; at least I remember the firmware guys complaining a lot. I have the toolchain for Sun-3/Sun-4 on a tape someplace,and someone who will preserve it. I wasn't aware of a GCC port...congrats again to the GCC guys for producing something that adaptable. 34010 is a strange architecture. Bit-level address-ability is pretty important to getting the most out of it; I would think that would make a C compiler 'interesting' (though the 8051 folks figured it out, so clearly it's a solved problem). Compilers aren't my area of expertise, though, so I'll leave that analysis for the experts.

Basically, the TMS34010 was everything that the 'experts' thought the market wanted from a graphic chip at the time (highly programmable, general purpose GPU) if they ignored everything that was actually happening at the time (fast, cheap, fixed function graphics primitives...mostly bitblt, line drawing, etc.). There's a reason TIGA had a really short lifespan.

  • TIGA was wonderfully flexible. In the days before Windows domination, you could implement all kinds of wacky graphics systems on it. However, hardware just for Windows could be simpler and cheaper, and sold in far greater volumes. – John Dallman May 9 at 19:29

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.