Turbo9 - Pipelined 6809 Microprocessor IP

Join Kevin Phillipson and Michael Rywalt as they explain the design of the Turbo9 Microprocessor IP. The Turbo9 is a compact high performance microprocessor IP that executes the Motorola 6809 instruction set. It is a modern implementation using a pipelined micro-architecture to give performance matching or exceeding modern RISC microprocessors while being smaller and more compact.
Kevin Phillipson has a BSEE and a MSEE from the University of Florida and is currently pursuing a PhD from UF. Michael Rywalt has a BSCS and a BSSE from Florida Institute of Technology and is currently pursuing a MSEE from UF. The Turbo9 is the basis of their research.
Link to Kevin's TH-cam Channel: / @turbo9team
VCF Southwest is made possible by the efforts of the Vintage Computing Collective of North Texas, an 501(C)3 non-profit. Please consider a donation to help us continue to produce shows and content like this. www.paypal.com/donate/?hosted...

[@winstonsmith478]

6809 - great choice with a highly optimized instruction set.

[@8bitwiz_]

This shows that those Motorola guys were right after all. The 6809 deserves to be one of the few CISC architectures to get a microcode translation architecture like x86 did.
I'm still sad that they abandoned 6809 in favor of extending 6800 for future chips (68HC11 etc.)

[@JoelReesonmars]

I don't think they made a mistake in focusing on the 6805. That made them a lot of money and they were able to extend the IPA and keep it relevant. It's still a good architecture, even though I would have added a separate parameter stack when they added the push and pop (PSH/PUL) instructions to the 68HCS08 derivative. (Or, actually, I would have added it at least as early as 146805.)
I also don't think they made a mistake in extending the 6800 architecture in the 6801, except that I would have split the 6801 into the one we have and another that moved some op-codes around to squeeze in direct-page addressing mode unary instructions, and added at least one address line to allow moving the direct page out of the absolute address map. (Among other things.) I think the market would have supported both directions.
They did make a mistake in failing to supply the 6809 as an SOC core to general customers, and in failing to extend it. Just adding segment registers and a few more small tweaks (completing the DP addressing, adding the FDIV and IDIV instructions that the 68HC11 got) would have pitted the 6809 head-to-head against the 8088.

[@timothyp8947]

Very interesting material on producing a modern design for an older (but much loved) ISA aimed at a industrial-grade implementation. I wish the team - and the TurboR - much success.
From the PoV of the retro hobbyist, and as touched on in the video, it would be harder to reap the benefits as a ‘drop in’ due to the timing dependence of ‘80s software, but one could easily imagine a new ‘retro style’ machine using this, a Dragon or CoCo ‘++’ if you will.
Thanks for a great presentation!

[@JamesJones-zt2yx]

That's why there are operating systems. I'm hoping for a version of NitrOS-9 that can run on Turbo9.

[@AlexEvans1]

CoCo software is generally a lot less timing tight (for whatever reason) than a lot of other 8-bit machines

[@JoelReesonmars]

@@JamesJones-zt2yx It should just run.
Well, sort-of. A few drivers would need to be modified, to deal with timing issues.

[@properjob2311]

6809 is a great cpu

[@rabidbigdog]

The best thing about this is that you could/can run a real operating system; OS-9.

[@alexloktionoff6833]

Will Turbo9 fit into ice40?

[@turbo9team]

We have not synthesized the Turbo9 for any of the Lattice Ice40 family, but it should fit in one of the versions... LUT/Logic cell utilization is hard to translate between Altera/Xilinx/etc technologies, but a good measure of device utilization is number of flops. The Turbo9 uses less than 500 flops. Keep an eye on our youtube page for the pending GitHub release. Thanks!

[@FlyingPhilUK]

Very nice talk...
Is the IP downloadable from anywhere?
It would be interesting to see how this compares to the 65C816 in terms of efficiency
- the '816 looks to be very efficient, but you need to keep track of the Mode bits
Also, any thoughts on extending it beyond 16-bit addressing
- I think this was one of the limiting factors for the oriinal 6809.. the world wanted to move beyond 64K by the time it arrived...

[@VCFSW]

@turbo9team Should be able to answer your question.