Raptor Computing Systems Community Forums (BETA)
OpenPOWER ISA => Third Party CPU Discussion => Topic started by: mparnaudeau on August 17, 2024, 02:50:33 am
-
I have this question that stays in my mind for a long time ...
Power ISA is open, MicroWatt was announced and published in 2019, with the main actor Anton Blanchard saying that "the goal for Blanchard was to see if he could make it, and as a software developer, taking on a very low level hardware project was a challenge" (from Wikipedia but I attended the OpenPower Summit Europe the same year and understood quite the same). That did not reassured me but I had the hope that after the proof of concept, this enabled will be used as a base for real products and for example a small board to allow developers to play with it at low cost, and promote the architecture. And even more after the project evolved enough to make run Linux and Zephyr operating systems.
I think that's very sad that no board was created, that IBM has not officialy adopted this project or allowed resources to help structuring a side organism (OpenPower Foundation, Raptor CS ...) to do that.
What was missing? Why don't we get a board? Who could be interested in doing that? (if it's not too late)
@ClassicHasClass In 2020, you wrote "The possibility of a single-board Microwatt-based system (and fully reprogrammable, too) gets closer every day" so I suppose that you hoped the same.
It seems that the only implementation that went to the end is Kestrel. All other known initiatives seemed to be attempts that led nowhere unfortunately. There is an empty page about PowerPI at OpenPower Foundation (https://openpowerfoundation.org/groups/powerpi/ (https://openpowerfoundation.org/groups/powerpi/)). At Euro BSDCon 2023 (https://www.youtube.com/watch?v=Tj4Q-m_WEh0&t=220s&ab_channel=EuroBSDcon (https://www.youtube.com/watch?v=Tj4Q-m_WEh0&t=220s&ab_channel=EuroBSDcon)), there is also a PowerSBC mentioned, for which we find zero information elsewhere ...
-
I think everyone interested in processor development went towards RISCV instead, which has way more community energy. FPGA development is also very niche skill and it's way harder to do anything compared to pure software, so the general pool of people able to work on projects like microwatt is much smaller.
Open source projects like this only get done if a company decides to do them (and it seems like no companies think they could make money on it), or someone has the skills AND time AND interest (which is all a very large investment for an individual).
Maybe someone could develop a minimal carrier board for the Arctic Tern module, which does use microwatt. It has a enough peripherals to use as a standalone linux system. $600+ for that would be a tough sell given the limited performance, though.
-
Well, you kind of can: Arctic Tern is a Microwatt device. The main issue there is the expense, and it's intended as a replacement BMC instead of a cheap screw-around-with board.
-
I'd say cy is right, RISC-V has all the meme status. And hey, it is an open standard ISA; but people just don't know PowerPC is just as open as it so it's mostly just enthusiasts coming from the Power Mac or Amiga or the die-hard of die-hard free hardware enthusiasts, who might still find something to prefer about the RISC-V boards, like the Pioneer's Micro ATX form factor giving it more expansion slots and replaceable RAM. Though that's really rubbing up to the cost of a Blackbird by that point...
-
If RCS sought to develop and sell RISC-V systems, I would have no reservations about buying any. It's not that I have any particular loyalty to PowerPC or anything. It just happened to be the most mature, complete and performant solution available.
-
If RCS sought to develop and sell RISC-V systems, I would have no reservations about buying any. It's not that I have any particular loyalty to PowerPC or anything. It just happened to be the most mature, complete and performant solution available.
Honestly? I'd love to, if the resulting device were decent quality, priced more in line with the performance expectations of RISC-V, and open source.
C'mon, Raptor. Sell us a RISC-V Shrike.
-
You can already buy RISC-V systems today.
Milk-V (https://www.crowdsupply.com/milk-v/milk-v-pioneer)
DeepComputing (https://store.deepcomputing.io/products/dc-roma-riscv-laptop-ii-with-octa-core-cpu)
framework (https://frame.work/products/deep-computing-risc-v-mainboard)
But the performance is really sluggish.
-
Yeah, I haven't found a RISC-V system remotely in POWER9's class, let alone Power10 or (presumably) S1. So far it's a lot of sizzle and very little steak.
I wouldn't mind a Raptor RISC-V system, but I wouldn't want it to distract from whatever the nextgen OpenPOWER box is. Although I say that as a PowerPC bigot, I think I would get a lot more out of that than any RISC-V system right now.
-
If they could make one in the ~$800 range, I'd consider it as a NAS board. Don't need that much power for one of those.
-
HiFive Unmatched $740 (https://www.crowdsupply.com/sifive/hifive-unmatched)
-
I completely agree with Classic, I don't think Risc V can compete with the Power architecture, let's say it's a world apart.
-
HiFive Unmatched $740 (https://www.crowdsupply.com/sifive/hifive-unmatched)
If you want to play with the Unmatched you'll want to order from Mouser - they have the same board on sale for $299. (https://www.mouser.com/ProductDetail/SiFive/HF105-001?qs=Imq1NPwxi75JBw6ulD0quQ%3D%3D) Caveats apply: this is being clearanced out because SiFive's Premier P550 has been announced and will likely ship before the end of the year. And they are not fast - performance is north of a Pi 3B+ but well south of a Pi 4, and there is no SIMD whatsoever. Just about any GPU made in the last decade will be bottlenecked in one way or another, though I believe patches are landing about now to facilitate RISC-V operation with current Radeons. I can't emphasize the sluggishness enough - maybe it was on IRC where a dev talked about how his Power9 could emulate a RISC-V machine that was at least as fast as the Unmatched, and possibly with more threads.
-
I completely agree with Classic, I don't think Risc V can compete with the Power architecture, let's say it's a world apart.
That's why I went with raptor, riscv just isn't suitable for a desktop imo.
-
I looked into this a long time ago and determined that the minimal investment required to produce actual silicon based on Microwatt was somewhere around $20 million. If Microwatt (or POWER in general) had the kind of interest RISC-V has, perhaps someone could raise that amount of money. As it stands right now, the only place we could ever hope of getting an actual chip from is IBM itself and I don't see that happening.
As for boards, there are people around who are skilled enough with KiCAD that an excellent open-source board could produced if a chip existed. For now, those of us who like to experiment with Microwatt have to be satisfied with FPGA soft cores running at extremely slow speeds (100 MHz and less).
-
Banana Pi BPI-F3 (https://www.banana-pi.com/en/banana-pi-sbcs/175.html) approx. $99
-
You can already buy RISC-V systems today.
Milk-V (https://www.crowdsupply.com/milk-v/milk-v-pioneer)
DeepComputing (https://store.deepcomputing.io/products/dc-roma-riscv-laptop-ii-with-octa-core-cpu)
framework (https://frame.work/products/deep-computing-risc-v-mainboard)
But the performance is really sluggish.
Yes and no. I did a lot of work profiling several SBCs with ARM, RISCV and x86 CPUs. I found that the SG2000-based Milk Duo-S had single thread performance comparable to the 1GHz ARM Wandboard Quad and Celeron-based Udoo x86 Advanced boards for general purpose tasks. The difference is that the SG2000 used 0.6 watts at idle and 0.8 watts at full power, while the WB Quad used 2.7 and 3.4 watts and 9.1 and 10.1, respectively. Another difference is that the SG2000-based Milkv Duo-S cost $13.
Of course, both the Wandboard Quad and Udoo x86 Advanced have multiple cores, and easily outperformed the single core SG2000 in multithreaded tasks. In most cases, the power use scaled with threads to the maximum core count, but less than 1:1 (it was more efficient to bring up additional cores). I did the same tests on Raptor's POWER9 Talos II systems, and obviously performance was dramatically better, and the power usage was much higher, but not like the G5 and Cell systems I tested (both used over 100 watts at full power). For comparison, I did the same tasks on a 1.25GHz Mac mini (PowerPC G4), and while performance was almost 1/3 better, it used 17.0 and 31.6 watts, respectively (and was faster than the 2.2GHz PowerMac G5 I tested).
I think the SG2000 suffers from very small i- and d-cache sizes. Multicore support would obviously be a huge benefit. What is also notable about the SG2000 is that the entire boot process can be replaced, including the firmware. This is done as part of Lup Lee's successful efforts to port NuttX to the board.
Personally, while I am a huge fan of PowerPC/POWER architectures, in my efforts to port and/or build compiler and assembler pieces for the architecture, I have become dismayed at the sheer number of instructions the CPUs support. I have been reading through the OpenPOWER specifications, and with each version new instructions are added. I feel like the "Reduced" part of RISC has been lost, and the boundary between CISC and RISC seems very blurry. I have not been privy to the conversations about the additional instructions, but I do wonder if they are truly orthogonal and if they can maintain one instruction per cycle completion. If there is evidence of stalls or synchronization issues, perhaps it would be better to let the compiler build more steps from fewer instructions.
In comparison, RISC-V sticks to a limited instruction set and maintains one instruction per cycle. Yes, there is less pipelining and yes, no SMT-type re-use of CPU resources, but that also makes for a simpler design, which makes the entire production process cheaper.
I would really like to see a fully open PowerPC/POWER SBC with a goal of making it into a laptop and cellphone, targeting human rights defenders and other groups that are increasingly targeted by authoritarian governments. Although the market size would almost certain make production units expensive, as the adage goes, freedom isn't free.