Swordfish II
Advanced Member
- Joined
- May 20, 2015
- Messages
- 1,175
1. Linux support
2. Likely accept small volume orders
3. Way more powerful than omap5
4. 1.5x better than A72
Seems all around worth a look. If not it's cool
The Future of Pyra's CPU
- Thread starter Rami
- Start date
-
- Tags
- cpu pyra replacement
1. Linux support
2. Likely accept small volume orders
3. Way more powerful than omap5
4. 1.5x better than A72
Seems all around worth a look. If not it's cool
Letalis Sonus
Advanced Member
- Joined
- Mar 5, 2009
- Messages
- 1,530
Nice PR stunt you got there, it would be a shame if something happened to it...
Incredibly Scalable High-Performance RISC-V Core IP
www.sifive.com
RISC-V (open source) and 1.5x better than an ARM a72
You're talking about a startup there whose most powerful product on the market (the devboard costs 1k US$!) still gets absolutely mutilated by those ARM CPUs, only trust their words as far as you can throw them.
Last edited:
Phlyra
FLOSSing
- Joined
- Jul 25, 2019
- Messages
- 1,517
Hopefully it’ll get better with future updates of this board: https://www.cnx-software.com/2020/0...pen-source-alternative-to-raspberry-pi-board/
levi
Still fresh, damnit!
I'll wait until the actual adverts on the side are actually selling risc-V boards.
levi
Still fresh, damnit!
Speaking of risc-V board but not really about a future SoC for Pyra needs (but this is still the best thread for this post, I think), the Indians have produced a new RISC-V core which they claim is clocked at 75MHz and a little above:
The only fly in the ointment is that screen update looks about ARM2 level or maybe ARM3. IIRC ARM2 chips were fabbed at 180nm, and only ran at 4MHz as I recall. But we're looking at a serial connected screen here I guess, so that might actually be the limiting factor*.
I've downloaded a copy of the core ABI for RISC-V chips, but I've not really grokked it yet. I'm not sure what sort of pipelining it has and how long it can run at full pelt without having to clear that pipeline (simple non-predicting CPUs generally invalidate their pipelines when the do a jump, then it takes pipeline length ticks to execute the next instruction after the branch). Maybe it also had indirect addressing modes which require more than 1 tick to execute, I can't remember.
But as the third RISC-V CPU fabbed in India using entirely Indian design this can't be sniffed at.
* In my current terminal font, I get about 145 characters wide on this widescreen laptop. Assuming therefore that that screen is 145x30, and the screen paint finished in about 2 seconds by my untimed reckoning and it filled most of the width, that's a baud rate of a little over 2kcps.
The only fly in the ointment is that screen update looks about ARM2 level or maybe ARM3. IIRC ARM2 chips were fabbed at 180nm, and only ran at 4MHz as I recall. But we're looking at a serial connected screen here I guess, so that might actually be the limiting factor*.
I've downloaded a copy of the core ABI for RISC-V chips, but I've not really grokked it yet. I'm not sure what sort of pipelining it has and how long it can run at full pelt without having to clear that pipeline (simple non-predicting CPUs generally invalidate their pipelines when the do a jump, then it takes pipeline length ticks to execute the next instruction after the branch). Maybe it also had indirect addressing modes which require more than 1 tick to execute, I can't remember.
But as the third RISC-V CPU fabbed in India using entirely Indian design this can't be sniffed at.
* In my current terminal font, I get about 145 characters wide on this widescreen laptop. Assuming therefore that that screen is 145x30, and the screen paint finished in about 2 seconds by my untimed reckoning and it filled most of the width, that's a baud rate of a little over 2kcps.
MrConfusion
Very Active Member
- Joined
- Apr 11, 2013
- Messages
- 331
Yesyesyesyesyes... But can it run Crysis?
;-)
And when even RISC-V is not enough you will apparently soon have the option to use a handheld where you can replace your RISC-V with a Microwatt or whatever (just guessing, I haven't checked whether the FPGA can hold it).
Good old bunnie from Singapor seems to be after a Precursor for those willing yet moar coolness through obscurity (not really, it looks all very transparent, that's precisely why there's so much to understand).
Spoiler: no game controls in this handheld (unless you hack your own into it). Also, the crowdfunding is not even open yet.
Yet, I don't think we want an FPGA as our next Pyra CPU, do we ?
Last edited:
I only know what's in the Precursor crowdsupply page I linked above. It links to some update pages. I haven't read them all. Basically a small computer including wifi, USB-C, keyboard and screen (a little Blackberry-like) with an FPGA and basic crypto hardware to be used for signature, 2FA, wallet, etc. But since it's an FPGA you could change it all and use it to emulate a 6502, Z80 or something. I think it says it's around Nintendo DS powerful. The FPGA is running a RISC-V 32 core as shipped, I think.
God Ginrai
Godmaster
To be quite honest, I think having FPGA tech on the next Pyra CPU board would be great. Obviously we'd still need a main CPU. But if we could get our CPU to dynamically set an FPGA depending on what system we are emulating, then that would be awesome. Not sure if FPGA tech (and powerful enough) is sufficiently small enough for this to be useful yet, tho'.
-God Ginrai
Of course an FPGA is nice to have. But, even if I'm no expert, I don't see how it can fit. FPGAs are flexible, But the space, cost and power budget will always mean than a no FPGA board will be faster, last longer on battery and give functionality closer to what the PC equivalent of the time would, compared to a board with part of that budget allocated to a FPGA. Otherwise people would only use FPGAs for any application (there would be no ASICs, no general purpose CPUs or SOCs...). I don't imagine the flexibility (or in case of Precursor, the trust provided by the transparency of open cores) will weight enough for a Pyra like device to offset cost, efficiency or throughput.
The Xilinx Zynq Ultrascale series could be interesting, they have ARM Cortex-A53 cores, an ARM Mali-400 GPU and a Xilinx FPGA fabric. So you could run a useful Linux including games on the ARM cores and GPU, and add any extra hardware you want to emulate in the FPGA fabric. The chips seem to be very expensive, though.
Letalis Sonus
Advanced Member
- Joined
- Mar 5, 2009
- Messages
- 1,530
The UltraScale+ (not to be confused with the old UltraScale chip family!) gets way too hot, though, even if you don't use the beefy FPGA on that chip. I've been working with one of those beefy 10k € evaluation boards for a while some time ago.
My opinion on this whole topic are :
Why do whe need a new CPU Board yet?
Whe ditnt even know what this current OMAP 5 Board is apple to do as there wherend that much Pyrae out..
Why do whe need more power? The Pyra is a Pocket PC after all, so if you want to make something that needs more power, use a RAPI at Home that is connectet to the Wall Plug..
Great Power tends to Great Current Draw, thats an OLD Yeti Whisdom..
My other concern is compability and that a newer, faster CPU Board draws the attention of the Defs away from the OMAP 5 Pyra, so they get abandoned..
Also: Lets Say, i play Final Fantasy VII on my Omap 5 Pyra, then i get to the Middle of the Game, and then i should upgrade my Pyra to a faster CPU Board because the newer Update for PCSX Rearmed Pyra want be on Omap 5 anymore,,
How it is guarantiet that i can use my OMAP 5 Savegame??
I know, i dont have that much knowledge, i ditnt have a high Computer Background, but therefore i have to ask..
Why do whe need a new CPU Board yet?
Whe ditnt even know what this current OMAP 5 Board is apple to do as there wherend that much Pyrae out..
Why do whe need more power? The Pyra is a Pocket PC after all, so if you want to make something that needs more power, use a RAPI at Home that is connectet to the Wall Plug..
Great Power tends to Great Current Draw, thats an OLD Yeti Whisdom..
My other concern is compability and that a newer, faster CPU Board draws the attention of the Defs away from the OMAP 5 Pyra, so they get abandoned..
Also: Lets Say, i play Final Fantasy VII on my Omap 5 Pyra, then i get to the Middle of the Game, and then i should upgrade my Pyra to a faster CPU Board because the newer Update for PCSX Rearmed Pyra want be on Omap 5 anymore,,
How it is guarantiet that i can use my OMAP 5 Savegame??
I know, i dont have that much knowledge, i ditnt have a high Computer Background, but therefore i have to ask..
We don't, I agree. I'm not sure whether we'll ever need another CPU, but surely we don't need it now.
I don't know about games or proprietary software, so I can't help you with your question. I tend to think that free software communities tend to mix and match and can use the contributions of other groups for new hardware, so fragmentation could not be so much of a problem in freedom as it could be in proprietary platforms, but it's still a lot of investment without having yet recovered the first investment. so no hurries.Let's get the Pyra out first, then grow the software ecosystem a little, and then we'll see.
Last edited:
Thanks for the info. I figured as much as the $250 Ultra96 board needs a 8V 3A PSU.
levi
Still fresh, damnit!
Open source communities generally built their own source if they need to. That can be done on the Pyra too, but there's a bigger problem IMO. Certainly for games and emulators it's not just about can it run, but can it run well. I might expect some hypothetical future pyra cpu to be throttled to fit into a similar power envelope, so actually they might not behave so differently in practice, something more like a 600MHz Pandora (that can be overclocked stably to about 800MHz) versus a 1GHz Pandora (which can be overclocked a little above 1100MHz). Then again you've also got problems if the next chip supports Vulkan while our existing chip is a little over OGL2-ish.
Kev2442
Still French
I think the thing is, we have no idea how long it would take to release a new CPU board, so it might be valid to get a head start on it. If it takes two years, by then new use cases could arise where the current CPU just won't cut it. I'm afraid it already is a bit short on the video side.
Soon enough many houses will be equiped with 4K screens, especially in the tech enthusiast communities we're a part of. So being able to output 4K from the Pyra's HDMI out would be great. And as more and more software is multithreaded and not requiring high clockspeeds, more cores would also be a nice thing. This is just two of the current CPU's limits, we might not feel constrained by its limits yet, but that doesn't mean there are none.
I see.
But then, if we start choosing a SOC before someone is ready to put the work and money to produce a new board, the same can happen once again, can it not ?
I mean the SOC we pick now can be outdated by the time the investment on the original Pyra is recovered and the user base expanded, and ecosystem blosomed, and we don't know now what SOC options will there be...
Like we pick now a SOC that can output 4K on HDMI but then by the time the board starts to be designed people have moved to 6K or something.
Anyway, I'm not putting the money nor the hardware expertise, so I have no idea how soon or late is now to talk. So if you want to talk about alternative SOCs, go ahead.
levi
Still fresh, damnit!
I'm a tech enthusiast, but not like that. This laptop I'm typing on now is 12 years old, and similarly my video server is of a very similar vintage (at least according to the CPU date; it was second hand). I like being able to do modern stuff on old kit, and I never buy things just because they're the newest shiny. I buy infrequently and try to buy the best when I do so that it'll last me. The Pyra was something of an exception to that for me, but I enjoyed the journey with the Pandora once I actually got one in my mitts and my Pandora was a bit tired by the end. The Pyra should be closer to the ideal of something the lasts for me.
