> RISC-V is an open-source instruction set architecture (ISA) poised to make personal computing more, well, personal
Eerrr, how exactly? RISC-V is poised to make companies not pay the ARM fee. That's it.
Eerrr, how exactly? RISC-V is poised to make companies not pay the ARM fee. That's it.
Starting with the opening paragraph
> BURIED IN THE INNER workings of your laptop is a secret blueprint, dictating the set of instructions the computer can execute and serving as the interface between hardware and software. The instructions are immutable and hidden behind proprietary technology. But starting in 2025, you could buy a new and improved laptop whose secrets are known to all. That laptop will be fully customizable, with both hardware and software you’ll be able to be modified to fit your needs.
ending with the closing paragraph
> ”If we can open up even the chip design, then one day, even students at schools and universities can come into class and design their own chips, with open tools,” says Liang. “With openness, you can choose to build things yourself from zero.”
It's easy to run your own RISC-V core at 50 or 100 MIPS in a cheap FPGA.
You can make a custom chip for $150 at TinyTapeout. A pretty small chip, but you can fit a simple RISC-V cpu in if you use multiple tiles.
https://tinytapeout.com/runs/tt05/tt_um_kianV_rv32ima_uLinux...
It boots (noMMU) Linux:
https://x.com/splinedrive/status/1874225296335134867/photo/1
> as slow as an $80 raspberry pi 5
The current Framework board is similar to an Arm A55 e.g. Odroid C4, somewhere between Pi 3 and Pi 4. More like Pi 4 in practical terms because Pi 3 only has 1 GB RAM which is extremely restricting, while current RISC-V boards can take 8 or 16 GB -- or 32 GB on the Milk-V Megrez and SiFive HiFive P550 which are probably a bit faster than a Pi 4, other than not having SIMD.
That lets you do serious work, even if not quite as fast as one might like.
Something comparable to the Pi 5 needs to wait another 12 months maybe. Deep Computing are talking about making one late this year in the article.
The small market is a major reason for the high price -- limited units to recoup the development cost from.
It's a chicken and egg thing, and you have to start the cycle somewhere, assuming you want to actually do something new, not just make internet posts.
The same goes for FrameWork itself, even the Intel and AMD ones.
Someone needs to find a "tesla of risc-v chips" that's insanely fast, even if its a premium price. Once the image is shaken, we can move into commodity chips.
But I guess we have the (slow) golf cart for now.