Show HN: OpenRISC JavaScript emulator emulates up to 16 cores
github.com
github.com
Please do get in touch with me if the router goes down. I can't believe it's stayed up this whole time!
Edit: Please do read the README.md on GitHub, but demo links are here:
Demo (1 core): http://s-macke.github.com/jor1k/
Demo (2 cores): http://s-macke.github.io/jor1k/index.html?cpu=smp&n=2
And increase the "n" querystring argument for number of cores up to 16.
EDIT: Looks like the RISC-V FAQ references the same blog post since so many people ask the same question.[3]
[1] https://blog.riscv.org/2014/10/why-not-build-on-openrisc/
[2] http://en.wikipedia.org/wiki/David_Patterson_(computer_scien...
Joking apart, as a developer of jor1k I am also interested in RISC-V and constantly follow their progress. I am still not sure, if RISC-V is just OpenRISC 2.0 or a competitor. The technical keypoints of criticism of the OpenRISC architecture are true, but show me an architecture, which is free of bad design decisions?
The problem is usually not to build a new instruction set. It is in fact the easiest part I would say. The most time consuming part is to build the toolchain including, emulators, Linux, gcc, binutils, libc and tons of architecture dependent libraries and programs (gdb, strace, ltrace, libffi, xpcom, ...) . And here RISC-V begins again from scratch which is a little bit sad.
You forgot to tell, that the RISC-V team build their own Javascript based emulator at
They're all somewhat MIPS-like, and that's mostly due to their heritage and the relative ease of creating them. What's more, they all seem to have similar (not so great) performance characteristics. Currently the only "RISC" ISA that has good performance/efficiency characteristics is the ARM, and that one has some very un-MIPS-like predicated instructions, combined shifts, and more complex addressing modes (including post-inc/decrement.)
Meanwhile, what AMD and Intel have done with x86 both in terms of performance and efficiency has been quite amazing, so I think what would be interesting to see is an OpenCISC/x86 that draws some inspiration from them, or at least something more ambitious and imaginative than yet-another-MIPS-clone. Otherwise, seeing all these projects attempting to revive the RISC fad, which is long over, feels rather misguided.
http://s-macke.github.io/jor1k/compile.html
So you can use it in a C programming course for SMP systems. OpenMP is supported for example.