I just installed the 1.6.7 julia on my Raspberry Pi 400. It runs but the
TTFX is excruciating. I would like to install 1.10.x+ but there appear
to be no binaries from the julialang.org site.
Has anyone successfully built the latest version of julia for the RPi 4
hardware? I’m in the process of determining whether to go with a
native build or to try one with cross compiling.
Any suggestions are welcome!
Seems to exist under the downloads tab…
That said, you might want to install via Juliaup which will make future updates much easier.
The Raspberry Pi 400 is arm7 with hardware float and not arm8
which is the AArch64 version. I did try downloading and then discovered
the incompatibility. Thanks for the thought.
The Raspberry Pi 400 hardware is ARMv8. Perhaps you are running a 32bit version of the OS?
Check the output of uname -m, if you are running the armv7 version, try installing the 64 bit version of Raspbian
Thanks for the tip!
It never in a million years would have occurred to me that the default
OS for 64-bit hardware would be 32-bit. I’ll give this a try as soon as
I can burn a new SD card.
I guess there is a reason that the last five letters on default are f-a-u-l-t.
A final follow up:
I was successful installing the latest version of the Raspberry Pi OS desktop,
64-bit version and was able to install julia-1.10.4.
Unfortunately, trying to install Makie triggered an LLVM crash which I am not
able to debug at this time. Seems likely that the problem was running out of
memory as the RC1 install had a 2000MB swap file size set.
I then installed the Upcoming release: v1.11.0-rc1 (June 25, 2024) and was
successful building Makie and CairoMakie.
What was the error message precisely? Was it by any chance
JIT session error: Cannot allocate memory
?
Sorry, I wiped that OS install on the Raspberry Pi and cannot
reproduce the run to get the precise error message.
If I get another LLVM error, I’ll try to copy and paste the text
so I can provide the exact output message.
I have used the Raspberry PI always with the Raspberry PI OS. I have thought of using Ubuntu but never have. This is an option however.