| 23 Dec 2023 |
 raitobezarius | But it's not really SMP | 18:46:12 |
 zrsk | anyway if someone wants to satisfy my curiosity and try to build a random derivation (I don't know, nix or hello maybe?) on an SBC and give me the time it takes (just the order of magnitude is enough) I would be grateful | 18:48:35 |
| zrsk | In reply to @raitobezarius:matrix.org
I don't know what you mean by emulated or see how it is relevant here Yeah don't worry, I've no really idea what I'm talking about probably. Thank you very much however :) | 18:49:21 |
| zrsk | Anyway I don't know if somebody already tried it but building RISC-V docker images with dockerTools.buildImage worked like a charm. I was impressed (as always) by Nix. | 18:53:35 |
 fgaz | In reply to @aciceri:nixos.dev
anyway if someone wants to satisfy my curiosity and try to build a random derivation (I don't know, nix or hello maybe?) on an SBC and give me the time it takes (just the order of magnitude is enough) I would be grateful
- hello: 2m dominated by configure, likely faster on nvme
- nix: I started the build, I'll let you know tomorrow
| 20:52:00 |
| fgaz | *
- hello: 2m dominated by configure, likely faster on nvme (I'm using mmc)
- nix: I started the build, I'll let you know tomorrow
| 20:54:37 |
| fgaz |
- It takes 3 days to build https://sr.ht/~fgaz/nixos-star64
| 21:20:29 |
| raitobezarius changed their display name from raitobezarius to raitobezarius (DECT 2128). | 22:22:30 |
| zrsk |
I'll let you know tomorrow
Not the best premises lol
Just tried hello and it takes 136s for me, probably using a Star64 as a builder is not worth it. In my case the disk speed wasn't the bottleneck surely (it's a fast nvme).
| 23:45:53 |
| zrsk | *
I'll let you know tomorrow
Not the best premises lol
Just tried hello and it takes 136s for me, probably using a Star64 as a builder is not worth it. In my case the disk speed wasn't the bottleneck surely (it's a fast nvme).
| 23:45:57 |
| zrsk | *
I'll let you know tomorrow
Not the best premises lol
Just tried hello and it takes 136s for me, probably using a Star64 as a builder is not worth it. In my case the disk speed wasn't the bottleneck surely (it's a fast nvme ssd).
| 23:46:40 |
| zrsk | In reply to @fgaz:matrix.org
- It takes 3 days to build https://sr.ht/~fgaz/nixos-star64
Your work there is great, do you keep images updated? Don't you host a cache also? | 23:52:18 |
| 24 Dec 2023 |
 Pratham Patel (you can mention me) | In reply to @raitobezarius:matrix.org
Then you can buy the MilkV Milk-V Pioneer? | 02:50:34 |
| raitobezarius | yes | 02:50:41 |
| Pratham Patel (you can mention me) | There’s also Oasis, lunching in under a year. No adjusted ETA after the bump in memory bus. | 02:51:47 |
| raitobezarius | there's also new JH chips around | 02:54:25 |
| raitobezarius | and the new SG too | 02:54:30 |
| Pratham Patel (you can mention me) | Yeah, JH8100 should be as fast as RK3588, at least in raw computer. Which is what we need for compilation. | 02:56:25 |
| Pratham Patel (you can mention me) | *raw compute power | 02:59:09 |
| Pratham Patel (you can mention me) | sorry, I just woke up :) | 02:59:17 |
| fgaz | In reply to @aciceri:nixos.dev
Your work there is great, do you keep images updated? Don't you host a cache also? Not yet but I plan to | 06:32:01 |
| fgaz | * I upload a new image from time to time. I don't have a cache yet (that's why I build fat images) but I plan to | 07:31:58 |
| fgaz | * I upload a new image from time to time. I don't have a cache yet (that's why I add some build dependencies to the images) but I plan to | 07:32:24 |
| fgaz | zrsk: nix took 96 minutes | 07:32:49 |
| fgaz | * Thanks! I upload a new image from time to time. I don't have a cache yet (that's why I add some build dependencies to the images) but I plan to | 07:33:42 |
 Steven Keuchel | In reply to @aciceri:nixos.dev
Hey folks!
I'm working with some RISC-V derivations and I'm a bit annoyed by the building speed (I'm using binfmt on a quite powerful x86_64 machine). I know that there is this cache (indeed I'm using it, with the same exact nixpkgs used to populate it) but still many derivations are not cached and at the same time I would like a more updated nixpkgs.
Isn't there a community builder, right? Also I tried looking for cloud providers that offer RISC-V without luck, maybe there is one instead?
I was also considering buying an SBC but they don't look so powerful (I checked the VisionFive2 and the Star64), I fear that I would get performances similar to my emulated RISC-V. I have a VisionFive2 and a Lichee Pi 4A. The VF2 is slower than QEMU user mode emulation on recent x86 CPUs (AMD 5950X / 3990X). The Lichee Pi 4A is slightly faster than QEMU. | 10:57:03 |
| zrsk | In reply to @skeuchel:matrix.org
I have a VisionFive2 and a Lichee Pi 4A. The VF2 is slower than QEMU user mode emulation on recent x86 CPUs (AMD 5950X / 3990X). The Lichee Pi 4A is slightly faster than QEMU. I've a comparable x86 CPU (AMD 7900x) so thank you for confirming that probably it's not worth it | 12:05:11 |
| zrsk | In reply to @thefossguy:matrix.org
Yeah, JH8100 should be as fast as RK3588, at least in raw computer. Which is what we need for compilation. Oops I didn't see this message, thank you too. Just tried and building hello on the Rock5B takes ~45 seconds (EMMC), it would mean almost one third of what I get with QEMU now, not bad. But I've to consider that building hello seems mainly a single thread operation (looking at htop while it builds). Probably when using 24 threads my x86 cpu would gain ground. | 16:16:55 |
| Pratham Patel (you can mention me) | Somewhat of a follow-up to that message:
JH8110 SoC: 2x dubhe-80 (low-perf, energy efficient cores) and 4x dubhe-90 (high-perf, said to be equivalent to ARM's A76 cores (what RK3588 has as high-perf cores))
Even though the CPU IP has the V 1.0 (vector) extension, the SoC doesn't. I don't know why. The H (hypervisor) extension is missing too, so forget h/w accelerated KVM on this.
| 16:25:38 |
| Pratham Patel (you can mention me) | Here's the [rejected] dtsi from the first patchset: https://lore.kernel.org/linux-riscv/20231127013602.253835-8-jeeheng.sia@starfivetech.com/ | 16:26:36 |
