#asahi-dev · 2022-02-19 11:23
Glanzmannpovik: I think it is unrelaed, maybe I had it just on in my usually broken config without any reason whatsoever.#asahi-dev · 2022-02-19 11:21
Glanzmannmps: Debian bootstrap.sh needs 8 minutes on the air, 4 on the ryzen ...#asahi-dev · 2022-02-19 11:20
Glanzmannmarcan: I see.#asahi-dev · 2022-02-19 11:20
GlanzmannWe don't need that, do we? https://pbot.rmdir.de/u/V5wo5cz6XyHacO6BGpYEzQ#asahi-dev · 2022-02-19 11:17
Glanzmannnot*#asahi-dev · 2022-02-19 11:16
Glanzmannmarcan: Yes, it did, but I started testing kernel options from the bottom up, now top down.#asahi-dev · 2022-02-19 11:15
GlanzmannYou found it faster than I, because that was the last thing that I checked.#asahi-dev · 2022-02-19 11:15
Glanzmannmarcan: Yes, I did. Forgot that my build box is now an amd64 ...#asahi-dev · 2022-02-19 11:12
GlanzmannI also get this: https://pbot.rmdir.de/aKbAxtC0HVErbRD03D9-mw maybe it is helpful.#asahi-dev · 2022-02-19 11:12
GlanzmannThanks.#asahi-dev · 2022-02-19 09:09
Glanzmannjannau: I see, thanks. I did not know.#asahi-dev · 2022-02-19 08:54
Glanzmannjannau: So I did not saw these patches, but maybe I was not looking right: https://tg.st/u/9ce9060dea91951a330feeeda3ad636bc88c642c.patch https://tg.st/u/256f5efbf23ff68c489dad92f99d1cecfb021729.patch https://tg.st/u/8737955a0263d09ffa8550658dfcac1df3d0665c.patch#asahi-dev · 2022-02-19 08:52
Glanzmannjannau: Are your dtsi and spi patched for the device tree for u-boot in?#asahi-dev · 2022-02-19 08:50
Glanzmanntoo.#asahi-dev · 2022-02-19 08:39
Glanzmannknow*#asahi-dev · 2022-02-19 08:39
Glanzmannj`ey: You should no him better than that.#asahi-dev · 2022-02-19 08:38
GlanzmannYes, he did. Small performance improvement.#asahi-dev · 2022-02-19 08:33
Glanzmannmarcan: \o/#asahi-dev · 2022-02-19 08:31
Glanzmannrkjnsn: Yes, he did.#asahi-dev · 2022-02-19 08:25
GlanzmannYep, without the first patch with cpufreq 'nvme' hangs. Second one for performance.#asahi-dev · 2022-02-19 08:24
Glanzmannmarcan: https://git.kernel.dk/cgit/linux-block/commit/?h=m1-test&id=e226b395bc2d49a6a1411765875e0864c9a0e0da#asahi-dev · 2022-02-19 08:24
Glanzmannhttps://git.kernel.dk/cgit/linux-block/commit/?h=m1-test&id=e493225b2692e663729a79754065c3fcbe2d3243#asahi-dev · 2022-02-19 08:23
Glanzmannmarcan: Two patches from axboe: #asahi-dev · 2022-02-19 07:46
Glanzmannmarcan: This was with axboes patch, I just wanted to show the file system overhead.#asahi-dev · 2022-02-19 07:42
Glanzmannmarcan: I don't know if that is relevant, but if you repeat the random test with the files in place I get 70k IOPS: https://pbot.rmdir.de/NAZqGckf6JqZoyrwOvlMnQ#asahi-dev · 2022-02-19 06:17
Glanzmannpovik: mini and air target are online.#asahi-dev · 2022-02-18 14:48
Glanzmanntoo.#asahi-dev · 2022-02-18 14:47
GlanzmannYep, it doesn't.#asahi-dev · 2022-02-18 14:47
GlanzmannI see, I I copied it from mps. :-)#asahi-dev · 2022-02-18 14:46
Glanzmannaxboe: About jumpy touchpads that helped me a lot: https://git.zerfleddert.de/cgi-bin/gitweb.cgi/m1-debian/blob/ae43a6ec466a273a5867834c6e066d7158078d99:/files/quickstart.txt#l47#asahi-dev · 2022-02-17 07:30
Glanzmannmarcan: axboe said yesterday that there is no such thing as 'barrieres' for 15 years. And that it is impossible due to multiqueue support IIUC.#asahi-dev · 2022-02-16 21:03
Glanzmannsven: Have a good night sleep. I'm curious if you find something.#asahi-dev · 2022-02-16 20:15
GlanzmannI'm also of in favor for the patch, but I begged for months for it.#asahi-dev · 2022-02-16 20:13
Glanzmanntpw_rules: https://pbot.rmdir.de/4emSWC5TySRwHeJBhexadg Yes#asahi-dev · 2022-02-16 20:07
Glanzmanntpw_rules: I notice when I install debian packages (which syncs after extracting every package) that it is dog slow without a measuring tool. That is how I noticed that we had the problem in the first place. Because the nvme felt like a spinning disk.#asahi-dev · 2022-02-16 19:54
GlanzmannI see, will do that in the future. I normally have scripts rming the files between runs ...#asahi-dev · 2022-02-16 19:54
Glanzmannaxboe: Okay it must have been an overwrite, because: https://pbot.rmdir.de/INSyOFFbbiBinZsyUYoPpA#asahi-dev · 2022-02-16 19:52
Glanzmannnoticed*#asahi-dev · 2022-02-16 19:52
GlanzmannThat was the first fio run I ever did on this particular installation. before I did not benchmark because I already notices that the issue was gone due to how fast apt was installing packages.#asahi-dev · 2022-02-16 19:52
GlanzmannNope.#asahi-dev · 2022-02-16 19:51
Glanzmannhttps://pbot.rmdir.de/IyW42yKcd5J6v7Dcy9jsLA#asahi-dev · 2022-02-16 19:51
Glanzmannaxboe: I have a macbook air with worse specs than yours.#asahi-dev · 2022-02-16 19:50
GlanzmannI see. Btw. with your patch I don't notice any additional latency. And my fio test comes also back with: 64K, how can that be? https://pbot.rmdir.de/Bizcz1sc3Vs3EZ1raP31kA#asahi-dev · 2022-02-16 19:48
GlanzmannAfter reading up on it.#asahi-dev · 2022-02-16 19:48
Glanzmannaxboe: I wasn't aware of O_DSYNC. But I get the idea.#asahi-dev · 2022-02-16 19:47
GlanzmannI see. I saw that probably while benchmarking nfs volumes.#asahi-dev · 2022-02-16 19:45
GlanzmannYep, but when doing I/O tests I sometimes notice that while the test is running throughput and iops are faster than when the fsync is commenced and than the results will be 1/3 or so less than what was shown during the workload.#asahi-dev · 2022-02-16 19:44
Glanzmanns/and/an/#asahi-dev · 2022-02-16 19:44
Glanzmannaxboe: Maybe but there was also and end-fsync ...#asahi-dev · 2022-02-16 19:41
Glanzmannaxboe: About the 12k. Do you have an explanation why macos can do 18.9k and OpenBSD can do 31.2k? (OpenBSD don't do any flushes at all IIRC, but kettenis knows for sure).