Slackware ARM -current - updates & performance - post-June 2019
I've been working with Slackware ARM -current for a number of years and during that time I have very much become accustomed to, and familiar with, the way it generally works and runs. I run the OS on several ARM devices and I'm seeing a very noticeable speed increase 'across the board'. [forgive the pun] :>
With the latest updates [post-June 2019] something has fundamentally changed. For sure, it's the same wonderful Slackware ARM that we all know and love, but it's like MoZes has turbocharged the code, or strapped a rocket-booster to the build farm, or something. Is anyone else finding that Slackware ARM -current seems to want to run the commands before you finish typing them? When processes are initiated are they faster to complete? Everything on the system just seems to happen "super-allegro fashion" with added Va-va-voom! I'm finding the output from simple tasks like 'ls' and 'stat' used to happen in the blink of an eye but now it appears the INSTANT you hit the return key. Compiling is quicker too. I have a test-build bash script which takes 17-18 minutes to complete on the Raspberry Pi 4 under Slackware ARM -current. Or should I say it used to take that long. Now, when I run that same script it completes within 12-13 minutes. The hardware and configuration is the same. The only thing that's changed is the Slackware ARM updates and the RPi firmware. It's not the firmware because the Orange Pi and Hummingboard doesn't use RPi closed-source firmware and they exhibit the same [similar] speed increase too when running Slackware ARM -current. Let's face it, Slackware ARM has never been slow or sluggish on any device [unless the device was struggling and under-powered by design] but I find that now, compared to before June 2019, it is A LOT quicker than what I'm used to. Is it just me or does anybody else think/feel the same? I have consumed quite a large amount of Jaffa cakes recently. Perhaps that has something to do with it. Hmmm. |
Any kind of something like stress we can run and compare notes on?
|
Quote:
|
Quote:
Code:
#!/bin/bash Code:
mkdir -p /tmp/_build-dir && cd /tmp/_build-dir && wget -nc http://sarpi64.fatdog.eu/files/extra/SARPi64.SlackBuild-aarch64-cc.txt \ Code:
############################################ |
Quote:
|
Quote:
glorsplitz, you are NOT going to believe this. I swear to you something is going on, and MoZes is NOT saying what it is/was/might be. This is what happened this morning... I installed a brand new fresh version of Slackware ARM -current on the Raspberry Pi 4 for testing. It was much much much much quicker than usual. I did not make a note of the exact time but I'd say it was approx. 1 hour for a full installation. It was unusual but not really noteworthy. Just unexpectedly quick. I installed all the available updates via slackpkg from slackware.uk repo. However, I got around to the main purpose which was to test some revisions to the SARPi64.SlackBuild-aarch64-cc.sh. So, I ran the script and fully expected it to take about 2.5 hours to complete, which would be considered "normal amount of time" for this process over the past month or so. Code:
############################################ MoZes is [or has been] up to a lot of something! I am not imagining this. Hand on my heart, I swear, to all that is pure and Holy, the above is the unequivocal truth and facts. Try it for yourself!!! |
Exaga, yes 69 compared to 153 minutes is definitely something's going on noticeable difference.
Kind of have busy work schedule weekly, hoping to try something this weekend. |
Quote:
|
From Which Pi have I got?
Code:
cat /proc/cpuinfo | grep 'Revision' | awk '{print $3}' | sed 's/^1000//' Quote:
Quote:
Isn't the better performance because the RPi4 devices are just better faster stronger? |
It looks like you are including download time in your timing, that's gonna change a lot. Download first, then time only the compile, and I'll bet you get more consistent times.
|
Quote:
|
I ran a script that usually takes several hours to complete, and it's completed in under 30 mins.
I don't know what's changed to facilitate that though - I suspect the Kernel. |
Quote:
Which device was this on MoZes? |
Quote:
Anyway, in this test the same; Raspberry Pi 3 Model B+, 32Gb microSD card, [as near as dammit] kernel .config, cables, power supply, was used on Slackware ARM 14.2 and -current. Both systems were installed, with an exact set of packages, on the same SD card and updated with the latest everything available. Bearing in mind Slackware ARM 14.2 uses gcc 5.4.x and -current uses gcc 9.1.x which is one exception and major difference here. I used the 'SARPi3 - Slackware ARM installer disk image' build process [bash script] as a test-subject. Identical in every detail by way of execution - the only difference in these two build processes is the version of minirootfs downloaded from your archive. Everything in the kernel/modules/etc. was pre-compiled in order to carry out this test on a quick-turnaround basis. The purpose of this bash script is to pull everything together and fashion a disk image that can be written to a SD card, which will then boot into the Slackware ARM installer on a Raspberry Pi 3/3B+. It's not a long process but involves a lot of data moving about the system in very quick succession. I thought it would be perfect candidate for this test. The results: Slackware ARM 14.2: Code:
SARPi3.SlackBuild sarpi3-installer_slack14.2_10Aug19_sp1.img ... Done Code:
SARPi3hf.SlackBuild sarpi3-installer_slackcurrent_10Aug19_sp1.img ... Done Whatever the reasons or causes may be, I think this is GREAT! :D Thanks a lot MoZes, and thanks to the Slackware team as well. :hattip: |
I'll bet its GCC adding new support and new optimizations for ARM. Lots of differences between GCC 5 and 9. And it would affect everything, kernel, apps, drivers. Everything.
If you were really curious, you could try to install different versions of GCC, and recompile something (like zip/bzip/gzip) and see if that made a difference. |
All times are GMT -5. The time now is 12:47 PM. |