01:35 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [rpardini](<https://github.com/rpardini>) - [Link](<https://github.com/armbian/build/commit/0c7596bd5a92c303abb4ccfbea5de1cefc9bddea>): *export-logs: use multiple paste servers in order until one works*

05:14 <DC-IRC> [Discord] <Werner> I personally have a kvm instance with 16G of memory to build stuff but I also build images on an Orangepi5 with 8G of memory.

07:55 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [chainsx](<https://github.com/chainsx>) - [Link](<https://github.com/armbian/build/commit/b86cf1adbb5dc8263b481d06f5bffa7256075727>): *sun55iw3-syterkit: update kernel config (#7855)*

08:56 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [igorpecovnik](<https://github.com/igorpecovnik>) - [Link](<https://github.com/armbian/build/commit/bc69b1d3eb2f74ff7c30bf5e96db285431ed54a3>): *sakurapi-rk3308b: bump dts and vop patches to kernel 6.13*

08:56 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [igorpecovnik](<https://github.com/igorpecovnik>) - [Link](<https://github.com/armbian/build/commit/ecfde0c3d7ee27ec49692ab34a0737ca6a2e088b>): *sunxi-6.6: Add i2c3_pa, all pwm pins*

10:17 <DC-IRC> [Discord] <mecoblock> https://www.apple.com/de/shop/product/FMFJ3D/A/refurbished-mac-mini-apple-m2-chip-mit-8%E2%80%91core-cpu-und-10%E2%80%91core-gpu

10:17 <DC-IRC> [Discord] <mecoblock> sold out but when it’s back up again it might be an option for you. I think it can run asahi too

10:17 <DC-IRC> [Discord] <Werner> Never had an apple device and probably never will

11:13 <DC-IRC> [Discord] <gambl0r> shit not even an ipod?

11:45 <DC-IRC> [Discord] <nexusguy59> Good Morning Folks! ☕

12:26 <DC-IRC> [Discord] <Werner> Nope

13:04 <DC-IRC> [Discord] <menu> this is £579 here 😭

13:05 <DC-IRC> [Discord] <mecoblock> 369€ from Apple directly was such a good deal

13:05 <DC-IRC> [Discord] <menu> the M4-10-10 is actually *cheaper*

13:06 <DC-IRC> [Discord] <menu> i live in a nonsense country

13:06 <DC-IRC> [Discord] <menu> https://cdn.discordapp.com/attachments/857944478269702214/1343207381173014581/image0.jpg?ex=67bc6edb&is=67bb1d5b&hm=cc3a06ed95962f75cbab60b7714f5d8ace098f8d340b0f0d24c45843ffd3f291&

13:07 <DC-IRC> [Discord] <mecoblock> well that’s a no brainer if you were looking for a mac. Only downside is no asahi but VM's work great and that’s also how I run linux on mine

13:15 <DC-IRC> [Discord] <menu> asahi's future is uncertain now anyway, but i actually like macOS, it's my preferred desktop OS! unfortunately i don't have a spare £600 to upgrade, so the 2015 macbook pro will continue to overheat 😂

13:19 <DC-IRC> [Discord] <mecoblock> those 2015 tanks are a classic but also deserve to be sent to retirement

13:21 <DC-IRC> [Discord] <menu> i was hanging it out of the window the other night to update some dependencies, they build from source, all four cores at 100% for 24 hours got real damn hot

14:25 <DC-IRC> [Discord] <.grippy> @mecoblock Regardless of opinions on Apple, M series silicon is crazy fast and in the context of the Mac Mini specifically super good value, especially for dev work if you don't care about GPU. I'm sitting on my couch with my M3 Pro MBP and I just timed it - < 1 min to build a server image with Ubuntu base for example, 1:20 if you count generating the .img . My 7950x3D desktop ta <clipped message>

14:25 <DC-IRC> [Discord] <.grippy> kes significantly longer, haven't timed it but I don't even bother at this point.

15:03 <DC-IRC> [Discord] <johnthecoolingfan> > takes significantly longer, haven't timed it but I don't even bother at this point.

15:03 <DC-IRC> [Discord] <johnthecoolingfan> So no data to support your point?

15:04 <DC-IRC> [Discord] <johnthecoolingfan> Also to make it a fair test caching and reuse of already buidl components should be disallowed (or cache and reuse the same things on both stations)

15:04 <DC-IRC> [Discord] <johnthecoolingfan> Also to make it a fair test caching and reuse of already buidl components should be disallowed (or cache and reuse the same things on both stations), plus the target image configuration shoudl be the same

15:05 <DC-IRC> [Discord] <johnthecoolingfan> Easy to skew the times if you don't match the conditions

15:05 <DC-IRC> [Discord] <johnthecoolingfan> Easy to skew the results if you don't match the conditions

15:09 <DC-IRC> [Discord] <.grippy> Fair enough, I'll clear both docker instances and do a timed test on both a bit later today and report back

15:12 <DC-IRC> [Discord] <johnthecoolingfan> that would be great

15:13 <DC-IRC> [Discord] <menu> i would expect the Apple Silicon to perform significantly better, purely because it's not cross-compiling

15:14 <DC-IRC> [Discord] <johnthecoolingfan> build for risc-v on both?

15:15 <DC-IRC> [Discord] <menu> that would be a fair test imo

15:15 <DC-IRC> [Discord] <johnthecoolingfan> Is cross-compiling inherently slower than compiling for the host architecture?

15:17 <DC-IRC> [Discord] <johnthecoolingfan> I don't really understand how the target and host architectures matching boost compilation times...

15:18 <DC-IRC> [Discord] <johnthecoolingfan> Also can just run twice more tests, one build for x86, anotehr for arm

15:18 <DC-IRC> [Discord] <johnthecoolingfan> Also can just run twice more tests, one build for x86, another for arm

15:19 <DC-IRC> [Discord] <johnthecoolingfan> the more data teh better

15:19 <DC-IRC> [Discord] <johnthecoolingfan> the more data the better

15:21 <DC-IRC> [Discord] <johnthecoolingfan> Because now I want to know what it would be like, will x86 compile for x86 faster than for arm and arm compile faster for arm than for x86 or would it be that x86/arm target compiles faster than the other arch

15:21 <DC-IRC> [Discord] <menu> typically yes, but it would be good to benchmark a range of hardware

15:22 <DC-IRC> [Discord] <johnthecoolingfan> Because now I want to know what it would be like, will x86 compile for x86 faster than for arm and arm compile faster for arm than for x86 or would it be that x86/arm target compiles faster than the other arch, and risc-v as always cross-compilation (even if the target arch is relatively obscure and not a common target) as a control

15:22 <DC-IRC> [Discord] <johnthecoolingfan> Because now I want to know what it would be like, will x86 compile for x86 faster than for arm and arm compile faster for arm than for x86 or would it be that x86/arm target compiles faster than the other arch, and risc-v as always cross-compilation (even if the target arch is relatively obscure and not a common target) as a control group or smth

15:39 <DC-IRC> [Discord] <.grippy> Time to Image Generated :

15:39 <DC-IRC> [Discord] <.grippy> - Mac : 1m 31s

15:39 <DC-IRC> [Discord] <.grippy> - PC: 4m 31s

15:39 <DC-IRC> [Discord] <.grippy> Build Test -

15:39 <DC-IRC> [Discord] <.grippy> 
Target Board - PocketBeagle 2 (TI AM62)

15:39 <DC-IRC> [Discord] <.grippy> Image - Ubuntu Noble 24.04

15:39 <DC-IRC> [Discord] <.grippy> Image Type - Image with console interface (server) - Standard

15:39 <DC-IRC> [Discord] <.grippy> Machine 1 - MacBook Pro 14” - M3 Max - 36GB RAM - 1TB Internal SSD - MacOS Sequoia 15.3.1

15:39 <DC-IRC> [Discord] <.grippy> Machine 2 - Custom PC - AMD 7950x3D, 64GB DDR5 6000Mhz CL30, 4TB NVMe Gen 4 SSD - Windows 11 Pro 24H2 - WSL2 Ubuntu 22.04.5 LTS

15:39 <DC-IRC> [Discord] <.grippy> Build env - Docker Desktop 4.38.0 (181591) - all volumes/caches/etc cleared

15:39 <DC-IRC> [Discord] <.grippy> 2.5Gb/s Wired Ethernet - no local network caching

15:40 <DC-IRC> [Discord] <.grippy> So yeah, significantly slower cross compiling, also let's not talk power consumption lol

15:40 <DC-IRC> [Discord] <.grippy> This was the better of 2 runs for the PC, first one came in over 5 min

15:41 <DC-IRC> [Discord] <.grippy> @johnthecoolingfan I'll try an x86 test too sure

15:41 <DC-IRC> [Discord] <johnthecoolingfan> awesome, thank you

15:50 <DC-IRC> [Discord] <johnthecoolingfan> was about to point out that docker on windows is a VM and a such has more overhead, but from the sources I found macs with apple silicon also use a virtual machine.

15:51 <DC-IRC> [Discord] <.grippy> Cleared all caches/volumes etc again.

15:51 <DC-IRC> [Discord] <.grippy> Uefi-x86 -> edge -> noble 24.04 -> Img with console interface standard

15:51 <DC-IRC> [Discord] <.grippy> Mac - 5 min 59s

15:51 <DC-IRC> [Discord] <.grippy> PC - 1m 55s

15:52 <DC-IRC> [Discord] <johnthecoolingfan> now that's interesting.

15:52 <DC-IRC> [Discord] <.grippy> Ha so exactly reversed, that's funny

15:52 <DC-IRC> [Discord] <johnthecoolingfan> now that's exciting

15:53 <DC-IRC> [Discord] <.grippy> Mind you the mac is sipping power and the PC is pulling like 400W lol. But interesting that it's exactly the opposite

16:01 <DC-IRC> [Discord] <johnthecoolingfan> So for compiling for arm M3 mac is great

16:01 <DC-IRC> [Discord] <johnthecoolingfan> And PC is great for x86 compilation

16:03 <DC-IRC> [Discord] <johnthecoolingfan> And if it was taking the PC to compile x86 the same amount of time as it was taking mac to compile for arm (if the x86 took less time overall), mac would approximately take 284-ish seconds, or 4m44s

16:03 <DC-IRC> [Discord] <johnthecoolingfan> And if it was taking the PC to compile x86 the same amount of time as it was taking mac to compile for arm (if the x86 took less time overall), mac would approximately take 284-ish seconds, or 4m44s for x86 corss-compile

16:03 <DC-IRC> [Discord] <johnthecoolingfan> And if it was taking the PC to compile x86 the same amount of time as it was taking mac to compile for arm (if the x86 took less time overall), mac would approximately take 284-ish seconds, or 4m44s for x86 cross-compile

16:07 <DC-IRC> [Discord] <johnthecoolingfan> If the results were not dependent on whether it's a cross-compile, the conslusion would be more straightforward, but with this the conslusion is that it's better to match target arch to host arch, and since armbian has more arm boards, M3 mac is generally better for this purpose

16:07 <DC-IRC> [Discord] <johnthecoolingfan> I suspect a risc-v compilation test would show mac being slower

16:08 <DC-IRC> [Discord] <menu> they do, i'd be curious to see how it performed using UTM with native Apple Virtualization though

16:09 <DC-IRC> [Discord] <menu> they do, i'd be curious to see how it performed using UTM with native Apple Virtualization though - with prefer_docker=no, because i find docker inherently slow

16:10 <DC-IRC> [Discord] <johnthecoolingfan> cross-compiling without docker is a nightmare. I remember how nightmarish it was on arch to install teh toolchain

16:10 <DC-IRC> [Discord] <johnthecoolingfan> cross-compiling without docker is a nightmare. I remember how nightmarish it was on arch to install the toolchain

16:10 <DC-IRC> [Discord] <johnthecoolingfan> Gentoo seems to have better cross-compiling support though, `cross` is a great tool

16:10 <DC-IRC> [Discord] <johnthecoolingfan> Gentoo seems to have better cross-compiling support though, `crossdev` is a great tool

16:11 <DC-IRC> [Discord] <menu> does the build script not handle this? i only build for ARM at present

16:13 <DC-IRC> [Discord] <johnthecoolingfan> Never tried without docker. No idea how it handles it, but I expect one of two things:

16:13 <DC-IRC> [Discord] <johnthecoolingfan> 1. Install the toolchain via package manager (on ubuntu/debian, can't handle EVERY package manager in existence in the long run)

16:13 <DC-IRC> [Discord] <johnthecoolingfan> 2. Fail if teh toolchain is missing on host

16:13 <DC-IRC> [Discord] <johnthecoolingfan> Never tried without docker. No idea how it handles it, but I expect one of two things:

16:13 <DC-IRC> [Discord] <johnthecoolingfan> 1. Install the toolchain to host via package manager (on ubuntu/debian, can't handle EVERY package manager in existence in the long run)

16:13 <DC-IRC> [Discord] <johnthecoolingfan> 2. Fail if the toolchain is missing on host

16:15 <DC-IRC> [Discord] <menu> 1 is what i'd *expect* to happen, i haven't actually read the build script but i'd have thought it grabs deps the same as the docker image

16:16 <DC-IRC> [Discord] <johnthecoolingfan> Never tried without docker. No idea how it handles it, but I expect one of two things:

16:16 <DC-IRC> [Discord] <johnthecoolingfan> 1. Install the toolchain to host via package manager (on ubuntu/debian, can't handle EVERY package manager in existence in the long run, otherwise go to 2)

16:16 <DC-IRC> [Discord] <johnthecoolingfan> 2. Fail if the toolchain is missing on host

16:16 <DC-IRC> [Discord] <johnthecoolingfan> Also option 3: skip the package manager and download/install the toolchain into a temp dir

16:16 <DC-IRC> [Discord] <johnthecoolingfan> Also option 3: skip the package manager and download/install the toolchain into a temp/cache dir

16:17 <DC-IRC> [Discord] <johnthecoolingfan> And docker is a glorified temp dir

16:55 <DC-IRC> [Discord] <.grippy> I’ll try a riscv crosscompile on both tonight but not gonna do full tool chain haha. Part of the whole appeal of docker is how easy it is to un-f it

16:57 <DC-IRC> [Discord] <.grippy> But quite surprising results yeah

18:21 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [igorpecovnik](<https://github.com/igorpecovnik>) - [Link](<https://github.com/armbian/build/commit/11430de9a06e4ab542c70151343948f53c35f65f>): *Bump Vim1-2 u-boot binaries (stopgap)*

18:23 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [igorpecovnik](<https://github.com/igorpecovnik>) - [Link](<https://github.com/armbian/build/commit/a0bc90c360cc74064030776b477bcde71c13c02e>): *BananaPi BPI-F3: `Alphabetize DTS`*

18:24 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [igorpecovnik](<https://github.com/igorpecovnik>) - [Link](<https://github.com/armbian/build/commit/745955110be1a57eb6ada15062dc4c67b5b15e29>): *rockchip64:loaderimage: fix rk3308 uboot loader offset*

18:26 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [igorpecovnik](<https://github.com/igorpecovnik>) - [Link](<https://github.com/armbian/build/commit/d19444bf8a252aa5c2e1a8b5dd029da82d812092>): *rockchip64: fix wrong GPIO direction in DP virtual extcon driver*

18:27 <DC-IRC> [Discord] <Github> ✅ **Merged** into [armbian/build](<https://github.com/armbian/build>) by [igorpecovnik](<https://github.com/igorpecovnik>) - [Link](<https://github.com/armbian/build/commit/86691d17e4d9c607e5ed661a9209be36233cb6a4>): *rockchip/64: fix again periodic dma and audio stuttering*

19:25 <DC-IRC> [Discord] <rpardini> for kernel/uboot builds: no, cross compiling is generally just as fast.

19:25 <DC-IRC> [Discord] <rpardini> for rootfs/image builds: yes, as native does not need qemu. native should be 3-4x faster or more.