Chromium build time 36 hours??

Message

e8root · Post by **e8root** » Mon May 06, 2024 7:29 am

Chromium 124.0.6367, no LTO and no PGO builds 1h 24m on 13900KF with HT disabled.

Judging by the times like 23 hours on something like Core i7 7700 and difference in benchmarks like Cibebench R23 I would say there is not enough memory.
Imho better to use zram than tmpfs because source files are very compressible. Though if it makes performance sense on 32GB system with 4 core CPU versus just using disk directly I am not sure. Personally I use zram to just minimize writes on my NVMe drive.

lyallp · Post by **lyallp** » Mon May 06, 2024 9:16 am

1 day, 7 hours and counting using all 8 cores.

35634 source files compiled out of 59628

Sheesh, I am only just over half way!

logrusx · Post by **logrusx** » Mon May 06, 2024 9:28 am

lyallp wrote:Still building.... sigh.... now I remember why I dropped chromium.
Still, my firefox is broken, despite my best efforts, so other than konqueror...
Code: Select all
# uname -a
Linux Lyalls-PC 6.6.21-gentoo #14 SMP PREEMPT_DYNAMIC Fri May  3 19:07:10 ACST 2024 x86_64 Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz GenuineIntel GNU/Linux
root@Lyalls-PC package.use
# genlop -cte

 Currently merging 3 out of 3

 * www-client/chromium-124.0.6367.60 

       current merge time: 23 hours and 54 seconds.
       ETA: unknown.
root@Lyalls-PC package.use
# 
32G RAM, of which 24G is devoted to /tmp (tmpfs) and portage configured to use /tmp, not /var/portage/tmp.
I figure RAM disk would be faster than SSD...

lyallp wrote:1 day, 7 hours and counting using all 8 cores.

35634 source files compiled out of 59628

Sheesh, I am only just over half way!

What you're doing does not make sense. You're running 8 cores (or rather 8 threads as you CPU has 4 cores) with 8GB of RAM. And that makes no sense even more given that you have 32GB RAM but wasting 24 of them on a tmpfs. Do you realize what insane amount of swapping is taking place?

Best Regards,
Georgi

Post by **NeddySeagoon** » Mon May 06, 2024 9:36 am

Chromium on an 8G Pi4 takes about 3d 12h.
On an 8G Pi5, its just over 24h.

lyallp · Post by **lyallp** » Mon May 06, 2024 9:42 am

Code: Select all

root@Lyalls-PC temp
# free -m
               total        used        free      shared  buff/cache   available
Mem:           32036        7631        1272       17030       23132        6906
Swap:          32505        1669       30836
root@Lyalls-PC temp
# htop
    0[|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||100.0%]   4[|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||100.0%]
    1[|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||100.0%]   5[|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||100.0%]
    2[|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||100.0%]   6[|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||100.0%]
    3[||||||||||||||                                                            16.7%]   7[||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||        83.3%]
  Mem[||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||24.2G/31.3G] Tasks: 159, 444 thr, 173 kthr; 8 running
  Swp[|||||                                                               1.59G/31.7G] Load average: 6.16 6.35 6.34 
                                                                                       Uptime: 1 day, 08:59:56

  [Main] [I/O]
  PID USER       PRI  NI  VIRT   RES   SHR S  CPU%▽MEM%   TIME+  Command
 4284 portage     20   0  675M  453M  148M R  94.1  1.4  1:53.38 x86_64-pc-linux-gnu-clang++ -MD -MF obj/components/feedback/feedback/feedback_uploader.o.d -DUSE_UDEV -DUSE_
 4609 portage     20   0  467M  243M  141M R  78.4  0.8  0:30.49 x86_64-pc-linux-gnu-clang++ -MD -MF obj/components/history/core/browser/browser/expire_history_backend.o.d -
 4652 portage     20   0  452M  226M  140M R  78.4  0.7  0:24.02 x86_64-pc-linux-gnu-clang++ -MD -MF obj/components/history/core/browser/browser/features.o.d -DUSE_UDEV -DUS
 4676 portage     20   0  426M  200M  138M R  78.4  0.6  0:15.85 x86_64-pc-linux-gnu-clang++ -MD -MF obj/components/history/core/browser/browser/history_backend.o.d -DUSE_UD
 4677 portage     20   0  419M  193M  138M R  78.4  0.6  0:15.15 x86_64-pc-linux-gnu-clang++ -MD -MF obj/components/history/core/browser/browser/history_constants.o.d -DUSE_
 4720 portage     20   0  378M  151M  133M R  78.4  0.5  0:02.05 x86_64-pc-linux-gnu-clang++ -MD -MF obj/components/history/core/browser/browser/history_database.o.d -DUSE_U
 4749 root        20   0  9348  5504  3456 R  78.4  0.0  0:00.10 htop -n 1
    1 root        20   0  2384  1504  1504 S   0.0  0.0  0:01.00 init [3]

Not a lot, nearly used all the main memory but not all, a bit of swap, only 1 G.
htop shows swap is hardly being used at all, memory is only using 23G of 32G

adcr · Post by **adcr** » Mon May 06, 2024 2:37 pm

lyallp wrote:

Code: Select all

 Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz GenuineIntel GNU/Linux
(...)
       current merge time: 23 hours and 54 seconds.

(...)
32G RAM(...)

That machine should finish within less than 11 hours if Turbo Boost is not disabled.

For reference, my i7-8650U, from a similar gen, half the TDP, finishes in 10 and a half hours with Turbo Boost on.

I normally disable turbo on this laptop, limiting the clock to 1.9Ghz to keep temps cool (<65°C in a world upgrade) and the fan barely audible. The compile time is still no more than 16-18 hours.

IMO no matter how little swap use there is, it'll still be a bottleneck. Just try filling up your memory with whatever and opening wikipedia in a web browser. There'll be no more than 1G of swap in use, yet the system will be unusable.

BTW I don't use LTO and my portage files are all on my (unencrypted, for now) M.2 SSD.

I had once a really weird bug which will likely not apply here. I ran out of space while compiling chromium. After freeing up plenty space I restarted the process, yet it would remain visibly slowed down and my SSD would overheat (from the normal ~20°C while emerging to 60°C+). The end resolution was to run fstrim, which otherwise would only have ran next week, on schedule. I'm guessing the poor SSD was trying to move the new files around on a tiny small space it was aware of being free. I've since moved to just mounting using the discard option.

logrusx · Post by **logrusx** » Mon May 06, 2024 5:55 pm

lyallp,

You make it look like it's useless to give you advice. After all, if you know better, what's the point of you posting here? Complain and rant about it? If you're not seeking help, you better stop it here. And if you are, you're disrespecting the people who try to help you.

Best Regards,
Georgi

eccerr0r · Post by **eccerr0r** » Mon May 06, 2024 9:36 pm

How long did that i5-4590 take to build rust or gcc? 36 hours for chromium seems a bit high though I've yet to try to build it.

Unfortunately my build time logs for things like qtwebengine is really contaminated as distcc helps considerably for it...

I was wondering whether I should get a E5-2697v2 or E5-2690v2 (E5-2687Wv2 can't be had cheaply...) to upgrade my i7-3820. After some weird chain of events, my i7-3820 has 64GiB RAM in it and the computer is not really taking advantage of it, not much different to my i7-2700k with 24GiB....

lyallp · Post by **lyallp** » Mon May 06, 2024 10:54 pm

Just an fyi: 2/3 through, 44 hours down, looks like 20 hours to go.

I am just agahst that it takes so long to build one program, when I can build my entire system with --emptytree in 24 hours (not including chromium, firefox and thunderbird). Nearly 50,000 source files, 14 million lines of code, including tests. How does someone wrap their minds around that sort of thing?

I can't reduce my tmpfs because chromium requires 20G temp, and writing to disk is slower than memory. It's been a while, but I don't think I can add more RAM to my motherboard.

So, bottom line, unless I upgrade, it's just something I will have to put up with.

eccerr0r · Post by **eccerr0r** » Mon May 06, 2024 11:08 pm

lyallp wrote:Just an fyi: 2/3 through, 44 hours down, looks like 20 hours to go.

It takes my Atom around 120 hours to do its yearly updates...so a long time is relative.

adcr wrote:IMO no matter how little swap use there is, it'll still be a bottleneck. Just try filling up your memory with whatever and opening wikipedia in a web browser. There'll be no more than 1G of swap in use, yet the system will be unusable.

It totally depends on how your using your RAM, can't say swap is always bad. If you're using RAM as a buffer (like tmpfs) and it swaps out, swap actually helps significantly. 1GB of swap is not a big deal.

However if you're using that ram for lto, then that may be a problem if it's forced to swap out.

Anything not actively being used in RAM could be swapped out and used for stuff that's actively being used, where actively means being used at the moment right then and there. That's what swap is for - to make a tradeoff of what does not need to be in RAM at the moment.

C5ace · Post by **C5ace** » Tue May 07, 2024 4:20 am

Installed chromium on a virgin Xfce4 desktop with 7500RPM rotating rust HDD. CPU and RAM usage monitored using mate-extra/mate-system-monitor-1.26.2. RAM used shown is within +-0.5GB.

Chromium and related build times and RAM used:
lscpu:

Code: Select all

Architecture:             x86_64
  CPU op-mode(s):         32-bit, 64-bit
  Address sizes:          43 bits physical, 48 bits virtual
  Byte Order:             Little Endian
CPU(s):                   32
  On-line CPU(s) list:    0-31
Vendor ID:                AuthenticAMD
  BIOS Vendor ID:         Advanced Micro Devices, Inc.
  Model name:             AMD Ryzen 9 3950X 16-Core Processor
    BIOS Model name:      AMD Ryzen 9 3950X 16-Core Processor             Unknow
                          n CPU @ 3.5GHz
    BIOS CPU family:      107
    CPU family:           23
    Model:                113
    Thread(s) per core:   2
    Core(s) per socket:   16
    Socket(s):            1
    Stepping:             0
    Frequency boost:      enabled
    CPU(s) scaling MHz:   48%
    CPU max MHz:          4761.2300
    CPU min MHz:          2200.0000
    BogoMIPS:             7002.61
    Flags:                fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge m
                          ca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall
                           nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep
                          _good nopl nonstop_tsc cpuid extd_apicid aperfmperf ra
                          pl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 
                          movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_leg
                          acy svm extapic cr8_legacy abm sse4a misalignsse 3dnow
                          prefetch osvw ibs skinit wdt tce topoext perfctr_core 
                          perfctr_nb bpext perfctr_llc mwaitx cpb cat_l3 cdp_l3 
                          hw_pstate ssbd mba ibpb stibp vmmcall fsgsbase bmi1 av
                          x2 smep bmi2 cqm rdt_a rdseed adx smap clflushopt clwb
                           sha_ni xsaveopt xsavec xgetbv1 cqm_llc cqm_occup_llc 
                          cqm_mbm_total cqm_mbm_local clzero irperf xsaveerptr r
                          dpru wbnoinvd arat npt lbrv svm_lock nrip_save tsc_sca
                          le vmcb_clean flushbyasid decodeassists pausefilter pf
                          threshold avic v_vmsave_vmload vgif v_spec_ctrl umip r
                          dpid overflow_recov succor smca sev sev_es
Virtualization features:  
  Virtualization:         AMD-V
Caches (sum of all):      
  L1d:                    512 KiB (16 instances)
  L1i:                    512 KiB (16 instances)
  L2:                     8 MiB (16 instances)
  L3:                     64 MiB (4 instances)
NUMA:                     
  NUMA node(s):           1
  NUMA node0 CPU(s):      0-31
Vulnerabilities:          
  Gather data sampling:   Not affected
  Itlb multihit:          Not affected
  L1tf:                   Not affected
  Mds:                    Not affected
  Meltdown:               Not affected
  Mmio stale data:        Not affected
  Reg file data sampling: Not affected
  Retbleed:               Mitigation; untrained return thunk; SMT enabled with STIBP protection
  Spec rstack overflow:   Mitigation; Safe RET
  Spec store bypass:      Mitigation; Speculative Store Bypass disabled via prctl
  Spectre v1:             Mitigation; usercopy/swapgs barriers and __user pointer sanitization
  Spectre v2:             Mitigation; Retpolines; IBPB conditional; STIBP always -on; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
  Srbds:                  Not affected
  Tsx async abort:        Not affected

make.conf:

Code: Select all

COMMON_FLAGS="-O2 -march=znver2 -mtune=znver2 -pipe"
CFLAGS="${COMMON_FLAGS}"
CXXFLAGS="${COMMON_FLAGS}"
FCFLAGS="${COMMON_FLAGS}"
FFLAGS="${COMMON_FLAGS}"

MAKEOPTS="-j32"
FEATURES="-preserve-libs"
ACCEPT_KEYWORDS="amd64"
ACCEPT_LICENSE="*"
LINGUAS="en"
INPUT_DEVISES="libinput"
VIDEO_CARDS="amdgpu radeonsi"
SANE_BACKENDS="hp"

DESKTOPUSE="X a52 aac acl acpi alsa amd64 bluetooth branding bzip2 cairo cdda cdr cli crypt cups dbus dri dts dvd dvdr elogind encode exif flac fortran gdbm gif gpm gtk gui iconv icu ipv6 jpeg lcms libnotify libtirpc mad mng mp3 mp4 mpeg multilib ncurses nls nptl ogg opengl openmp pam pango pcre pdf png policykit ppds qt5 readline sdl seccomp sound spell split-usr ssl startup-notification svg test-rust tiff truetype udev udisks unicode upower usb vorbis vulkan wxwidgets x264 xattr xcb xft xml xv xvid zlib " 

USE="${DESKTOPUSE}-kde -gnome -qt5 -ipv6 -bluetooth -systemd -multilib lm_sensors alsa"

PORTDIR="/var/db/repos/gentoo"
DISTDIR="/var/cache/distfiles"
PKGDIR="/var/cache/binpkgs"

GRUB_PLATFORMS="pc"

emerge -av chromium

Code: Select all

These are the packages that would be merged, in order:

Calculating dependencies... done!
Dependency resolution took 11.17 s (backtrack: 0/20).

[ebuild  N     ] acct-group/pipewire-0-r2::gentoo  0 KiB
[ebuild   R    ] media-libs/libpng-1.6.43:0/16::gentoo  USE="-apng* -static-libs -test" CPU_FLAGS_X86="sse" 0 KiB
[ebuild  N     ] media-libs/openh264-2.4.1:0/7::gentoo  USE="plugin -test -utils" CPU_FLAGS_X86="avx2" 0 KiB
[ebuild  N     ] net-libs/nodejs-20.12.1:0/20::gentoo  USE="icu inspector npm snapshot ssl system-icu system-ssl -corepack -debug -doc (-lto) -pax-kernel -test" CPU_FLAGS_X86="sse2" 0 KiB
[ebuild  N     ] dev-util/wayland-scanner-1.22.0::gentoo  225 KiB
[ebuild  N     ] dev-libs/wayland-1.22.0::gentoo  USE="-doc -test" 0 KiB
[ebuild  N     ] dev-libs/wayland-protocols-1.33::gentoo  USE="-test" 90 KiB
[ebuild  N     ] media-libs/libva-2.21.0:0/2::gentoo  USE="X wayland" 0 KiB
[ebuild  N     ] dev-build/gn-0.2157::gentoo  731 KiB
[ebuild   R    ] media-libs/mesa-24.0.4::gentoo  USE="X gles2 llvm (opengl) proprietary-codecs vulkan wayland* zstd -d3d9 -debug -gles1 -lm-sensors -opencl -osmesa (-selinux) -test -unwind -vaapi -valgrind -vdpau -vulkan-overlay -xa (-zink)" CPU_FLAGS_X86="sse2" LLVM_SLOT="17 -15 -16" VIDEO_CARDS="radeonsi -d3d12 (-freedreno) -intel -lavapipe (-lima) -nouveau (-panfrost) -r300 -r600 -radeon (-v3d) (-vc4) -virgl (-vivante) -vmware" 0 KiB
[ebuild   R    ] x11-libs/gtk+-3.24.41:3::gentoo  USE="X cups examples gtk-doc introspection wayland* (-aqua) -broadway -cloudproviders -colord -sysprof -test -vim-syntax -xinerama" 0 KiB
[ebuild   R    ] dev-lang/rust-1.77.1:stable/1.77::gentoo  USE="profiler* rustfmt (-big-endian) -clippy -debug -dist -doc (-llvm-libunwind) -lto (-miri) (-nightly) (-parallel-compiler) -rust-analyzer -rust-src (-system-bootstrap) (-system-llvm) -test -verify-sig -wasm" CPU_FLAGS_X86="sse2" LLVM_TARGETS="(X86) -AArch64 -AMDGPU -ARC -ARM -AVR -BPF -CSKY -DirectX -Hexagon -Lanai -LoongArch -M68k -MSP430 -Mips -NVPTX -PowerPC -RISCV -SPIRV -Sparc -SystemZ -VE -WebAssembly -XCore -Xtensa" 0 KiB
[ebuild   R    ] virtual/rust-1.77.1:0/llvm-17::gentoo  USE="profiler* -rustfmt" 0 KiB
[ebuild  N     ] media-video/pipewire-1.0.4:0/0.4::gentoo  USE="X dbus readline ssl -bluetooth -doc -echo-cancel -extra -ffmpeg -flatpak -gsettings -gstreamer -ieee1394 -jack-client -jack-sdk -liblc3 -lv2 -man -modemmanager -pipewire-alsa -roc (-selinux) -sound-server (-system-service) -systemd -test -v4l -zeroconf" 1,692 KiB
[ebuild  N     ] media-video/wireplumber-0.4.17-r1:0/0.4::gentoo  USE="elogind (-system-service) -systemd -test" LUA_SINGLE_TARGET="lua5-4 -lua5-3" 323 KiB
[ebuild  N     ] sys-devel/lld-17.0.6:17::gentoo  USE="zstd -debug -test -verify-sig" 0 KiB
[ebuild  N     ] sys-devel/lld-toolchain-symlinks-17:17::gentoo  USE="native-symlinks -multilib-symlinks" 0 KiB
[ebuild  N     ] www-client/chromium-124.0.6367.60:0/stable::gentoo  USE="X cups hangouts official proprietary-codecs screencast system-harfbuzz system-icu system-png system-toolchain system-zstd vaapi wayland widevine -bindist -custom-cflags -debug -ffmpeg-chromium -gtk4 (-headless) -kerberos (-libcxx) (-lto) -pax-kernel (-pgo) -pulseaudio -qt5 -qt6 (-selinux)" L10N="af am ar bg bn ca cs da de el en-GB es es-419 et fa fi fil fr gu he hi hr hu id it ja kn ko lt lv ml mr ms nb nl pl pt-BR pt-PT ro ru sk sl sr sv sw ta te th tr uk ur vi zh-CN zh-TW" 3,352,285 KiB

Total: 18 packages (13 new, 5 reinstalls), Size of downloads: 3,355,343 KiB

The following USE changes are necessary to proceed:
 (see "package.use" in the portage(5) man page for more details)
# required by www-client/chromium-124.0.6367.60::gentoo[system-png]
# required by chromium (argument)
>=media-libs/libpng-1.6.43 -apng
# required by www-client/chromium-124.0.6367.60::gentoo[vaapi]
# required by chromium (argument)
>=media-libs/libva-2.21.0 wayland
# required by www-client/chromium-124.0.6367.60::gentoo
# required by chromium (argument)
>=media-libs/mesa-24.0.4 wayland
# required by www-client/chromium-124.0.6367.60::gentoo[-gtk4]
# required by chromium (argument)
>=x11-libs/gtk+-3.24.41 wayland
# required by www-client/chromium-124.0.6367.60::gentoo
# required by chromium (argument)
>=net-libs/nodejs-20.12.1 inspector
# required by www-client/chromium-124.0.6367.60::gentoo[system-toolchain]
# required by chromium (argument)
>=virtual/rust-1.77.1 profiler
# required by virtual/rust-1.77.1::gentoo
# required by www-client/chromium-124.0.6367.60::gentoo[system-toolchain]
# required by chromium (argument)
>=dev-lang/rust-1.77.1 profiler

Would you like to add these changes to your config files? [Yes/No]

Added USE flags and ran
emerge www-client/chromium

Result:

Code: Select all

qlop
2024-05-06T15:47:47 >>> acct-group/pipewire: 9s
2024-05-06T15:47:56 >>> media-libs/libpng: 13s
2024-05-06T15:48:09 >>> media-libs/openh264: 18s
2024-05-06T15:48:27 >>> sys-devel/clang-common: 32s
2024-05-06T15:48:59 >>> sys-devel/clang: 10′03″ (Max RAM used: 29GB)
2024-05-06T15:59:02 >>> sys-devel/clang-toolchain-symlinks: 12s
2024-05-06T15:59:14 >>> net-libs/nodejs: 7′32″ (Max RAM used: 21GB)
2024-05-06T16:06:46 >>> dev-util/wayland-scanner: 8s
2024-05-06T16:06:54 >>> dev-libs/wayland: 9s
2024-05-06T16:07:03 >>> dev-libs/wayland-protocols: 7s
2024-05-06T16:07:10 >>> media-libs/libva: 10s
2024-05-06T16:07:20 >>> dev-build/gn: 22s
2024-05-06T16:07:42 >>> media-libs/mesa: 1′05″
2024-05-06T16:08:47 >>> x11-libs/gtk+: 3′09″
2024-05-06T16:11:56 >>> sys-libs/compiler-rt: 26s
2024-05-06T16:12:22 >>> sys-libs/compiler-rt-sanitizers: 54s
2024-05-06T16:13:16 >>> sys-libs/libomp: 24s
2024-05-06T16:13:40 >>> sys-devel/clang-runtime: 5s
2024-05-06T16:13:45 >>> dev-lang/rust: 17′11″ (Max RAM used 23GB)
2024-05-06T16:30:56 >>> virtual/rust: 5s
2024-05-06T16:31:01 >>> media-video/pipewire: 27s
2024-05-06T16:31:28 >>> media-video/wireplumber: 11s
2024-05-06T16:31:39 >>> sys-devel/lld: 44s
2024-05-06T16:32:23 >>> sys-devel/lld-toolchain-symlinks: 5s
2024-05-06T16:32:28 >>> www-client/chromium: 2:08:54 (Max RAM used: 29GB)

logrusx · Post by **logrusx** » Tue May 07, 2024 5:12 am

lyallp wrote:Just an fyi: 2/3 through, 44 hours down, looks like 20 hours to go.

I am just agahst that it takes so long to build one program...

The stop building chromium and stop complaining about it. It's that simple.

lyallp wrote:I can't reduce my tmpfs because chromium requires 20G temp, and writing to disk is slower than memory.

You're under the influence of some really unfounded beliefs and you don't want to let go of them. That's the root cause of your problems.

Best Regards,
Georgi

lyallp · Post by **lyallp** » Tue May 07, 2024 5:56 am

logrusx wrote:
lyallp wrote:Just an fyi: 2/3 through, 44 hours down, looks like 20 hours to go.

I am just agahst that it takes so long to build one program...
The stop building chromium and stop complaining about it. It's that simple.

lyallp wrote:I can't reduce my tmpfs because chromium requires 20G temp, and writing to disk is slower than memory.
You're under the influence of some really unfounded beliefs and you don't want to let go of them. That's the root cause of your problems.

Best Regards,
Georgi

And how about you butt out, I have gotten over how long it takes and simply reporting, to those that are interested, how long it takes.
If you are not interested, don't read this thread.

e8root · Post by **e8root** » Tue May 07, 2024 9:01 am

lyallp wrote:Just an fyi: 2/3 through, 44 hours down, looks like 20 hours to go.

I am just agahst that it takes so long to build one program, when I can build my entire system with --emptytree in 24 hours (not including chromium, firefox and thunderbird). Nearly 50,000 source files, 14 million lines of code, including tests. How does someone wrap their minds around that sort of thing?

I can't reduce my tmpfs because chromium requires 20G temp, and writing to disk is slower than memory. It's been a while, but I don't think I can add more RAM to my motherboard.

So, bottom line, unless I upgrade, it's just something I will have to put up with.

Your reported compile times feel way too long for your system specs.
It should comfortably build over half a day or something like it e.g. during the night.

Actually I have in one of my computers Xeon E3-1285L v4 machine with 16GB memory and 2TB A-DATA SU800 SSD and this CPU scores very similarly to Core i7 7700 (https://technical.city/en/cpu/Core-i7-7 ... 3-1285L-v4) so I can check how long Chromium builds on it. I first need to install Gentoo on it with all the necessary tooling to build Chromium so might take a while (unless I use 13900KF to rip through this stuff in a matter of minutes

) but if I am right and I manage to configure everything to launch emerge chromium on this PC before night then Chromium should be up and running by tomorrow

lyallp · Post by **lyallp** » Tue May 07, 2024 9:31 am

It was only a year or so ago I gave up on chromium cos it took 4-6 hours, which I thought was a long time then, back when I had spinning iron disks, not SSD or even a TMPFS /tmp filesystem.

Almost the same compile times as OpenOffice.

Lately, I notice Thunderbird, Firefox and now Chromium compile times take an absolute age compared to what they used to take.

Chromium, taking the cake at 2 days 7 hours and counting

How times change...

lyallp · Post by **lyallp** » Wed May 08, 2024 1:08 am

e8root, It's awsome you are going to all that trouble.

2 days, 23 hours and counting, still a long way to go. (over 10,000 files to go, according to the build log).

It appears my system is CPU bound, as htop shows all my CPU's generally in excess of 80% (portage runs at low priority so my fluxbox desktop is minimally affected) with a minimum of 6 clang++ processes running at a time. A typical invokation of the clang++ compiler is lasting about 1 minute 30 seconeds of CPU, which, just seems ridiculous.

It is quite possible, because I have built my system with debug enabled, top to bottom, that may be the cause. I was under the impression I would suffer a slight degredation of performance for doing this but I was attempting to generate meaninful core files of Firefox, all the way down to glibc.

The only reason I am building chromium is because Firefox is broken and I wanted an alternative to Konqueror, whilst I tried to figure out what was wrong with Firefox.

When Chromium finishes compiling (probably next week), I will re-build my entire system, optimised and see if that makes a difference.

I mistakenly (apologies above) thought I had 8 cores cos Linux boots up with 8 penguins and Htop shows 8, but, apparently, I only have 4 cores - it's been a while since I did a CPU/Mobo refresh, the current one, up until now, seemed to be quite adequate. It's a shame that the solution always seems to be 'throw more iron at it' to solve performance problems...

Iguana8649 · Post by **Iguana8649** » Wed May 08, 2024 11:18 am

Jesus H Christ!

Thinking of returning to Gentoo and those build times make me happy we now have binary packages! (fear not, I will still compile the whole system including Chromium, for teh rice, but first package install will be binary followed by a `emerge -e @world` during the first possible weekend)

lyallp · Post by **lyallp** » Wed May 08, 2024 11:21 am

Don't get too excited.

I plan on re-building my entire system without debug and optimisation turned on.

It's entirely possible this may have a positive effect on my build times!

In the meantime, I am investigating replacing my 5 year old system with something a touch more beefy....

3 days, 9 hours so far.... 12,000 odd source files to build yet....

Post by Hu » Wed May 08, 2024 3:21 pm

Depending on exactly what you mean by "debug", yes, that can have a massive effect on build performance. Template-heavy C++ programs are infamous for generating massive debug symbols when you enable those, and maintaining all that symbol information is a cost for the compiler. If you don't need debug symbols, you are much better off not generating them.

logrusx · Post by **logrusx** » Wed May 08, 2024 3:34 pm

Iguana8649 wrote:Jesus H Christ!

Thinking of returning to Gentoo and those build times make me happy we now have binary packages! (fear not, I will still compile the whole system including Chromium, for teh rice, but first package install will be binary followed by a `emerge -e @world` during the first possible weekend)

Hello and welcome back to Gentoo! Indeed the binhost has been very useful.

However this is not the place to have a chat. Please use Gentoo Chat or #gentoo-chat for that

Please read the forum guidelines.

Best Regards,
Georgi

e8root · Post by **e8root** » Thu May 09, 2024 5:27 am

Managed to build Chromium on my Broadwell Xeon E3-1285L v4
I botched first time by not changing number of threads in make opts from 24 to 8 and it ran out of memory (no swap is the best guarantee no swap was used). I also had to run Gentoo's LiveUSB because it was late and I didn't have time installing kernel.

With correct setting it took ~608 minutes so 10 hours and ~8 minutes.
Given this Xeon has roughly comparable performance but still lower than Core i7 7700 it would seem there is a configuration issue

Code: Select all

www-client/chromium-124.0.6367.60::gentoo was built with the following:
USE="X cups custom-cflags hangouts official proprietary-codecs pulseaudio qt5 screencast system-harfbuzz system-icu system-toolchain system-zstd vaapi wayland widevine -bindist -debug -ffmpeg-chromium -gtk4 (-headless) -kerberos (-libcxx) (-lto) -pax-kernel (-pgo) -qt6 (-selinux) -system-png" ABI_X86="(64)"

Code: Select all

COMMON_FLAGS_TMP="-O2 -pipe ${COMP_CPU_CFLAGS} -mfpmath=both -fomit-frame-pointer -DNDEBUG

COMP_CPU_FLAGS is output from resolve-march-native

If I may suggest quick to build web browser: www-client/seamonkey
It is classic style Mozilla browser with much less bloat. So little on my 13900KF it builds in like 5 minutes.
YT works, JS with JIT work, WebAssembly and WebGL work - really it seems its full featured.
Only not sure about supporting multiple threads because it seems to run in single thread... but it also means it should use less memory.

Chromium, not sure if its due to settings or anything but fonts on it look wonky. Less sharp than FF. I have similar issue with Tidal-HiFi so they probably screwed something up. I haven't investigated the issue because I don't use Chromium.
Chrome seems to fare better and at least supports google profile.
Either I don't use because I use FireFox 99% of the time anyways.

eccerr0r · Post by **eccerr0r** » Fri May 10, 2024 8:43 am

Trying to build Chromium ...

using Sandybridge 4C8T, Haswell 2C4T, Skylake 2C4T, and a Core2Quad, all ancient hardware. 12 physical cores/20 virtual cores. The Sandybridge is the build host, rest are distcc helpers. Not using tmpfs, SSD on build host.

I think the Sandybridge is still probably the fastest of the four but don't think the Core2Quad is that far slower than any of them.

It took 6 hours. I noticed it didn't use all the cores/threads available at all times and it looks like chromium is slightly rusty too (uses rust?) I wonder if just getting a E5v2 with 12 cores would be just as fast as all these machines combined when building Chromium...

So I would guess the Sandybridge should be able to finish in 18 hours or so without distcc help... alas it probably will always have help. Now if I had to build on my laptops away from home, indeed it will take over a day... :(

lyallp · Post by **lyallp** » Tue May 14, 2024 3:43 am

Gave up after 3 and a half DAYS.

Completely rebuilt my system from the ground up (--empty-tree @world @system) took 30 hours

Re-build chromium, took 9 hours! Huge improvement!

e8root · Post by **e8root** » Tue May 14, 2024 9:16 am

9h looks correct
My Broadwell system wasn't even running at full speed (it's an ITX system and I slightly lowered max multiplier) and its slightly slower CPU anyways. So 9h on 7700 vs 10h on my Xeon looks about what I would expect.

I also did compile time test for Seamonkey Browser.
It took 25 minutes to build on this Broadwell

25 min vs 600 minfor Chromium... YT 4K 60fps works like a champ on Seamonkey with zero skipped frames.
I am not entirely sure what I gain by using Chromium other than ridiculous build times. Same is true for Firefox though it builds faster. How long I didn't test yet.
More testing is needed.

lyallp · Post by **lyallp** » Tue May 14, 2024 9:39 am

Never tried seamonkey, keep me informed.