| View previous topic :: View next topic |
| Author |
Message |
fturco
Veteran
Joined: 08 Dec 2010
Posts: 1181
Location: Italy
|
 Posted: Mon Jan 21, 2019 10:32 pm Post subject: Doubts about CONFIG_NR_CPUS and MAKEOPTS options |
 |
|
I have a Gentoo desktop system with an old dual-core Intel Core 2 Duo CPU:
| Code: | $ lscpu
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 36 bits physical, 48 bits virtual
CPU(s): 2
On-line CPU(s) list: 0,1
Thread(s) per core: 1
Core(s) per socket: 2
Socket(s): 1
NUMA node(s): 1
Vendor ID: GenuineIntel
CPU family: 6
Model: 23
Model name: Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz
Stepping: 10
CPU MHz: 2989.401
BogoMIPS: 5978.79
Virtualization: VT-x
L1d cache: 32K
L1i cache: 32K
L2 cache: 6144K
NUMA node0 CPU(s): 0,1
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx lm constant_tsc arch_perfmon pebs bts rep_good nopl cpuid aperfmperf pni dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm sse4_1 xsave lahf_lm pti tpr_shadow vnmi flexpriority dtherm |
Since my CPU is dual-core, at first I built my kernel with the CONFIG_NR_CPUS=2 option. But after booting my system I get a warning message in the logs:
| Code: | | Sep 10 12:11:11 desktop kernel: smpboot: 4 Processors exceeds NR_CPUS limit of 2 |
If I set CONFIG_NR_CPUS=4 the previous warning message disappears, but I don't understand where the number 4 comes from.
Moreover if I set both CONFIG_NR_CPUS=4 and MAKEOPTS="-j4", emerge will be slightly slower at building packages than if I stay with CONFIG_NR_CPUS=2 and MAKEOPTS="-j2".
Does anyone have an idea? What are the optimal settings for my CPU? |
|
| Back to top |
|
 |
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54232
Location: 56N 3W
|
| Posted: Mon Jan 21, 2019 10:49 pm Post subject: |
|
|
fturco,
Well, Flags: includes ht, which is hyper-threading.
Is that disabled in your BIOS?
CONFIG_NR_CPUS=4 is correct with your Dual Core CPU if it really has hyper threading.
This Intel Page says no hyper threading.
CONFIG_NR_CPUS=4 costs you a few kB of waste RAM if its not real, so go with it.
MAKEOPTS= Is not related only to your CPU. Each MAKEOPTS thread can want 2G RAM on bigger packages.
When you run out out of RAM, one way or another, you start swapping.
How many tux logos do you get on your framebuffer boot screen?
If you get four, the kernel thinks you have four logical cores.
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
|
mike155
Advocate
Joined: 17 Sep 2010
Posts: 4438
Location: Frankfurt, Germany
|
|
| Back to top |
|
|
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54232
Location: 56N 3W
|
| Posted: Tue Jan 22, 2019 12:03 pm Post subject: |
|
|
mike155,
That's the page I linked too.
I wonder if hyperthreading was added with a microcode update ?
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
|
fturco
Veteran
Joined: 08 Dec 2010
Posts: 1181
Location: Italy
|
| Posted: Tue Jan 22, 2019 12:47 pm Post subject: |
|
|
/usr/src/linux/.config
I have an Intel DQ35JO motherboard.
I don't upgrade the microcode.
With the next reboot I'll check any reference to hyper-threading in the BIOS and the number of Tux logos on screen.
Edit #1:
I haven't found any reference to hyper-threading in the BIOS.
CONFIG_LOGO is currently disabled in my kernel. Should I enable it?
Edit #2:
Even if I enable CONFIG_LOGO in the kernel it seems I cannot display it at boot. I tried removing the "quiet" option and adding "nomodeset", to no avail.
dmesg
Edit #3:
I forgot to tell you I'm not using a Gentoo kernel. I'm using linux-libre. I'm not sure if this matters. |
|
| Back to top |
|
|
mike155
Advocate
Joined: 17 Sep 2010
Posts: 4438
Location: Frankfurt, Germany
|
| Posted: Tue Jan 22, 2019 1:50 pm Post subject: |
|
|
@NeddySeagoon:
I'm sorry - i didn't see the link in your post.
It's quite possible that Intel added hyperthreading with a microcode update. But it's unlikely. Furthermore, Google doesn't show any articles about such an event...
@fturco: thanks for the files you posted. I don't see anything wrong in your kernel config or in dmesg.
Intel ark shows that the mainboard was released earlier than the processor. It may well be that your mainboard needs a BIOS update: https://ark.intel.com/products/50382/Intel-Desktop-Board-DQ35JO |
|
| Back to top |
|
|
fturco
Veteran
Joined: 08 Dec 2010
Posts: 1181
Location: Italy
|
| Posted: Tue Jan 22, 2019 2:09 pm Post subject: |
|
|
| I'll try updating the BIOS. Thanks for the suggestion. |
|
| Back to top |
|
|
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54232
Location: 56N 3W
|
| Posted: Tue Jan 22, 2019 4:25 pm Post subject: |
|
|
fturco,
Your dmesg output would be useful.
CONFIG_DRM_I915=y gets you a frambuffer console on Intel GPUs
CONFIG_FRAMEBUFFER_CONSOLE=y says you prefer a frambuffer console if there is one.
# CONFIG_LOGO is not set says that you don't want any tux logos at boot. Its not really functional.
Its decoration on the console and a diagnostic aid as you get one image for each detected logical core.
linux-libre only prevents you using the in kernel firmware blobs. Some Intel GPUs need firmware some don't
If they need firmware and its not provided, missing functionality varies.
Its not an issue here.
Take care updating the BIOS. If the update fails, it can leave the system unable to boot.
This is called 'bricking' your motherboard. Recovery is non trivial.
A few systems do provision for recovery but its very ugly.
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
|
mike155
Advocate
Joined: 17 Sep 2010
Posts: 4438
Location: Frankfurt, Germany
|
| Posted: Tue Jan 22, 2019 4:34 pm Post subject: |
|
|
| @NeddySeagoon: fturco posted his dmesg output (see Edit #2 above) |
|
| Back to top |
|
|
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54232
Location: 56N 3W
|
| Posted: Tue Jan 22, 2019 5:21 pm Post subject: |
|
|
mike155,
Thank you.
| dmesg: | [ 0.261938] smp: Bringing up secondary CPUs ...
[ 0.261938] x86: Booting SMP configuration:
[ 0.261938] .... node #0, CPUs: #1
[ 0.262969] smp: Brought up 1 node, 2 CPUs
[ 0.262969] smpboot: Max logical packages: 2
[ 0.262969] smpboot: Total of 2 processors activated (11957.78 BogoMIPS) |
and no complaints about more CPUs than the kernel can drive but it is set up for 4 CPUs
| dmesg: | [ 1.117779] fbcon: inteldrmfb (fb0) is primary device
[ 1.273260] Console: switching to colour frame buffer device 240x75
[ 1.289987] i915 0000:00:02.0: fb0: inteldrmfb frame buffer device |
That's good.
If you turned on logos, I think you would get two.
I recall something about Intel setting the ht flag in some CPU that were not hyperthreading capable but it was a long time ago.
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
|
fturco
Veteran
Joined: 08 Dec 2010
Posts: 1181
Location: Italy
|
| Posted: Tue Jan 22, 2019 6:20 pm Post subject: |
|
|
The latest BIOS for my motherboard is from 2010 (!). Anyway, I already have the latest version (1143).
Edit #1: I went ahead and filed a bug upstream: https://bugzilla.kernel.org/show_bug.cgi?id=202385
Edit #2: I forgot to thank you both for the help! |
|
| Back to top |
|
|
|
Kernel & Hardware
