Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #175896 > unrolled thread

checking that all cpus are being used in my installation....

Back to article view | Back to linux.debian.user

Contents

  checking that all cpus are being used in my installation.... Michael Fothergill <michael.fothergill@googlemail.com> - 2016-12-23 21:50 +0100
    Re: checking that all cpus are being used in my installation.... Ben Caradoc-Davies <ben@transient.nz> - 2016-12-24 01:20 +0100
      Re: checking that all cpus are being used in my installation.... Michael Fothergill <michael.fothergill@googlemail.com> - 2016-12-24 13:30 +0100
    Re: checking that all cpus are being used in my installation.... Dan Ritter <dsr@randomstring.org> - 2016-12-24 12:40 +0100

#175896 — checking that all cpus are being used in my installation....

[Multipart message — attachments visible in raw view] — view raw

​Dear Folks,

I am running Debian stretch on a pc with an AMD FX 8350 chip and a
sabertooth FX 990 motherboard.

I am interested to check a few things are working as they would be expected
to here.

I ran cat /proc/cpuinfo

and got

processor : 7
vendor_id : AuthenticAMD
cpu family : 21
model : 2
model name : AMD FX(tm)-8350 Eight-Core Processor
stepping : 0
microcode : 0x600084f
cpu MHz : 4013.300
cache size : 2048 KB
physical id : 0
siblings : 8
core id : 3
cpu cores : 4
apicid : 23
initial apicid : 7
fpu : yes
fpu_exception : yes
cpuid level : 13
wp : yes
flags : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat
pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb
rdtscp lm constant_tsc rep_good nopl nonstop_tsc extd_apicid aperfmperf
eagerfpu pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 popcnt aes
xsave avx f16c lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a
misalignsse 3dnowprefetch osvw ibs xop skinit wdt lwp fma4 tce nodeid_msr
tbm topoext perfctr_core perfctr_nb cpb hw_pstate vmmcall bmi1 arat npt
lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists
pausefilter pfthreshold
bugs : fxsave_leak sysret_ss_attrs null_seg
bogomips : 8026.82
TLB size : 1536 4K pages
clflush size : 64
cache_alignment : 64
address sizes : 48 bits physical, 48 bits virtual
power management: ts ttp tm 100mhzsteps hwpstate cpb eff_freq_ro


​Does this seem reasonable?

Should the number of cpu cores be 8 rather than 4 (listed above)?

If you look at the discussion below:

https://ubuntuforums.org/showthread.php?t=2233301

Only one cpu was working out of all 8 for the original poster until they
turned the ACPI setting to ACPI=force apparently.

I suppose I could check the grub.cfg file in /boot or wherever it lives and
see what the ACPI setting is in my case.

In gentoo you set makeopts to -j9 and make =-j9 when you compile a fresh
kernel and then it should run quickly and as long as your power supply unit
is in good shape then things should go smoothly....

In Debian I think there is something called makeflags but I am not sure
(comments appreciated here).

I am using kernel 4.8.2 I think.

I could download e.g. kernel 4.9.0 and then manually compile it using eg
make menuconfig and the existing .config file.

If I did that would it be make -j9 by default in debian or do I need to
specify that in the same way as you would in gentoo?

Also it should then compile pretty swiftly and smoothly (provided the power
supply is a good one) should it not?

Suggestions on performance testing the FX 8350 chip on this install are
welcome.

Regards

Michael Fothergill

[toc] | [next] | [standalone]

#175900

On 24/12/16 09:45, Michael Fothergill wrote:
> Suggestions on performance testing the FX 8350 chip on this install are
> welcome.

For an open source multithreaded CPU load with a handy benchmark mode, 
install povray, then run:

povray -benchmark

and compare the resulting time in seconds with:
https://openbenchmarking.org/showdown/pts/povray

(FX-8350 listed as 812 seconds.)

Your cpuinfo suggests that all 8 processors are detected. The "cpu 
cores" might refer to the four "modules" of this package, each with two 
integer and one floating point unit. POV-Ray is FPU intensive so expect 
quad-core performance.

Kind regards,

-- 
Ben Caradoc-Davies <ben@transient.nz>
Director
Transient Software Limited <http://transient.nz/>
New Zealand

[toc] | [prev] | [next] | [standalone]

#175909

[Multipart message — attachments visible in raw view] — view raw

On 24 December 2016 at 00:10, Ben Caradoc-Davies <ben@transient.nz> wrote:

> On 24/12/16 09:45, Michael Fothergill wrote:
>
>> Suggestions on performance testing the FX 8350 chip on this install are
>> welcome.
>>
>
> For an open source multithreaded CPU load with a handy benchmark mode,
> install povray, then run:
>
> povray -benchmark
>
> and compare the resulting time in seconds with:
> https://openbenchmarking.org/showdown/pts/povray
>
> (FX-8350 listed as 812 seconds.)
>

​Many thanks for this advice.  I installed povray and gkrell and then ran
povray -benchmark as you suggested.

I also checked the cpu temperature as well as the cpu core activity.

As the test ran I noticed two things:

1.  All 8 cpu cores ran simultaneously at 100% capacity.

2. The cpu temperature rose gradually until the benchmark was about 85%
complete - by then the temperature exceeded 84C and the cpu automatically
cut out and the pc shut itself down.

When I bought the motherboard and cpu I got the retailer to switch the
standard stock cooler for an Arctic Freezer 13 model.

They installed it for me and I fitted the motherboard into the PC cabinet
with the cooler.

Concerning the Arctic Freezer 13 cooler, I would have thought it should be
good enough to cool this cpu adequately in this situation, but I am open to
comments and suggestions on this.

I would have thought the cooler should work adequately if the following
conditions were met in my case:

1. The cooler was correctly installed and attached to the motherboard and
cpu.

2. Adequate thermal contact between the heat sink and the cpu was achieved
using e.g. properly swabbed and sautéed thermal paste.

3. The fan is turned on an dynamically controlled to respond to temperature
changes vigorously as needed in this test case.

Or it would fail.

I could open the cab door and check by eye that the fan does actually start
whirring vigorously when the temperature increases in the povray test. I
think that gkrell should probably output the fan speeds continuously as
well temp and cpu load etc.

However there is one other mysterious factor here.  The Arctic Freezer 13
cooler apparently has 9 different fan settings.  I have tried googling
around but I don't know how to change the fan settings on it.......   This
seems to be an enigma.

Does anyone know if it would be possible to change the fan speed settings
in Debian in some way?  Or would I have to manually adjust them in the
cabinet after shutting the off?

Comments appreciated on this.

Regards

MIchael Fothergill


>
> Your cpuinfo suggests that all 8 processors are detected. The "cpu cores"
> might refer to the four "modules" of this package, each with two integer
> and one floating point unit. POV-Ray is FPU intensive so expect quad-core
> performance.
>
> Kind regards,
>
> --
> Ben Caradoc-Davies <ben@transient.nz>
> Director
> Transient Software Limited <http://transient.nz/>
> New Zealand
>

[toc] | [prev] | [next] | [standalone]

#175906

On Fri, Dec 23, 2016 at 08:45:29PM +0000, Michael Fothergill wrote:
> ​Dear Folks,
> 
> physical id : 0
> siblings : 8
> core id : 3
> cpu cores : 4
> 
> ​Does this seem reasonable?

Yes.
 
> Should the number of cpu cores be 8 rather than 4 (listed above)?

It is 8. See the siblings field? cat /proc/cpuinfo shows you all
8 cores, but a lot of it will scroll off your screen.

The 8350 is arranged as two 4-core CPUs sharing the same module.

> If I did that would it be make -j9 by default in debian or do I need to
> specify that in the same way as you would in gentoo?

Specify it.

> Also it should then compile pretty swiftly and smoothly (provided the power
> supply is a good one) should it not?

RAM and disk play a big part, too.

If the irqbalance daemon is not running, you may want to install
it. irqbalance spreads interrupt load over as many CPU cores as
you have, which in most *but not all* situations leads to better
overall performance. 

-dsr-

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web