Re: dmesg - cpu, smt, core, package

2022-02-13 Thread Stuart Henderson
On 2022-02-12, Mihai Popescu wrote: > Doing a wrap up, how an Intel i5 family cpu is seen by OpenBSD? > The family has 4 cores / 4 threads per package. > If someone can check please, with dmesg and sysctl. There are many different i5 CPUs, at least with 2, 4 or 6 cores, and usually (maybe

Re: dmesg - cpu, smt, core, package

2022-02-12 Thread Mihai Popescu
Doing a wrap up, how an Intel i5 family cpu is seen by OpenBSD? The family has 4 cores / 4 threads per package. If someone can check please, with dmesg and sysctl. Thanks.

Re: dmesg - cpu, smt, core, package

2022-02-12 Thread Bryan Steele
pre-Ryzen AMD CPUs did not have SMT, but they had "CMT" or "clustered multithreading" which is the shared-FPU stuff, hw.smt=0 disables that too on these CPUs. I believe this was intentional as this kind of resource sharing between cores comes with inherent risk-- FPU state can contain things like

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Mihai Popescu
I did set hw.smt=1 and all are shown now "online". I'm a little bit worry since I switched to this cpu from a 2 core AMD. Total mess in the CPU manufacturer yard. But we cannot produce our own cpu in the kitchen, can't we? # sysctl hw.smt=1 hw.smt: 0 -> 1 hw.machine=amd64 hw.model=AMD A8-5500B

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Stuart Henderson
On 2022-02-10, Mihai Popescu wrote: > I see in the AMD spec that I have a 4 core CPU, no *-threading. So it > is 4 core, 4 threads, no multi, simultaneous, hyper, etc. > Still the sysctl hw.ncpuonline reports 2, but it found 4 and top > displays CPU0 and CPU2 only ( that is 2). I am aware that

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Hrvoje Popovski
On 10.2.2022. 20:03, Mihai Popescu wrote: >> you mean gaps because HT is disabled ? > > I think they are disabled from the factory, cores that are not 100% > functional, i.e defects. > There is one line for a family, the luckiest ones have the maximum > number of cores and $$$, the rest are lower

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Mihai Popescu
> you mean gaps because HT is disabled ? I think they are disabled from the factory, cores that are not 100% functional, i.e defects. There is one line for a family, the luckiest ones have the maximum number of cores and $$$, the rest are lower but still functional on the advertised cores.

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Hrvoje Popovski
On 10.2.2022. 16:38, Todd C. Miller wrote: > On Thu, 10 Feb 2022 08:46:37 +, Stuart Henderson wrote: > >> The numbers come from what's reported by the relevant CPUID instruction, >> the only one actually used by OpenBSD is smt to disable all but one >> thread in a core, otherwise they're just

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Mihai Popescu
I see in the AMD spec that I have a 4 core CPU, no *-threading. So it is 4 core, 4 threads, no multi, simultaneous, hyper, etc. Still the sysctl hw.ncpuonline reports 2, but it found 4 and top displays CPU0 and CPU2 only ( that is 2). I am aware that each pair of cores is using a shared FPU. My

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Stuart Henderson
On 2022-02-10, Mihai Popescu wrote: >> ... gaps ... > > What gaps are you referring to? The ones in dmesg for Intel in OP? in Hrvoje's dmesg for AMD. > Also the smt is from simultaneous multithreading? yes (hyperthreading on Intel CPUs), these are what are disabled by the default setting

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Mihai Popescu
> ... gaps ... What gaps are you referring to? The ones in dmesg for Intel in OP? Also the smt is from simultaneous multithreading? top also has a strange report, but i think is is known. sysctl hw. has 2 cpu online from 4 found. load averages: 0.37, 0.43, 0.21

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Todd C . Miller
On Thu, 10 Feb 2022 08:46:37 +, Stuart Henderson wrote: > The numbers come from what's reported by the relevant CPUID instruction, > the only one actually used by OpenBSD is smt to disable all but one > thread in a core, otherwise they're just for information. > > I'm not sure the reason for

Re: dmesg - cpu, smt, core, package

2022-02-10 Thread Stuart Henderson
On 2022-02-09, Hrvoje Popovski wrote: > On 9.2.2022. 19:04, Kapetanakis Giannis wrote: >> On 09/02/2022 19:48, Mihai Popescu wrote: >>> $ dmesg | grep smt >>> cpu0: smt 0, core 0, package 0 >>> cpu1: smt 1, core 0, package 0 >>> cpu2: smt 0, core 1, package 0 >>> cpu3: smt 1, core 1, package 0

Re: dmesg - cpu, smt, core, package

2022-02-09 Thread Hrvoje Popovski
On 9.2.2022. 19:04, Kapetanakis Giannis wrote: > On 09/02/2022 19:48, Mihai Popescu wrote: >> $ dmesg | grep smt >> cpu0: smt 0, core 0, package 0 >> cpu1: smt 1, core 0, package 0 >> cpu2: smt 0, core 1, package 0 >> cpu3: smt 1, core 1, package 0 >> >> for >> >> AMD A8-5500B APU with Radeon(tm)

Re: dmesg - cpu, smt, core, package

2022-02-09 Thread Kapetanakis Giannis
On 09/02/2022 19:48, Mihai Popescu wrote: > $ dmesg | grep smt > cpu0: smt 0, core 0, package 0 > cpu1: smt 1, core 0, package 0 > cpu2: smt 0, core 1, package 0 > cpu3: smt 1, core 1, package 0 > > for > > AMD A8-5500B APU with Radeon(tm) HD Graphics, 3194.46 MHz, 15-10-01 > > What could be the

Re: dmesg - cpu, smt, core, package

2022-02-09 Thread Mihai Popescu
$ dmesg | grep smt cpu0: smt 0, core 0, package 0 cpu1: smt 1, core 0, package 0 cpu2: smt 0, core 1, package 0 cpu3: smt 1, core 1, package 0 for AMD A8-5500B APU with Radeon(tm) HD Graphics, 3194.46 MHz, 15-10-01 What could be the "smt" thing?

dmesg - cpu, smt, core, package

2022-02-08 Thread Hrvoje Popovski
Hi all, in one supermicro box in dmesg i'm seeing this smc24# dmesg | grep smt cpu0: smt 0, core 0, package 0 cpu1: smt 0, core 1, package 0 cpu2: smt 0, core 2, package 0 cpu3: smt 0, core 3, package 0 cpu4: smt 0, core 4, package 0 cpu5: smt 0, core 5, package 0 cpu6: smt 0, core 8, package 0