I am running an 11.2 SuSE Linux system. I had to upgrade it from a dual core AMD CPU/MOBO (thank you ERCOT – if you are from TX USA you will understand).
Ok, first problem was the replacement Video card (GeFarce 8400 GS) failed, and the 7200 GS worked just fine.
Got the thing booted and VMware back up and running. Got the hard drives restored.
But I noticed that something was just not right. So I checked various things and it seems that I am only running on 1 core. I have gone into the BIOS and enabled all 6 cores, and I have set MAXCPUS=4 and cat /proc/cpuinfo shows that I am only running one core.
What do I need to do to this system to get it to run multiple “CPUs”?
I would like to “assign” 2 CPUs to VMware so that W/XP will have a bit better performance.
I have run yast looking for a different kernel, for anything to control the CPUs, and no joy.
How do you really know you’re running on only one core?
The easiest way to observe which core a process is running on is to install and run htop (You can do it with top as well with some extra configuration but IMO just running htop is easier and in the long run may be beneficial).
“Just so I understand this: you are having problems with multiple core on the VMWare Guest machines, not on the Host, correct?”
I wish to set VMware to use 2 CPUs. Can’t do that if there is only one being used by the system.
How do you really know you’re running on only one core?
I think I mentioned having run: cat /proc/cpuinfo . And I also use “System Monitor”.
“That does not help us to help you. We need facts, not interpretations. One way to provide some facts would be giving us the output of”
You gets what you gets. “and cat /proc/cpuinfo shows that I am only running one core.” I can read. However, also doing support on mainframes, yes, you do get some dunderheads. But, if I knew enough to discuss cat /proc/cpuinfo, and setting maxcpus at loader time (in my case GRUB), I’m probably not in the “dunderhead, can’t believe anything they say” group.
A bit of history:
I have run different versions of SuSE and when running 10.3 through 11.1, the monitor I use tracks each CPU on a moving graph, each CPU having a different color. System Monitor is only tracking 1 CPU. I mentioned mainframes. I do system level code. I am use to controlling the CPUs that the hardware supplies. So, I probably understand what I’m reading. And near as I can tell, the ONLY CPU being used is 0, not 1, 2, 3, 4, or 5 – numbers are relative to zero). Given that, I have also looked at top. I thought (but honestly don’t remember) that it would indicate if there were more than one CPU.
But again, this MOBO, and CPU are new. The AMD CPU I was using was a dual core, and when I check on that system (11.2) it showed 2 CPUs.
So is there something about a “6 core” CPU that will require special code/module/kernel?
Your VMWare references are confusing the issue: 1) is your host machine (the bare metal box) a Linux machine? 2) If so, is that machine not recognizing your multiple cores?
The HOST SCP (System Control Program) is Linux. It seems to only be able to see/handle CPU 0. VMware is so that I can run W/XP as a guest.
You did cause me to think about something. In testing 11.1, for RAID, I believe the 64bit systems had an install problem that the 32bit systems did not. This is a 64 bit system with 8GB of RAM.
In using Google and searching the Novell forums, I did not see any specific mention of a “multi-core”/“multi-CPU” problem. Perhaps I should check a few other distros.
Alright - so I understand now that you have a bare metal system that’s running openSUSE and somehow doesn’t recognize your CPU count. This is something I have never heard of and you’re going to have to help us help you. As per @gropiuskalle: post the output of “cat /proc/cpuinfo” back to this forum. Also, post the output of “uname -a” as well. Maybe there’s something we can find for you.
On 02/23/2011 11:36 PM, wylbur wrote:
>
> The HOST SCP (System Control Program) is Linux. It seems to only be able
> to see/handle CPU 0. VMware is so that I can run W/XP as a guest.
>
> You did cause me to think about something. In testing 11.1, for RAID, I
> believe the 64bit systems had an install problem that the 32bit systems
> did not. This is a 64 bit system with 8GB of RAM.
>
> In using Google and searching the Novell forums, I did not see any
> specific mention of a “multi-core”/“multi-CPU” problem. Perhaps I should
> check a few other distros.
The default MAX CPUs for an openSUSE kernel is 32, thus you should have no
problem with 6.
Have you looked at the dmesg output? Perhaps some condition is encountered due
to a faulty BIOS or a chip set that the code cannot handle.
The answer to my question is, there is a BIOS setting that needed to be changed. And in this case, it was in the <Power> tab. I set ACPI APIC support to ENABLED and then I set ACPI 2.0 Support to ENABLED.
Now top shows 4 CPUs, System Monitor shows 4 CPUS.
So, now I can go reset to a max of 6 and reboot and see if I can cut the boot time even further (I went from 2 minutes to ~50 seconds by picking up the other 3 CPUs).
I found the box and re-read what it has. This is a 4 core only CPU. The motherboard supports up to 6 cores.
So, mysteries solved.
I can read top output, and such.
I actually do know what it means.
I need to get reading glasses that let me read poorly printed 8 point font in the MOBO manual so I can see where you set certain things.
And now wife is ecstatic that her workstation runs very fast. And she now understands why we need the file server that I’m resurrecting, so that we have tape backup once again.