Possible, though I don't think likely in my case.
I'm not at home, so I can't give you the exact model, but I bought a
pretty powerful MSI motherboard for the CPU, got the dual RAM, a serious
aftermarket CPU fan, several case fans, etc. I don't overclock at all
because I've always been worried about over heating or that I'd get the
settings wrong and fry something. I've also run memtest86 for 24 hours,
though I don't think it does the HT thing. I'm not a hardware guru, I
just try and (barely) keep up with what's going on so that I can have
decent performance.
I know that anecdotal evidence isn't, but I've heard from at least one
other source on the redhat list back when I was trying this that he'd
had problems with smp kernels on HT processors as well, so I dunno.
We're running RHEL AS 2.1 on two Xeon HT processors and they don't seem
to have any probs. go figure.
Ben
Maciej ?enczykowski wrote:
>I'll just note here, that HT really pushes the proc, it might just be
>flaky hardware - especially if it's overclocked. I can run my computer
>for ours on end without problems overclocked 20% and everything is fine -
>but fire up two mersenne prime threads and it starts giving bad
>answers (a single one is ok and two work ok when overclocked to 15%). So
>even though for me the problem occurs at full throttle with 20%
>overclocking, your hardware might be worse and it might occur at full
>throttle (the kind which only the extra performance boost from HT can
>bring out) even without it being overclocked.
>
>Of course this is only 1 possibility... Maybe HT support on 2.4 is just
>plain buggy... don't know, never used HT on 2.4.
>
>On Thu, 3 Mar 2005, Benjamin J. Weiss wrote:
>
>
>
>>Well, when I had a process that ate a bunch of CPU on my P4 under the
>>2.4 kernel in HT mode, sometimes I'd get a kernel panic or other crash
>>in RHEL 3. For instance, if I was trying to run VMWare in smp mode
>>(running a Windows guest system on the RHEL 3 host system), sometimes
>>the vm would crash, twice the whole system died. If I ran the VMWare in
>>uni-processor kernel, it would run fine though it would eat resources.
>>
>>I haven't had that happen in XP yet, when running the RHEL/CentOS guest
>>operating systems on the XP host OS. I haven't had time to try it on
>>CentOS 4 as the host OS yet to see how the 2.6 kernel handles the load.
>>
>>Ben
>>_______________________________________________
>>
>>
