So far as I know, Celeron has 128K 2nd level cache with cache speed same as
processor and PII has 512K 2nd level cache with cache speed half the
processor. Anybody correct me if I am wrong. I think there should be some
more minor different but I don't know about them in details.
In my dual Pentium 233 MMX PC, I observed that the bogomips of the second
processor reported is lower than the first processor by around 10%. But I
cannot sure whether it is a problem or not since I am over-clocking them.
They are actually 200MHz.
I don't know a way to disable one processor under Linux and check out the
speed of the other one. May be other OS like BeOS or Windows NT can do so,
or I just don't know the way in Linux. You can test it under other OS to
see the different.
However, you will always observe that the overall system performance is not
close to twice a single processor system since there are a lot of SMP
overhead, both in hardware and software. My experience is: it is quite
significant, I observed 20-30% overhead in my PC.
That may be one of the reason why you feel the second processor is slower.
Both of them should be slower if they are running at the same time. If the
second processor really slower than the first one, it should be the reason
of the main board design again.
S.L.Mo
At 02:27 AM 11/16/99 -0700, Ondrej Florian wrote:
>
>...don't get me wrong. Celerons I use works just fine with my Linux
>installation. System is solid, no crashes or anything unusuall. It's just
>that cpuinfo will no recognize 2nd processor correctly. I think that what
>really is happening is that motherboard I use do not set cache for 2nd CPU
>correctly (my motherboard actually doesn't support dual Celeron
>officialy). BTW: does PII init differs so much from Celeron ?? maybe BIOS
>doesn't inicialize 2nd CPU at all, because when I boot into Windows NT
>(SMP mode), I don't see any difference in pernformace, but monitor shows
>2nd CPU to be working... In Linux on the other hand I can see some
>pernfornmace gain (compilation with "-j", many proccesse running at once,
>etc.) I guess mtrr fixed the problem (according to doc it should
>inicialize 2nd CPU) -- but it seems that 2nd CPU is a bit slower ...maybe
>it's my imagination, anyways is there any way how can I test speed of only
>one CPU in the SMP system ??
>
>
>-----------------------------------------------------------------------
> Ondrej Florian
> [EMAIL PROTECTED]
> ICQ: 5508146
>
>On Tue, 16 Nov 1999, S.L.Mo wrote:
>
>> I have two dual processor PC. Both of them I didn't intentionally buy
>> processors of the same batch but they all works fine. I think Intel won't
>> be that poor in QA for their processors.
>>
>> The problem in this case is not they didn't work but a wrong recognition of
>> processor type. Both bogomips and clock are reported correctly. So I
>> suspect the problem is in the BIOS or Main Board setting. Especially if it
>> is always the second processor recognized wrongly no matter which
>> processors you put it.
>>
>> I cannot sure Linux detect processor type directly or go through BIOS, or
>> be affected by the BIOS. But I don't expect that is a Linux kernel
>> problem, or at least, by its own.
>>
>> I heard about Celeron SMP support was removed. But if this case is using
>> such Celeron processor, I cannot sure it will work or just reporting a
>> wrong processor ID.
>>
>> S.L.Mo
>>
>> At 01:50 AM 11/16/99 -0600, Andrew J. Pavlis wrote:
>> >Likewise, I am runnning dual 400mhz celeron cpu's. From my understanding
>> >(perhaps I read it in the faq), dual proc's run best when they are the
>> EXACT same
>> >processor, right down to the manufacturing location. I bought my board
>> and procs
>> >at the same place and received the exact same processors, from the same
>> factory.
>> >So, my processors show up just fine. This won't really help with your
>> problem,
>> >just an FYI. I am running BP6 as well, perhaps your board has a problem
>> running
>> >dual celerons. . . . .
>> >
>> >Another aside, Intel cut the ability in celerons to run on dual processor
>> boards
>> >around August. So if your processors are manuf. from August on, this
>> could be
>> >the effect of that operation.......but then, I am really stretching on
>> this one.
>> >Intel's attitude annoys the h*ll out of me. If I buy your product, I
>> should be
>> >able to do whatever I want with it. Did you test two different processors
>> from
>> >the same genre?
>> >
>> >-Andy
>> >
>> >Geoff Cross wrote:
>> >
>> >> yes it should be recognized right because I'm running two celeron 500mhz
>> >> processors in a BP6 and linux recognizes both as celeron 128kb of
cache at
>> >> 500mhz.
>> >>
>> >> ----- Original Message -----
>> >> From: Ondrej Florian <[EMAIL PROTECTED]>
>> >> To: <[EMAIL PROTECTED]>
>> >> Sent: Tuesday, November 16, 1999 12:26 AM
>> >> Subject: Re: Dual Celeron problem
>> >>
>> >> >
>> >> > ...actually, I have tried Celerons 400 (differen CPUs), so problem
really
>> >> > seems to be in 2nd CPU inicialization
>> >> >
>> >> > ...I don't know how how cpuinfo really works under Linux but
shouldn't it
>> >> > read ID on CPU directly ...I mean should Celeron be Celeron no
matter how
>> >> > it was inicialized ?? ...ID is "burned in" right ??
>> >> >
>> >> >
>> >> >
-----------------------------------------------------------------------
>> >> > Ondrej Florian
>> >> > [EMAIL PROTECTED]
>> >> > ICQ: 5508146
>> >> >
>> >> > On Tue, 16 Nov 1999, S.L.Mo wrote:
>> >> >
>> >> > > Did you swap the two processors and still see the same thing?
What I
>> >> mean
>> >> > > is whether the wrong recognition is on the 2nd processor or for the
>> >> > > particular processor. It should be useful for further diagnosis.
>> >> > >
>> >> > > Regards,
>> >> > >
>> >> > > S.L.Mo
>> >> > >
>> >> > > At 10:50 PM 11/15/99 -0700, Ondrej Florian wrote:
>> >> > > >
>> >> > > >Hi I'm running M720 motherboard with two Celeron 433 processors.
>> >> > > >Everything seems to be OK except that when I look at the CPU
info in a
>> >> > > >short version I get this
>> >> > > >
>> >> > > >processor : 0
>> >> > > >vendor_id : GenuineIntel
>> >> > > >cpu family : 6
>> >> > > >model : 6
>> >> > > >model name : Celeron (Mendocino)
>> >> > > >stepping : 5
>> >> > > >cpu MHz : 434.322223
>> >> > > >cache size : 128 KB
>> >> > > >bogomips : 432.54
>> >> > > >
>> >> > > >processor : 1
>> >> > > >vendor_id : GenuineIntel
>> >> > > >cpu family : 6
>> >> > > >model : 6
>> >> > > >model name : Mobile Pentium II
>> >> > > >stepping : 5
>> >> > > >cpu MHz : 434.322223
>> >> > > >cache size : 0 KB
>> >> > > >bogomips : 434.18
>> >> > > >
>> >> > > >...one processor is recognized correctly, other one is
recognized as
>> >> Mobile
>> >> > > >Pentium II without cache. It does not matter what Kernel version
I run
>> >> or
>> >> > > >what if mtrr support is compiled in.
>> >> > > >
>> >> > > >my hardware is PCChips 720 MB (Intel LX chipset) with APIC chip.
>> >> > > >
>> >> > > >...I guess CPU do not get properly initialized during BIOS
startup --
>> >> > > >shouldn't mtrr fix it ??
>> >> > > >
>> >> > > >Thanx, Ondrej Florian
>> >> > > >
>> >> > > >
>> >> > >
>> >-----------------------------------------------------------------------
>> >> > > > Ondrej Florian
>> >> > > > [EMAIL PROTECTED]
>> >> > > > ICQ: 5508146
>> >> > > >
>> >> > > >-
>> >> > > >Linux SMP list: FIRST see FAQ at
>> >> http://www.irisa.fr/prive/dmentre/smp-howto/
>> >> > > >To Unsubscribe: send "unsubscribe linux-smp" to
>> >> [EMAIL PROTECTED]
>> >> > > >
>> >> > >
>> >> > > -
>> >> > > Linux SMP list: FIRST see FAQ at
>> >> http://www.irisa.fr/prive/dmentre/smp-howto/
>> >> > > To Unsubscribe: send "unsubscribe linux-smp" to
>> >> [EMAIL PROTECTED]
>> >> > >
>> >> >
>> >> > -
>> >> > Linux SMP list: FIRST see FAQ at
>> >> http://www.irisa.fr/prive/dmentre/smp-howto/
>> >> > To Unsubscribe: send "unsubscribe linux-smp" to
>> [EMAIL PROTECTED]
>> >> >
>> >>
>> >> -
>> >> Linux SMP list: FIRST see FAQ at
>> http://www.irisa.fr/prive/dmentre/smp-howto/
>> >> To Unsubscribe: send "unsubscribe linux-smp" to
[EMAIL PROTECTED]
>> >
>> >-
>> >Linux SMP list: FIRST see FAQ at
http://www.irisa.fr/prive/dmentre/smp-howto/
>> >To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
>> >
>>
>> -
>> Linux SMP list: FIRST see FAQ at
http://www.irisa.fr/prive/dmentre/smp-howto/
>> To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
>>
>
-
Linux SMP list: FIRST see FAQ at http://www.irisa.fr/prive/dmentre/smp-howto/
To Unsubscribe: send "unsubscribe linux-smp" to [EMAIL PROTECTED]
