CPU Internal Core Speed

[Daave]

My PC is 5½ years old. It's a barebones model with an EliteGroup
P6SET-ML motherboard with onboard audio and video and 32 KB primary
memory cache and 128 KB secondary memory cache. The CPU/SDRAM bus
frequency is 66/66 MHz. It has a 500 MHz PPGA (socket 370) Celeron
processor.

I've been looking at different things to tweak to eek out more power (I
know...not a lot by today's standards!).

One thing I thought was odd were the settings from the BIOS & CPU
Features Setup:

CPU Internal Core Speed: 233 MHz
CPU/SDRAM Bus Frequeny: 66/66 MHz
CPU Core: Bus Freq.Multiple: 3.5x

Okay, I know that 3.5 times 66 = 231 (close enough to the 233 listed).
But the speed of my processor is 500 MHz! What's going on?

 

[Paul]

My PC is 5½ years old. It's a barebones model with an EliteGroup
P6SET-ML motherboard with onboard audio and video and 32 KB primary
memory cache and 128 KB secondary memory cache. The CPU/SDRAM bus
frequency is 66/66 MHz. It has a 500 MHz PPGA (socket 370) Celeron
processor.

I've been looking at different things to tweak to eek out more power (I
know...not a lot by today's standards!).

One thing I thought was odd were the settings from the BIOS & CPU
Features Setup:

CPU Internal Core Speed: 233 MHz
CPU/SDRAM Bus Frequeny: 66/66 MHz
CPU Core: Bus Freq.Multiple: 3.5x

Okay, I know that 3.5 times 66 = 231 (close enough to the 233 listed).
But the speed of my processor is 500 MHz! What's going on?

There was a time, when the multiplier setting could be set via
jumpers or dip switch. But then Intel started locking the multiplier
inside the processor. This does two things. It gets around any issues
with the range of options offered by the jumpers. (Like processors where
the multiplier goes up to 14X.) But it also prevents overclocking via
multiplier jumpers only. To overclock, you could try changing the FSB clock,
with the downside of that being, that the old boards tied PCI and AGP
clock to the FSB clock, so the peripheral busses could end up overclocked
past their limits. On modern boards, the PCI and AGP clocks are "locked",
meaning they are independent and async to the other clocks in the system.
On a modern board, PCI and AGP stay at the proper frequencies, while you
adjust the FSB. On the older boards, they were still tied together.

(On the machine I'm typing on, I run at 100MHz, giving PCI=33 and AGP=66.
If I bump the clock to 112MHz, the PCI is 37.3 and AGP is 74.6MHz. Which
is a comfortable limit for the PCI at least. Some of the older AGP cards
could be run at 100MHz on the AGP bus, so an older AGP card may be less
of a limit, than proper operation of the PCI bus.)

Chances are, whatever you find as a substitute for your processor,
it will have a locked multiplier as well. So whatever the multiplier
is set at, you'll be able to continue to ignore it.

This page suggests a SECC2 form factor processor, as a faster alternative.
Mainly because the S370 "PPGA" suggests to me, that more modern S370
processors may not be compatible with the socket. It is possible there
were some hardware hacks you could do at the socket level (maybe some
drilled pins and wired jumpers), to allow a Coppermine S370 to be used.
But that is not likely to be a lot of fun, if this is your only computer.
And it may turn out, that it is easier to find an SECC2, than the right
S370 processor.

http://www.mainboard.cz/mb/ecs/P6SET-ML.htm

I think I can see where the 850MHz limit is coming from. If I look at
the Pentium III section of processorfinder.intel.com , and set the
selector to 100MHz only, the fastest SECC2 processor is 850MHz/100MHz/256KB.
That is a CPUID of 0683.

http://processorfinder.intel.com/List.aspx?ProcFam=25&sSpec=&OrdCode=
http://processorfinder.intel.com/details.aspx?sSpec=SL43F

If you were to buy a 133Mhz FSB processor, and run it at 100Mhz, I
think the top multiplier on those is lower than the 850MHz/100MHz
processor, so you'd be no further ahead. So at first glance, it
does look like the 850MHz/100MHz SECC2 might be your best alternative.
(At least, if I can trust that mainboard.cz listing as being correct.
You can hunt around ecs.com.tw for more info I supposed.)

It is also possible, you could plug a Powerleap processor, or a SECC2
slocket adapter with S370 socket on it. Yes, that does open a range of
new possibilities, up to 1.4GHz/100MHz level of S370 processor. But
some of the old BIOS had trouble with "high multiplier" processors,
and not all motherboard manufacturers, continued to produce BIOS
updates, to allow using devices like that. On my old board, I can use a
1.4GHz processor, but I have to manually add microcode using CTMC, so
my processor is not completely supported by using the latest BIOS.
Still, my BIOS is in better shape than some other motherboards of
that era. Maybe your 850MHz/100Mhz SECC2 module would have microcode
support, and fit within the limits of the BIOS.

If you could find an 850/100/256 processor for free, then doing the
upgrade may be worth it. But if you have to pay money for it, keep
in mind that your motherboard won't last forever. You might add a
new processor, only to have the motherboard die six months from
now. Keep that in mind, before going overboard with upgrades.

(Example of a recycled processor.)
http://cgi.ebay.com/PENTIUM-3-III-8...ryZ14292QQssPageNameZWDVWQQrdZ1QQcmdZViewItem

Now, another thing that comes to mind, is I don't see the mechanical
bits for holding the SECC2 processor in place ? There should be some
plastic to hold an SECC2 module upright. You don't want to rely on
the SECC2 connector to hold the processor in an upright position.
So if that is a real working SECC2 slot on your motherboard, it really
needs proper mounting hardware, so the processor doesn't fall over
and muck up the pins in the slot. Mine uses a different scheme than
is shown on this page.

http://www.tyan.com/support/html/secc2_install.html

Paul

 

[Daave]

There was a time, when the multiplier setting could be set via
jumpers or dip switch. But then Intel started locking the multiplier
inside the processor.

Paul, thank you *so* much for such a long and detailed reply. I printed
it and when I'm ready to stick in a faster CPU, I'll definitely use it
as a reference.

For now, I just want to explore what is already there and try to
determine what's going on. Specifically, I am trying to reconcile the
two different processor speeds. In the BIOS, the values are:

CPU Internal Core Speed: 233 MHz
CPU/SDRAM Bus Frequency: 66/66 MHz
CPU Core:Bus Freq.Multiple: 3.5x

The numbers add up, of course. However the core speed is less than half
of the 500!

And when I measure CPU speed in Windows, it indicates that it *is* 500
MHz (and Everett shows the correct 7.5 multiplier).

So far, someone has suggested that this Celeron-A processor is
"double-pumped," but I've researched this and doubt it is the case.

Is it possible that the person who built me this system merely entered
the wrong value in the BIOS. Would it be dangerous at all to change the
multiplier value in the BIOS to 7.5? (It is definitely possible to do
this; I'm guessing that Intel "top-locked" this particular
preocessor...)

I am very perplexed, to say the least!

 

[Michael Hawes]

My PC is 5½ years old. It's a barebones model with an EliteGroup
P6SET-ML motherboard with onboard audio and video and 32 KB primary
memory cache and 128 KB secondary memory cache. The CPU/SDRAM bus
frequency is 66/66 MHz. It has a 500 MHz PPGA (socket 370) Celeron
processor.

I've been looking at different things to tweak to eek out more power (I
know...not a lot by today's standards!).

One thing I thought was odd were the settings from the BIOS & CPU
Features Setup:

CPU Internal Core Speed: 233 MHz
CPU/SDRAM Bus Frequeny: 66/66 MHz
CPU Core: Bus Freq.Multiple: 3.5x

Okay, I know that 3.5 times 66 = 231 (close enough to the 233 listed).
But the speed of my processor is 500 MHz! What's going on?

See if you can set multiplier to 7.5. Use CPU-Z to check.
Mike.

 

[Daave]

There was a time, when the multiplier setting could be set via
jumpers or dip switch. But then Intel started locking the multiplier
inside the processor. This does two things. It gets around any issues
with the range of options offered by the jumpers. (Like processors
where
the multiplier goes up to 14X.) But it also prevents overclocking via
multiplier jumpers only. To overclock, you could try changing the FSB
clock,
with the downside of that being, that the old boards tied PCI and AGP
clock to the FSB clock, so the peripheral busses could end up
overclocked
past their limits. On modern boards, the PCI and AGP clocks are
"locked",
meaning they are independent and async to the other clocks in the
system.
On a modern board, PCI and AGP stay at the proper frequencies, while
you
adjust the FSB. On the older boards, they were still tied together.

(On the machine I'm typing on, I run at 100MHz, giving PCI=33 and
AGP=66.
If I bump the clock to 112MHz, the PCI is 37.3 and AGP is 74.6MHz.
Which
is a comfortable limit for the PCI at least. Some of the older AGP
cards
could be run at 100MHz on the AGP bus, so an older AGP card may be
less
of a limit, than proper operation of the PCI bus.)

Chances are, whatever you find as a substitute for your processor,
it will have a locked multiplier as well. So whatever the multiplier
is set at, you'll be able to continue to ignore it.

This page suggests a SECC2 form factor processor, as a faster
alternative.
Mainly because the S370 "PPGA" suggests to me, that more modern S370
processors may not be compatible with the socket. It is possible there
were some hardware hacks you could do at the socket level (maybe some
drilled pins and wired jumpers), to allow a Coppermine S370 to be
used.
But that is not likely to be a lot of fun, if this is your only
computer.
And it may turn out, that it is easier to find an SECC2, than the
right
S370 processor.

http://www.mainboard.cz/mb/ecs/P6SET-ML.htm

I think I can see where the 850MHz limit is coming from. If I look at
the Pentium III section of processorfinder.intel.com , and set the
selector to 100MHz only, the fastest SECC2 processor is
850MHz/100MHz/256KB.
That is a CPUID of 0683.

http://processorfinder.intel.com/List.aspx?ProcFam=25&sSpec=&OrdCode=
http://processorfinder.intel.com/details.aspx?sSpec=SL43F

If you were to buy a 133Mhz FSB processor, and run it at 100Mhz, I
think the top multiplier on those is lower than the 850MHz/100MHz
processor, so you'd be no further ahead. So at first glance, it
does look like the 850MHz/100MHz SECC2 might be your best alternative.
(At least, if I can trust that mainboard.cz listing as being correct.
You can hunt around ecs.com.tw for more info I supposed.)

It is also possible, you could plug a Powerleap processor, or a SECC2
slocket adapter with S370 socket on it. Yes, that does open a range of
new possibilities, up to 1.4GHz/100MHz level of S370 processor. But
some of the old BIOS had trouble with "high multiplier" processors,
and not all motherboard manufacturers, continued to produce BIOS
updates, to allow using devices like that. On my old board, I can use
a
1.4GHz processor, but I have to manually add microcode using CTMC, so
my processor is not completely supported by using the latest BIOS.
Still, my BIOS is in better shape than some other motherboards of
that era. Maybe your 850MHz/100Mhz SECC2 module would have microcode
support, and fit within the limits of the BIOS.

Wow! Thanks again for such a detailed and helpful reply!

I have been researching this a lot lately, only to get more confused.
:-( It appears that the highest multiplier (8) and the highest clock
speed (100) settings in the BIOS would suggest that the fastest
processor I could use is an 800 MHz one. Is this more a function of the
board or the BIOS itself? That is, does the board dictate to the BIOS
these figures?

I was thinking perhaps I should update the BIOS. But according to
http://www.ecsusa.com/downloads/p6set_ml.html , the newer features seem
to be ones I wouldn't be using anyway (concerning a Pentium II
processor, a daughter card, and a hard drive larger than 65 GB). I
suppose if I ever get a larger hard drive then it would behoove me to do
this.

Anyway, assuming that I want to replace my current CPU with an 800 MHz
Pentium III (Coppermine) processor, it looks like there are many types:

800/256/100, Slot 1
800/256/133, Slot 1
800/256/133, Socket 370
And I've seen some 370-style ones apparently mounted on a Slot 1 card.

Recall I currently have a 500 MHz Celeron (PPGA, Socket 370). There is a
bus speed limitation of 66 MHz. Is this a function of the processor
itself of the fact that it uses Socket 370? The reason I ask is that
some of the Pentium IIIs are *also* Socket 370, however they are FC-PGA
as opposed to PPGA (like the Celeron). Now, the Pentium III, has a much
faster bus speed, but will my board support it? That is, is FC-PGA
backward compatible? Can I plug it into the Socket 370? Might I still be
limited by a bus speed of 66 MHz? Sorry for all the questions; it's kind
of confusing to me.

Ah, I just reread your post; this must have been what you were referring
to when you mentioned the hardware hacks I could do at the socket level.
If so, I'm not prepared to go that far!

And since I am apparently limited to a bus speed of 100 MHz, does this
mean I should stay away from the CPUs that are 133 MHz? Or does it not
matter? If not, that would be good since those CPUs apparently are
selling for less!

Also I noticed that the voltages on these processors differ a bit (1.6
versus 1.7, I believe). I assume this shouldn't be a concern, no?

Thanks in advance, Paul.