Hi: I have a machine with a 1100MHz microprocessor (Celeron). However, it is clocked at 735MHz, according to the OS (it detects 735.020 to be precise, but the fraction must be accounted for lack of precission in the measurement).

It is clear for me 1100MHz is the maximum frequency the micro can ran at as given by the manufacturer's specs. But what about the 735MHz? If the micro can run at 1100 why make it run at a lower speed? Perhaps the chipset is the limiting factor? Perhaps the RAM is slow? These are all questions?

Celeron 1100MHz = 100MHz FSB x 11 *(clock multipiler)

735.02 / 11 = 66.82MHz

So instead of running 1100MHz with 100MHz FSB, its running 735MHz with a 66MHz FSB.

Bumping the FSB to 100MHz should get you 1100MHz. If its a celeron 1100MHz anyway, I've seen mistakes in packaging before with intel CPUs, and there were (and are) 'fake' CPUs around as well.

As for why its running at 66MHz FSB, not 100MHz- BIOS error, user error, mistaken CPU ID, jumpers not set, 'fake' CPU/motherboard. Could be any of those, or omething else I havent thought of.

Do you know what model motherboard yuo have? If yuo dont, post the output from this command (in 'code tags please, it keeps the forum neater)-

lshw

Make sure you run it as 'root'/su/sudo.

That will tell us the motherboard you have, and some other info that could help as well.

I have this datum most carefully written down: it's a M758LT-H. The manufacturer _seems_ to be PCCHIPS.

That could be bad news; there was a little-remarked scandal a few years back, with some PCCHIPS boards which wouldn't run at the correct frequency and PCCHIPS fixed the BIOS to lie about the frequencies, so it looked as if the clock frequencies were correct.

To what extent this applies to your board, as the case that I remember, concerned AMD processors, I don't know, but it is desperately underhand (and, possibly, illegal, depending on what they claimed for the board and your jurisdiction) and an organisation that would do that once, would probably do desperately underhand things in other circumstances, too.

@cascade9
Battery dead/low voltage, and BIOS losing settings, perhaps?

Guys, I don't believe it. Your several remarks about de 735/1100MHz stuff, made me go to the BIOS menu and see the CPU PlugnPlay entry. I simply change the multiplier freg to 'x11', CPU/DRAM Freq to 100/100MHz, and ... voilá, Linux when booting said:

Detected 1102.524 MHz processor.

Ain't that a funny thing? Bud luck just when the CPU fan blew up. Does any of you know how the OS can detect the CPU clock frequency? I'll bet you don't.

Alas! But why has the BIOS several multiplier freqs and several CPU/DRAM freqs. Perhaps because the same BIOS can run on different CPUs and RAMs? Am I not risking my motherboard by setting it to the maximum CPU freq?

If the CMOS battery is dead, the CPU speed should still go back to the normal settings in most cases.

Though its possible that if there was some BIOS problem, or if 'use failsafe values' (or similar) was selected in BIOS that the BIOS would downclock the CU from its standard 100MHz FS to 66MHz FSB.

I bet I do. :P

lshw

That will show you the current CPU MHz. If you have a CPU capable of changing its clocks, you can find the range of avaible frequencies with cpu-freq.

Finding info for the M758LT-H isnt fun, it seems to be a 'spanish speaking countries only' model (I only found results for that board in spanish). But it seems to be a socket 370 board, and pretty much all of them were made to be able to run 66MHz FSB (celeron), 100MHz FSB (P3 and celeron) and 133MHz FSB (P3). So yes, its setup to be able to run different CPUs.

As long as you have a real celeron 1100MHz, running it at the rated speed should give you no trouble.

I dont recall any major scandals over the way that PC-Chips (and others as well) would 'up' the frequency for the FSB but still have the BIOS report the standard speed- eg, CPU XXX should have been running a 100MHz FSB, but was actually running 105MHz FSB. Thats the only thing I recall happening to athlon setups, there was far worse with P3/P4/Celeron setups.

With the P3/P4/celeron 'fake' setups, the BIOS was flashed (and hardmounted so you couldnt replace the BIOS chip with a honest one) so that it would report a celeron XXX as a faster P3 or P4 CPU.

PC-chips were masters of that by the stage that the P3s came out, they were doing this sort of thing with 486s-

http://www.redhill.net.au/b/b-bad.html

Not all the PC-chips boards are bad, but some of them were really, really awful. BTW, I actually owned a PC-chips 'VX Pro+' with fake cache chips. One of the worst boards I've ever seen, and I've seen a lot of them.

@cascade9: thanks so much for the datum. But if you think you won the bet you are wrong. "How does the OS know" is the question. The CPU has an internal clock, completely independent of the external one. So by sampling one square wave with the other, a program can assert the external freq.

I had done a little program to measure f for the P1, and even found one written by Intel itself in the processor manuals, which extracted plenty of info out of the CPU. I do not want to know how much bigger the info must be today. So, you owe me.

Oh, so that wasnt an english mistake then?

AFAIK, the BIOS passes the CPU frequency (or avaible CPU frequencies) to the OS.

I dont think so.....BTW, if that was a joke, try adding a smilie, or j/k, or something to make it obvious. :P

You must excuse a passionate American movie goer. All my English comes from there and it is clear I many times think I understand the meaning when I don't. Regards.

No problem