Been reading this thread for the first time this morning. My answer might be long, I'll be writing it as if I'm taking notes.
I am a PCB designer for a well known telecom company (my location might help you there), I know these datasheets wordings and such. I will try to help out with the interpretation of the datasheet (refered to as DS from now on).
1- I would not try to go 100MHz with the 101 setting. On Page 3 of the DS, Note 2 says this would put you in Test Mode. Nobody wants that.
2- The next pages show the register description of the device. This is set through frimware/software (bios/os) via the SMBbus interface. If you jump to Page 8, Data Byte 6, Bits 0-1-2, you will notice that these are the frequency select bits. They are READ ONLY. Meaning that no external program will be able to change those on the fly. It has to be through the pins.
3- Figure 1-2-3 on Pages 9 and 10 are the key. That is because the clock frequency pins FS_A, FS_B and FS_C are multipurpose pins. Their "setting" is only sampled in State 2 (S2) of the state machine, VTT_PWRGD# has to go low for this to happen.
This means you can't go and hold a wire to the pin. The wire has to affect the pin during sampling time.
4- All three FS pins are multipurpose. FS_A is the scariest, it turns to an output after it has been sampled. If by mistake you try to wire this pin directly to VDD or VSS (GND), when the output starts to drive it might blow, sourcing too much current. FS_B and FS_C are inputs all the time, safer.
5- We want to change FS_B from a 0 value to a 1 value. Because of my number 4 here, you CANNOT tie FS_B with FS_A.
6- Since FS_B is multipurpose but its second purpose might be unused then it might be ok to override its default 0 value. It is very likely that this pin is pulled to VSS. Pulled means using a resistor in series, something like 1k or 3.3k from the pin to VSS.
(If someone could measure the impedance between pin 16 and 29 we would know the value of the pulldown resistor).
7- If the pin is pulled down by a resistor, to inject a 1 value and override the 0 you need to pull harder, a lot harder. You could try to wire pin 16 directly to pin 1. If the pin was not pulled with a series resistor, this will short power and ground. Not good.
(Note that no intelligent board designer would connect an input signal directly to power or ground. There's too many disadavantages to this and longterm reliability would be affected.)
8- The chip is a TSSOP package. With the right skills and tools it is easy to lift a pin, I've done it countless times. It then becomes easy to solder to the lifted pin, giving it the value we want. I would prefer that approach. I'd probably grab a resistor from 1kohms to 10kohms, whichever you have. And I'd wire the resistor from the pin to VDD (pin #1).
(To lift the pin: Use your soldering iron, heat up the pad on the pcb. Use a tool with a "hook", or make a hook with a wire. Lift pin. Careful of shorting the pins next to it.)
(When I do this, it's in our lab, with microscope, dentist-like picks. Yes I've shorted countless pins, but the tools allow me to fix and view each time).
Conclusion: It is not easy if you dont have access to the right tools. I carry my laptop with me to work every day, it'd take me at most 15 minutes to try #8. Unfortunately I bought it in September and don't want to void my warranty.
I will try to answer any questions you guys might have.