Welcome to the VMware Community formus. I think you may be the first to post a question in the forums about ESXi on an overclocked CPU. 
What does the CPU speed show on the summary page in the VI client for the host and have you run a CPU speed test within the VM or are you just looking at system info?
There's always one isn't there - and it sounds like it might be me
I'm actually the owner of an enthusiast technology publication, so as such overclocking - even on servers; is something that HAS to be done 
We currently run two Windows 2k3 Servers - both of which are Overclocked from 2.4ghz to 3.8Ghz and have been running stable for the past 2 years. The i7 965 is the latest addition to our mini-farm and with 12GB ram in it we thought we should give ESX a try :smileysilly:
To answer some of your questions, at the ESX console it mentions the CPU as being "Intel blah bha @ 3.2Ghz". This was somewhat expected as you rightly said it is read directly from the CPU and is similar to the what procinfo and Windows device manager would show.
However, I was quite surprised to see the Summary tab in the Infrastructure client reading 4x 3.207Ghz and also when provisioning CPU resources, the maximum available Mhz was ~1200Mhz. I also installed CPU-Z from within a virtual machine running Windows and that also reported the CPU speed as 3.207GHz
I know most people who use ESX probably don't overclock and i'm certainly not here to cause a stir, I'm just the inquisitive type.
Interesting ... I think the clue is the i7 processor here.
vmkernel calculates the CPU speed at boot using the timestamp counter register (TSC) embedded in the processor and seeing how many cycles that increases over a given period.
If you check out the processor manual () in 18.20.5 it says
The TSC, IA32_MPERF, and IA32_FIXED_CTR2 operate at the same, maximum-
resolved frequency of the platform, which is equal to the product of scalable bus
frequency and maximum resolved bus ratio.then later
• If XE operation is disabled, the maximum resolved bus ratio can be read in
MSR_PLATFORM_ID. It corresponds to the maximum qualified frequency.
• IF XE operation is enabled, the maximum resolved bus ratio is given in
MSR_PERF_STAT, it corresponds to the maximum XE operation
frequency configured by BIOS.
XE operation of an Intel 64 processor is implementation specific. XE operation can be
enabled only by BIOS. If MSR_PERF_STAT is set, XE operation is enabled. The
MSR_PERF_STAT field is read-only.
What this is saying is that the TSC isn't incrementing based on the core clock frequency of the processor, but calculated from the from the "(scalable bus freqeuncy) x (maximum resolved bus ratio)". It also says if XE mode is disabled (I think that has to do with the Extreme models), the maximum resolved bus ratio is fixed and corresponds to the maximum qualified frequency - i.e. the frequency written on the box you bought it in
So, maybe your BIOS has some options that can turn on XE mode and make the TSC run faster - we should then detect a higher speed.
Your tool on Windows is probably not using the TSC to measure the processor frequency, but a more fine-grained timer. I'd say you see the 3.2Ghz when running your CPU speed tool inside a VM because the monitor is also emulating based on the TSC counter.
This is kind of theoretical - it only relates to methods software can use to guess and report processor speed. If you run a benchmark that is actually doing work, I bet you'd see it run faster with overclocking on.