"Subject: Mac Pro Overclocking (Utility)
Maybe that is interesting to you and your readers.

"ZDNet overclocking tool enhances performance of Mac Pro"
You'll achieve the best results with good memory modules. (Yes we tried it, because we we developed it.)
(I saw a comment from a user (apple forum) about a restart correcting the incorrect system time/clock issue when OC'd and asked if the OC settings survived a warm restart.-Mike)
Yes, after a warm reboot the (2008) Mac Pro runs with the oc-frequency and mac os synchronizes its clock to the new frequency. The result is that the clock is running accurate and the mac pro runs with oc-frequency. Rebooting with oc-frequency does work with the 2008 model (MacPro3.1.) but doesn't work with 1st generation (MacPro1.1).
The fact that the latest version of Mac Pro (Mac Pro 3.1) can be booted after overclocking also means that a higher frequency is possible under other operating systems. That is of particular interest because there has to date been no functioning overclocking tool for Windows. In contrast, the first-generation machine (Mac Pro 1.1) can be overclocked under Windows with SysTool. But that doesn't work with the current generation of Mac Pro (MacPro3.1). (Also see Problems resulting from overclocking that mentions Mac OS time problems after overclocking, Problem with standby (sleep) mode, after waking clock speeds are reset.)

To get the best oc-result you'll need good memory mdules. We did some tests with CSX-Memory with a poor oc-result.
Regards, Kai"

The above page has a link to another page on Downloading and using ZDNet Clock (OS X leopard required).
As with any overclocking, YMMV (I'd suggest having up-to-date backup of the drive in case of corruption and the usual disclaimers/risks on OC'ing apply) but if you have tried it, let me know the details (including system and ram specifics).
Update: Here's a reply from Oliver on his initial experience with it:

" Hi Mike, I did try it just now with my 2008 Mac Pro (dual quad-core Xeon 2.8 GHz with 4x2 GiB brandname DIMMs), and it seems to work. I gradually increased the bus clock from 400 MHz by 1 MHz increments (the FSB transmits 4 signals per clock, which is the reason it's usually called a 1600 MHz FSB). Note that the CPU runs at 7 times the bus clock in my machine, so by increasing the bus clock by 1 MHz, the CPU clock increases by 7 MHz.

So I kept increasing the bus clock in 1 MHz increments, waiting maybe 20 seconds after each increade to see if any memory errors appeared (as described in the ZDNet article), and continued. At 458 MHz bus clock (3207 MHz CPU clock) the machine panicked and I had to shut it down.

After restarting I again tried running it at 457 MHz bus clock (3199 MHz CPU) while running mprime (stress test linked in ZDNet article), and after about 10 minutes it panicked again.

So I'd have to choose something a little lower and run a CPU/memory stresstest for a day to see wether it's really stable. I'm not going to do that however. Maybe it would be stable at say 10% higher clockrate (3,080 MHz), but for what I do this doesn't matter. I'd rather run the system (especially the memory) at the speeds it was designed/tested to work well.

Plus the downsides noted in the ZDNet article turned me off, most notably that the computer's system clock (time) runs too fast (easily verified with my wrist watch). (I asked Oliver to check after a warm reboot, as the 2008 Mac Pros (but not earlier models) retain OC settings after a warm reboot and the system clock is corrected also. See replies to my question on this above from the OC utility developer. Oliver later confirmed this - see his later comments below.-Mike)

Sidenote: it's rather hot in my room ATM: 27.5 ˚C (81.5 ˚F).

During the tests the CPU front and exhaust fans did increase their spinning frequency, also the maximum power used (of the "Power Supply 1 Line 1" as per Hardware Monitor) increased from a maximum of about 310W at 2.8 GHz to a maximum of about 340W at 3.2 GHz with mprime running at full blast (8 threads, all CPUs busy).

I did not run any benchmarks. According to the ZDNet article, some of them (CineBench & GeekBench) rely on the internal clock and thus report more or less the same results with and without overclocking, but by using a stopwatch one'd see that performance does increase (I guess one'd have to multiply the results by the "overclock factor" to get proper results).
(the next day he wrote he had tested a warm restart)
Ok, I tried this, overclocking somewhat less this time: to 443 MHz bus frequency (3101 MHz CPU). Yes, OC survives a warm start, and yes, the system clock seems to run normally after that. It even says so in the ZDNet article (this page), but I didn't realize it until now. However, this only works with 2008 Mac Pros reportedly - older Mac Pros cannot be rebooted after using the OC utility. (confirmed by developer's comments above) I guess the system clock "reads" the bus frequency at boot time and then makes use of it, assuming it won't change. Hmm, so if I were to decrease the bus frequency (now that I've restarted the OC'ed Mac Pro) the system clock should run slow... *decreasing from 443 to 400 MHz bus clock* - Yes, the system clock does run slow now (after downclocking speeds after a restart).
Best regards, Oliver"

Another mail with some first impressions (updated w/later comments):

" Hi Mike, Saw the posting on the MacPro/XServe Overclocking Utility. Downloaded and installed. Results so far...

System Specs:
MacPro2,1 (April 2007) 8 core 3.0Ghz
5GB RAM (Apple 512MBx2 in upper riser card slot 1&2 and Crucial 2GBx2 in lower riser slot 1&2) X1900XT
10.5.3 on 1TB(2x500GB) RAID 0, XP2 on drive 3 and Vista Ultimate 64 on drive 4.

Before starting to use the utility i used smcFanControl to set all fans to 50% of their max speed. I then started to clock up in the suggested 10MHz increments with a rest period of around 2-3 minutes in between jumps. I saw one RAM error very early on, but nothing more until a Bus Clock setting of 370Mhz (3338Mhz Proc Speed). AT this level, the errors began in a cascading manner with more errors reported each time the log file was updated. I clocked back to 369Mhz (3329Mhz Proc Speed) and have had no further memory errors.

I must admit that i have just been surfing the net and doing a few e-mails. Fans are also still at 50% of max. I intend to test the system a bit better tomorrow and will report back with further info.

Only problem i have seen so far is the fast forwarding of time, as mentioned on the utility download page.... (see notes above on this - for 2008 Mac Pros a warm restart corrects the time sync and retains OC, but for 1st gen Mac Pros, OC settings are not retained after a restart.)
(Kevan later wrote)
Shortly after i mailed you i set the fans back to normal. Within 5 minutes RAM errors started to appear. I clocked backwards in steps of 10 but didn't manage to get the RAM errors to stop. (Apple system profiler's memory section will show EEC errors/memory status) Eventually got back to stock settings. Restart failed with flashing power light. Forced shut down solved this. I think i will hold out for future versions of this utility.
Best Regards, Kevan G.
www.fixmymac.co.uk"

There was also a thread in Apple's Mac Pro forums on this utility.

"Toast 9.0.2 w/ Blu-Ray/HD Authoring Plugin
I was able to burn HDDVD data to a DVD and it played correctly on a Toshiba HDDVD player. No problems noted. Not able to test the BR disk as I don't have a player to run it on. The source of the video was from a HD station via Eye TV 500 (MPEG-2).
-Robert
This is my set up
Mac Pro Dual 2.8 GHz Quad-Core Intel Xeon
OS X Version 10.5.3 (at the time the disk was created)."

I think this is the same Robert that sent the post in April on playing Toast 9 created HD-DVD disc images with OS X DVD Player.
(I considered posting a separate page with User reports on Toast 9 HD/BD Authoring Plug-in that included the all reports on this plugin.)