Hi folks! So I have now played with v.16beta01. Well, no chit chat, straight to the nitty-gritty:
The overclock used for all testing is as follows:
10x 300FSB (3.0GHz) @ 1.365Vcore
1:1.5 CPU : DRAM ratio (450MHz, or DDR2-900) with 5-5-4-15-2T timings @ 2.1Vdimm
I ran five SuperPi 1M tests and a single test each of SuperPi 32M and WinRar's benchmark (time to crunch 100M, plus speed). Single tests were simply due to time constraints. Apologies for the formatting; the forum didn't like spaces.
SuperPi 1M test results:
BIOS----------1.-------2.-------3.-------4.-------5.--------Avg.
v.12-----------19.125--19.093---19.078---19.078---19.078---19.088
v.13beta01--19.140--19.079---19.079---19.094---19.094---19.089
v.14beta04--19.172--19.141---19.140---19.157---19.140---19.146
v.14beta05--19.078--19.094---19.156---19.078---19.094---19.089
v.14final------19.094--19.110---19.125---19.094---19.109---19.104
v.15-----------19.188--19.110---19.156---19.109---19.125---19.130
v.16beta01--19.047--19.032---19.031---19.047---19.047---19.042
*Averages were obtained by throwing out the highest & lowest time, then averaging the remaining three times.
SuperPi 32M test results:
v.12-----------17min, 26.344sec
v.13beta01---17min, 26.594sec
V.14beta04 --17min, 21.391sec
v.14beta05 --17min, 29.859sec
v.14final------17min, 26.875sec
v.15-----------17min, 32.922sec
V.16beta01--17min, 26.844sec
WinRar benchmark:
--------Time to reach 100M------Speed
v.12------------2:35--------------642KB/s
v.13beta01---2:35--------------643KB/s
v.14beta04---2:36--------------641KB/s
v.14beta05---2.36--------------641KB/s
v.14final-------2.35--------------642KB/s
v.15------------2.36--------------638KB/s
v.16beta01---2.34--------------647KB/s
So, at least from my testing, v.13beta01 is very close in speed to v.12 and v.14beta04 is slightly slower than the other two. v.14beta05 seems to be on par with v.12 & v.13beta01, with v.14final coming in right in the middle. Bringing up the rear is v.15.
The winner, by leaps and bounds, appears to be v.16beta01 (relatively speaking, of course). I actually had to open CPU-z to verify that I dialed in my OC correctly. I could have sworn I put in an incorrect (lower) memory timing or something. Nope; it was all identical. Looks like abit hit it out of the park with this one! As a bonus, it increases support for 45nm CPUs.
Additionally, the FSB reset issue when you remove power that presented itself with v.14 is still present in v.15. Shutting down & powering back on fixes it (restarting doesn't), but annoying nonetheless. Remarkably, whatever strange bug this was, I could NOT get it to reproduce itself with v.16beta01!
This one appears to be a win-win all around folks. For the first time since I began this testing, I can safely say v.12 will not be immediately re-flashed. Horray!!
Quick Disclaimer - If I had time to run more 32M tests, the obvious anomalies may work themselves out, but I don't.
In any case, all of these are too close to see any real world difference, but in case anyone was curious and/or you enjoy benchmarking or getting all you can out of your system for DC projects, this is how each of them perform.