OK, here are the preliminary results with those PC3200 TMSP BH-5 sticks. I can't push them really hard right now because I have switched to 3.3V line on my DFI (all that talk about alleged voltage spikes and the board killing RAM had me scared) so basically I have up to only 3.15V stable VDIMM.

Winchester 3000+ (0444 I think)
DFI NF4 Ultra-D, 510-2 BIOS, using Orange DIMMs for now.
TwinMOS TMSP400, 1A4T, 44D, BH-5/UTT
MyGenie DRAM timings:
2-2-2-8-1T-07-11-3-2-2-3-3120-1-Enable
Increase-200-Level 7-Level 3-7-Fast-5.0-016-Disable-16x-7x-Disable

First impression, I used to have CH-5/UTT versions of these same sticks, and the BH-5 ones seem to be able to hold 2-2-2-x timings with a lot less voltage. I had no problem using these new ones with 2-2-2-x at 200MHz with only 2.7V (even tried it on NF2 platform), the other ones needed about 3.0V to 3.1V to do that.

Stick 1:
2.70V .... 200Mhz .... Memtest 25 min pass
2.70V .... 210Mhz .... Memtest 25 min pass
2.70V .... 220MHz .... Fail
2.80V .... 220Mhz .... Memtest 20 min pass
2.80V .... 230Mhz .... Fail (freeze)
2.90V .... 230Mhz .... Memtest 20 min pass
2.90V .... 240Mhz .... Fail (freeze)
3.00V .... 240Mhz .... Fail (3 ers)
3.10V .... 240Mhz .... Fail (1 er)
3.10V .... 238Mhz .... Memtest 6+ hrs pass, Prime95 Custom Blend 11+ hrs pass

Stick 2:
2.70V .... 200Mhz .... Memtest 25 min pass
2.70V .... 210Mhz .... Memtest 25 min pass
2.70V .... 220MHz .... Fail
2.80V .... 220Mhz .... Memtest 75 min pass
2.80V .... 230Mhz .... Fail
2.90V .... 230Mhz .... Memtest 25 min pass
2.90V .... 240Mhz .... Fail
3.00V .... 240Mhz .... Fail
3.10V .... 240Mhz .... Fail
3.10V .... 238Mhz .... Fail
3.10V .... 235Mhz .... Fail
3.00V .... 235Mhz .... Fail
2.90V .... 235Mhz .... Fail
2.90V .... 230MHz .... Memtest 30 min pass, Prime Custom Blend 2 hrs+ (still running)

There you have it. I have had some problems with Stick 2 in the beginning. It wouldn't go any higher than 220 MHz. Raising voltage wasn't helping, actually it was making it worse. I was bewildered. I decided that I have a bad stick, so I moved on to testing Stick 1, and that one did excellent. The next day I went back to Stick 1, and now it passed 220 and 230, but no matter what I did it couldn't pass 235. Again, raising voltage from 2.9 to 3.0 and 3.1 didn't help move it up as little as 5Mhz. I might have to play with TREF, Drive Strength and Data Drive Strength values, as well as try Yellow slots to see if they make any difference. I am also planning to go to 5V line again (the RAM alleged RAM killing thing seems to have settled) and see how they scale with 3.2+ VDIMM.