Andreas,
It's an age old pheonema (sp) that running multiple dimms while overclocked greatly decreases the max FSB you can attain on BX boards.
With a single 128MB cas2 PC100 Infineon dimm, and BH6 1.1, I was able to run 142 FSB cas 2-2-2, and 150 FSB cas 3-2-2 rock solid (600e @ 900 cB0). Rock solid even with a case temp (in that miserable small place where I was living) as high as 50 C. Stick in a second 100% identical dimm, and I can only do 137 FSB cas 2-2-2, and 142 FSB cas 3-3-3. Note that I was unable to do 155 FSB stable at all with a single dimm, regardless of settings (the board seemed to have reached its limit at 150 for that dimm, although 155 FSB MIGHT have been possible with a PC133 cas2 dimm, as long as it was 128 MB and not 256).
When I went to a 256MB infineon dimm (briefly had a Micron PC133 cas3 dimm, that did 150 cas 3-3-3, although needed a case fan or Unreal would sometimes bomb with REALLY weird errors while loading, but I wanted cas 2, dangit), I could do 144 FSB cas 2-2-2 with case temp at 55 C (!), and 150 FSB cas 2-2-2 just fine, provided I blew a fan on the motherboard (it wasn't the RAM at fault here, as switching to cas 3-3-3 made no difference). 155 FSB was unstable regardless of timings.
I've heard from others (including, I believe, the author of burnBX), that a 128 MB dimm is more stable than a 256MB dimm, when overclocked, although I don't remember why. That might explain me being able to do 150 cas 3-2-2 on the 128MB dimm with my case as a raging inferno of heat, while the 256MB dimm required a fan for stability. 144 FSB with the 256MB dimm was as stable as granite, even when I tried heating everything up past 55 C with direct sunlight in summer, case fan off
Likewise, on my 1.4-S Tualatin, I can do 150 FSB somewhat stable (Prime95 seems to either run for a very long time or it gives a roundoff error after awhile--roundoff errors are almost always RAM related, lockups or illegal sumouts are CPU), with Kingmax PC133 256MB cas2 dimm, 3dmark likes to sometimes quit for no reason, but UT2003 runs forever, BF1942 has lasted a few 1.5 hour sessions without dumping to desktop since I put vcore at the powerleap default of 1.55, etc, and Serious sam 2 hasn't crashed YET, with the vcore bump to 1.55 (I had it at 1.50 originally). 144 FSB is as stable as granite. 150 FSB isn't QUITE as stable as it was on my 1000EB @ 1125; I suspect either the powerleap adapter or the CPU is putting more load on the memory controller or causing more crosstalk than the MSI 6905 2.3 did, with the 1000EB; on a different board, I might be able to do 155 FSB easily. The CPU is clearly capable of more than 1575 mhz.