So maybe it only affects ivy-e cpu's ?
Methods I've tried so far to get it working right:
Set cpu multi to 38
(Board powers off automatically after bios save)
Set cpu multi to 40 + set ram div lower at the same time
(Board powers off automatically after bios save)
Set ram div higher
Done, bandwith is back to normal
Alternate fix, for when the board doesn't power it's self off for a cpu re-init.
1. Set cpu multi from 44 to 42.
2. Manually Power off.
3. Set ram div to 2133 and cpu multi back to 44.
4. Manually Power off.
5. Set ram div back to 2400.
Done, bandwith is good again.
Now here's one that didn't work. (edit: I just realized I may not of did this right anyways, I didn't switch to cpu back UP and the mem Down at the same time)
Set cpu to 42x
Reset no power off (save and exit bios reset...)
Set mem to 2133
Reset no power off (save and exit bios reset...)
Set cpu to 44x and mem to 2400
Reset (save bios settings...) then manaully power off after the restart
Done, still bad bandwith...
Bios'es I've tried and I know it affects: 0403 (stock), and 0507.
I'm currently on 0507, I haven't gotten around to try 0601 yet.
I'm pretty sure that new one is affected too, I just don't feel like taking bios pics and re-applying my settings to a new bios right this sec (would have to port over my 100mhz and 110/114mhz configs).
The new one is essentially meant for 3930k's anyways.
And for some odd reason my bios isn't powering off when I change my cpu multiplier anymore, not sure what that's about but I kinda prefer it this way.
It just threw me off a bit on my bandwidth fix (hence the alt fix above).
But anyways, when the bandwith is good, it stays that way even during reset, but gets bad again when you do a power off.
And from there you have to repeat the fix.
Now for some reason alot of people aren't affected, maybe it's because they are using a diff strap from 100mhz I don't really know.
Like for example, tistou77's 2600+ screenshot above.
Everything seems normal there, then someone like me for example comes around wondering what is going on, so people start suggesting stability issues, timings or voltages.
And it's none of those.
I posted a set of 6 different screenshots on the XS forum, showing the differences between the 2 in memtweakit, not a single timing difference.
@tistou77
Sounds you managed to find the bug too, so what do you think about it?
I know some people are gonna think, oh it's nothing it's just a benchmark.
Well to me it's important :).
It's like a 30% bandwith loss.