Q9550 w/ S2E - VTT set to min. and still IBT stable, what?

[Bobly]

Hey guys,



Well I finally gave in and accepted the fact I would have to go computerless while I RMA'd my ram (had been running with only one module for months because I didn't want to have the rig out of order for a week :P) and sent it off. It arrived back and while I haven't mem-tested it I'm passing stability tests so it's all looking good (I'll memtest it overnight when I get the chance).


Anyway, so I'm back running on 4GB of ram now which effectively means my 2GB of ram overclock experiences are pretty much useless so I started from scratch (ish)

Jumped straight to FSB 450 (stock 333) so an effective clock speed of 3.8GHz (stock 2.83GHz) at 1.4V in the BIOS (1.36V in CPU-Z). Stable as a rock so I figured I'd lower the volts before I pushed it any further...



Now this is where it gets interesting, I've been keeping track of my attempts over the last 4 days in a spreadsheet, and while I was able to lower the volts to the CPU and NB as expected (seeing as I went with 1.4V on CPU and 1.6V on NB/SB), I was surprised with how low I could get the VTT and PLL volts while remaining stable:



Each line is a new attempt, either IBT or P95 stress testing which is indicated in the last columns, -- means no data for the specific method during that attempt while | means repeating data (ie. I didn't run any IBT tests at first and I haven't changed the ram settings much). A red setting is one that caused instability while a green setting is the minimum I've managed to reach for that setting while remaining stable.



I've now reached the point where I'm effectively on the lowest possible volts for the CPU and NB while retaining stability but also the lowest volts possible on SB, VTT and PLL as those are the lowest the MB allows.


What does VTT do and how come while set to the minimum possible it's still IBT stable?

I also noticed in the last test that the CPU volts reported in CPU-Z went down while I didn't change anything in the BIOS, I'm not sure if it was in the last test or one of the earlier ones where I was modifying VTT so I've just added it as a note.
I'm going to Prime it overnight see if it truly is stable, though I would've expected IBT to fail 90% of the time tbh :/

[sabe]

I have a Q9550 E0, I also need *no* VTT increment to achieve 4GHz 100% IBT stability either.

[Bobly]

I thought it would've been important seeing as a/On Auto Everest reads it as being 1.28V and b/ it's directly linked to GTL_Refs which are supposedly crucial when it comes to fine-tweaking Quad-Core overclocks :/
I'm priming now (gotta love 4GB of DDR3, I prime while playing Team Fortress 2 sometimes, still get a stable 50-100 FPS :P) and if by the time I go to bed (probably another 6-7h) it's not failed I'll just leave it at that and start pumping up the CPU again.

[Leeghoofd]

IBT is not as good as people think it is mate... sometimes you can pass it with flying colours yet prime or co caps out in minutes and vice versa... best stability tool remains daily usage !

Nice CPU there