This post is an offshoot from
here.
You can find a fuller description of my system
at SPCR. The essential components:
- Opteron 144
- Asrock 939Dual-SATA2
- 2x512MB PQI Turbo (TCCD on Brainpower)
- Seasonic Super Silencer 350 Rev A3
- Original Storm/G4
- 10W DDC
- Fedco 2-333 heater core
- Yate Loon D12SL
- Sunbeam Rheobus for both fan & pump
Something hit a wall at an unexpectedly low level. All claims of "stable" are based on hours of Memtest and/or Prime95. I've run other tests, but not at such length.
The current settings run the Opty with its 9x multiplier at 270FSB for 2.43GHz. This is with the voltage maxed out in BIOS at "1.45V" and CnQ enabled. Speedfan reports only 1.41V. If I disable CnQ, Speedfan shows 1.45V and can run 9x280 for 2.52GHz. RAM voltage is "high", with 3-4-4-8 2T timings.
To test for max mobo FSB, I reduced the CPU multiplier to 5x and the RAM with several dividers. These tests indicated stability at 270-280FSB but not 285-290.
Tests were run with various BIOS versions, including the OCWB betas and official Asrock versions. I can't report significant gains with any of them. Currently running 2.20.
I've experimented with greater cooling. OCs were found with pump & fan at top speed. In addition, I opened my case's bezel for a 4C improvement(!). I also added another 120 aimed at the CPU socket area and DIMMs. Neither of these steps added stability at 285FSB.
Okay, what else can I tell you?