eva2000

235bclk done in same session as my 32M Pi run @230mhz boot from bios, so setfsb was used after 32M pi finished to push to 235bclk.

eva2000

cstkl1, I asked DFI about the missing foxconn label from socket and backplate and it turns out DFI has two suppliers for the socket retention and backplate itself. One is Foxconn and one is Lotes. The actual motherboard itself is still the same manufacturer just some boards will have Lotes or Foxconn sourced cpu socket retention and backplate components.

eva2000

Both cstkl1 and my DFI contact suggested with 4/28 bios, I check out setting BOOT Up CPU Base Clock to around 180-185bclk for better perrformance and stability. Seems to work as first test has me reducing my required CPU VTT volts from 1.43v to 1.37v for 8x 32M Pi test!!!!





Bumped CPU VTT from 1.37v bios to 1.39v bios and IOH/ICH v1.1 from 1.11 to 1.13v for Linx stability

DrNip

Good lookin out eva2000. I was looking at that boot up clock yesterday wonderin about it. All I could figure is it would boot up at a lower blck then apply your full blck within windows. Is that what it is doing or is there more to it than that?

eva2000

AFAIK, on boot up it locks to what is set in Boot Up Base clock and sometime between boot up and initializing before windows boot it clocks up to set BCLK. Something to do with MCH straps according to cstkl1.

cstkl1

kekek eva keep somethings on the down low :P until the competition is over.. kekeke

eva i retested also on the usb issue. the 5v, 5vusb on a few things that might cause the lag

1. Having a higher powered DC fan on a corsair hx1000 molex connector together with the HDD/other peripherals.
2. The bios flashing i stated above is only on first brand new mobo flash.. that means u plug in and before u enter usually we assume its already on default. Nope still have enter the bios. Load optimized default on 1 GC. and then flash.

The 2nd one u will see the posting it will be long.. C1 then to something than back to C1 and then only will go ahead.

tested on one 920 D0, one 920 C0 and one 965.
now all of them to pass 2000mhz CL7 vtt must be set at 1.39
tested with Dk, and two UT's. the same. Dont care about the reported voltage on this mobo's its not accurate. the set voltage is the same. and yup one step of the ioh voltage for final stability on linx.
the one 920 is good with a sucky 920 C0 and 965 both i took from the office. so think maybe for this bios the myth of the golden IMC now can be questioned. until more ppl test. hoping gary, and the rest can share on this.

but for superpi32m with the 185 bclk up to 8 threads.. it will still work even if u reduce ure dram from 1.65v for example for the gt at 2000mhz CL7 to 1.635v and vtt at 1.37v set.

tested on the 1st UT and DK.

sourcing for the patriot LLK 2000mhz CL8 6gb kit.
recommended my bro to buy the 1600mhz CL8 from pccasegear and the quality is amazing. The pcb detail is exactly like the dominator variant.
and its cheap with xmp profile

DrNip

230 blck boot is wild. I can't get min e to boot past 220 blck. Got to keep on tweakin until I get it.

eva2000

hehe.. interesting about the lag.. no probs here with HD4870x2 CF so far anyway

keep at it..

First time using HD4870x2 CF on Win7 64bit RC OS and first time I've broken 30K mark in 3dmark06 and did it without even touching GPU/MEM clocks yet!



Personal best for 3D Vantage as well and with just stock GPU clocks!