Gigabyte EP45T-Extreme + E8500 C0 ES 5.4GHz

Push it to the max but make sure that you keep it cool!
Post Reply
Gouhan
Registered User
Posts: 516
Joined: 17 Dec 2002, 02:00
Contact:

Gigabyte EP45T-Extreme + E8500 C0 ES 5.4GHz

Post by Gouhan »

What a weekend!
Besides being impressed by the power of the 4870X2 I also realized how much I rely on Rivatuner and other 3rd party programs for overclocking dry.gif as fan control is a mission with the 4800 series. Besides the fan issues I'm even more excited to see what 4-way Crossfire can do with another 4870X2.
I had intended this to be my last session with DICE before I move to LN2 exclusively, but once again due to some ill preparation on my side not all runs were optimal. While I had perfect mount and near perfect sealing I only managed to run 3DMark05 when the system was on it's last legs. still impressive though considering that the last benchmark was done after 5PM and the system had been running at -65'C since 10:30AM in the morning. What made it take this log was flashing back to F3b bios, trying to find the right FSB/memory configuration which I never achieved but settled finally at the following settings:

Multiplier: 9X (5.4GHz)
FSB: 600MHZ
PCI-E: Auto (higher PCIe would have added performance but I forgot totally about this)
Graphics Robust Boost: Auto/Normal (Fast or Turbo would have increased performance slightly once again)
CPU Clock Skew: 0ps (anything higher is unstable)
MCH Clock Skew: 0ps (same as with CPU)
PCIe Clock Drive: 900mv (helped stabilize 3D over 700Mv)
CPU Clock Drive: 700mv (didn't try change it)

Frequency Latch/Strap: 400MHz (On X48/P45/X38 this is faster than 333 or any other strap. Also results in lower PL)
Memory Divider: 2.66D ("D" automatically selects 400 strap. Note!Lower divider puts more stress on the MCH which ends up relaxing Performance level, but higher divider like 3.33D or 4.0D may increasing RAM speed too much)
Memory Timings: CL7-5-5-19 2T is obviously best with 2.66D Divider with 600MHz FSB, but this changed from bench to bench)
Performance Level (Static Read Adjustment for both Channels): 10 (boot at 14 if CL7 is selected with 2.66D, 12 when CL8 is selected)
CPU VCore: 1.86V (True is 1.808V, load is 1.79V)
CPU Termination/VTT: 1.52 (tested as high as 1.66V)
CPU PLL Voltage: 1.72V
PCI/ICH: 1.6V (1.5V normal)
GTL Ref(s): NOT PRESENT IN DDR3 BOARD!

MCH Voltage: 1.44V (Tested as high as 1.62V)
VDimm: 2.28V
DRAM Termination: Auto

BIOS VERSION: F3b
Note about BIOS: F3b has high FSB, but RAM clocking not as good as F3e Bios, but I needed the FSB to get to the speed I wanted without sacrificing FSB

I have a better understanding of this motherboard than I have had before, so better results with same hardware and cooling at possible but I'm a little tired of DRY ICE right now so next session is LN2 or if it's Dry ICE again, it'll be X9650 with 2x 4870X2 or E8600 with 2x 4870X2 but I will not run ICE again with this exact configuration unless I have much to gain in any benchmark particularly in 01. More than the results which are nothing significant, the experience gained with the board is the biggest reward here.
Still believe more can be had form this motherboard. ;)

Image

Image

Image

Image
Image
I34z1k
Registered User
Posts: 8142
Joined: 05 Dec 2003, 02:00
Location: Dry Ice factory

Post by I34z1k »

Impressive man ;)

Top notch stuff as usual and long time no see :P
Post Reply