Tuesday, January 2, 2007

Overclocked at 3.2GHz

After being off for 2 weeks over the holidays, it was back to the salt mines for me.

Since my post yesterday, here are the changes I made to Nighthawk:

  1. Set CPU Frequency to 320 MHz (and verified that DRAM Frequency followed to DDR2-640 MHz.)
  2. Moved my Plextor PX-755SA SATA optical drive from the JMicron Controller to the Intel ICH8R Controller.
  3. Disabled the Nighthawk_VISTA RAID 0 Volume in Windows XP MCE 2005.

Here are my observations (I'll save my observations from overclocking to 3.2 GHz for last):

Moving Plextor PX-755SA SATA Optical Drive to the ICH8R Controller

Windows XP MCE 2005 now sees my PLextor PX-755SA DVD-RW optical drive!

Windows XP MCE 2005 also boots up significantly faster.

Based on this experience and my experience on Blackbird with my Samsung SH-S183L SATA optical drive connected to the JMicron Controller (Windows Vista Ultimate x64 (RTM Build 6000) does not recognize it as a writeable device), I highly recommend that you do not attempt to use a SATA optical drive with the JMicron Controller. It appears that this controller and/or its driver cannot properly handle SATA optical drives.

Unfortunately, I cannot connect the Samsung SATA optical drive to the ICH8R Controller on Blackbird since I already have 6 disk drives connected to it. I will probably be moving my Memorex IDE optical drive from Nighthawk to Blackbird.

Although the Plextor PX-755SA Motherboard Compatibility chart does not identify the ASUS P5B Deluxe WiFi-AP as being compatible, they are, indeed, compatible as long as you connect the Plextor PX-755SA optical drive to the Intel ICH8R Controller.

For additional helpful information about the Intel ICH8R Controller and the Intel Matrix Storage Console, check out Hiker's post in the ASUS Forums. (Highly recommended)

Disabling the 2nd RAID 0 Volume in Windows XP MCE 2005

Why did I want to do this? I am reserving this 2nd RAID 0 Volume for my test Windows Vista Ultimate x64 (RTM Build 6000) installation on Nighthawk. I have read accounts of Windows XP thrashing Windows Vista installations in a multi-boot environment. I simply do not want to take this chance.

I also wanted to know whether the Intel Matrix Storage Console in Windows XP 2005 MCE would have a problem with disabling one of its RAID volumes in the OS. I'm happy to report that the Intel Matrix Storage Console was still able to "see" the disabled RAID 0 volume.

The Nighthawk_VISTA RAID 0 Volume is now ready for my test Windows Vista Ultimate x64 (RTM Build 6000) installation. Now, if Nvidia would only release a Vista x64 driver for my BFG 8800GTS graphics card!

Overclocking to 3.2GHz

I ran Memtest86 overnight for 9 hours and 48 minutes at Nighthawk's new overclocked speed of 3.2GHz. I'm happy to report that no errors were reported after 19 passes.

Tonight, I will run Johnny Lee's SP2004 ORTHOS to further validate the stability of Nighthawk at this new speed.

The first test I ran was the Everest Ultimate Cache & Memory Benchmark. All results exceeded the 3.0GHz overclock benchmark.

Memory latency now matches the 71.3 ns observed prior to overclocking (compared to 76.1 ns at 3.0GHz overclock) and the memory read result (6469 MB/s) is closer to the 6781 pre-overclocked result (compared to 6178 MB/s at 3.0GHz overclock.)

However, the temperature readings bothered me. The motherboard temperature was now 3C higher at 3.2GHz overclock versus 3.0 GHz overclock.

The CPU temperature was 5C higher at 3.2GHz overclock versus 3.0GHz overclock.

By the way, I am using the stock HSF (Heat Sink & fan).

I read Tom's (Corn) post in the ASUS Forums about his overclocking experience and the effect of PECI (Platform Environment Control Interface) settings on temperature readings. He observed that temperature readings were lower when PECI is enabled in BIOS. I have PECI disabled in Nighthawk since I do not have any case fans connected to the motherboard, so I thought I would try changing it.

Sure enough, my CPU temperature went down to 39C from 49C. That's an incredible drop in temperature by just enabling PECI.

But, is the CPU actually running cooler? I'm not sure. Ultimate Everest shows both CPU cores still running at the same temperature (55C) with PECI enabled or disabled. Does anyone have an explanation for this anomaly?

I'll run the rest of the benchmarks tomorrow.

1 comment:

Michael Auslander said...
This comment has been removed by the author.