July 23, 2006
Hello all,
I have been continuing to pursue the battery life issues with PowerNow! enabled and disabled on my FX210. Let me recap what has transpired so far.
In December I purchased a new 900Mhz mobile Duron processor based on the Morgan core from Charles at IThinkComputers Inc. . I purchased this processor in the hope that AMD's PowerNow! feature would work and extend the poor battery performance of these laptops. I originally reported in December that I seemed to get less battery life, 105 minutes, with PowerNow! enabled than without PowerNow!, 120 minutes. link to previous post
The procedure I am using to measure battery life is as follows.
Laptop Power Settings
- Dim Display backlight to dimmest setting
- Disable Screen Blanking
- Hard Drive shutdown after 3 minutes
- System Standby Never
- System Hibernate Never
- Critical Battery Alarm set at 1% and set to Hibernate at activation
Battery Draw Down Methodology
- Scheduled task to load Yahoo web page.
- Scheduled task to do a DIR of C:\ Drive
- Scheduled task to record current time to text file.
All three (3) tasks are set to activate every 5 minutes. The third task makes use of the Win2K resource kit NOW utility to record current time to text file. This lets me run the test in an unattended manner and simply compare the first and last time stamp to see how long the laptop lasted. I know the test are not terribly real life or comprehensive but they are nicely repeatable which is very important.
Laptop Settings
- Bios -- R0107K5
- OS -- Win2K
- Ram -- 256MB
- Hard Drive -- Original 10GB drive
- Battery used -- PCGA-BP71 (2600mAh)
One thing that I have learned is that the battery life indicator is very inaccurate when PowerNow! is enabled. That is why my original tests had such odd results. In my original test I was only drawing the battery down to an indicated 10%. In my subsequent tests I ignored the battery indicator and let the laptop run until it shutdown from a dead battery. This created some additional problems because Win2K does not crash gracefully all the time. So I decided to run the laptop down to an indicated 1% and call it good.
Results of Battery Draw Down to indicated 1%
PowerNow! Setting PowerNow! Mode Run Time (minutes) Enabled Max Battery 140 to 150 Enabled Automatic 140 to 150 Disabled 140 to 150
Yep. The table is accurate. PowerNow! does not make a difference on my laptop. Yes, PowerNow! was functioning. During testing I ran the WCPUCLK utility from H.Oda! to monitor processor speed real time. With PowerNow! set for Max Battery the reported speed is 550Mhz. With PowerNow! set for Automatic mode the processor speed floated from 550 to 900Mhz based on demand. According to AMD's description of PowerNow! this is exactly what I am supposed to see.
I do have reports from people with the new FXA laptops that have gotten PowerNow! working. These people are reporting run times of up to 90 minutes with the wimpy PCGA-BP1N (1700mAh) battery.
Why don't I see a difference in battery life with PowerNow!? I am not sure. I do know that the FXA laptops have a new version of the Via KT133A Southbridge chip. I have a hunch that the FX210/FX215 laptops have the VT82C686A Southbridge chip while the FXA laptops have the VT82C686B Southbridge chip. According to chipset data on the Via web site the VT82C686B Southbridge has support for ATA100 and Second Generation PowerNow! while the VT82C686A has support for ATA66 and no mention of support for PowerNow!.
We know that the FXA laptops support ATA100 because the R0107K5 bios from the FXA laptops try to use ATA100 with ATA100 hard disks in our laptops. Jeremiah reported this problem when he first tried to use the R0107K5 bios in his FX210/FX215 with the IBM Travelstar 30GB drive. Therefore, I think that the FXA must use the VT82C686B Southbridge while the FX210/FX215 use the VT82C686A Southbridge.
Conclusion
It may be that the FX210/FX215 laptops do not properly implement PowerNow! and that is why I don't see a battery life increase with PowerNow! enabled.
If anyone else using the 900Mhz Duron or the 1.0Ghz Athlon has different results I would be very interested in seeing them.
Later,
Brad
