December 9, 2001
Hello All,
How is everyone? Well I went ahead and ordered a replacement processor, the 900Mhz, Morgan based, Duron for $125.00, from Charles at i Think Computers.
I received the processor on Tuesday December 4. The installation went smoothly and I have been testing ever since. From a stability standpoint I am very pleased with the new processor. From a battery life standpoint, which is the main reason for the upgrade, I am not so happy, read on.
In this update I am going to cover:
I placed the Disassembly and Installation process description in a separate page because it is so long. Go here for that information.
For the Benchmark data I used SiSoft Sandra Standard Version 2001.3.7.50. I also used Mad Onion's 3DMark2000 Version 1.1 and ran just the CPU test.
| Benchmark | 800 Mhz Duron | 900 Mhz Duron | 900 Mhz Batteries PowerNow! enabled (Auto mode) | 1.0Ghz Athlon (Tony's Machine) | OC 1.1Ghz Athlon (Tonys Machine) | OC 1.15Ghz Athlon (Tonys Machine) |
| CPU Benchmark | ||||||
| ALU (MIPS) | 2247 | 2519 | 1526 | 2800 | 3055 | 3193 |
| FPU (MFLOPS) | 1097 | 1243 | 763 | 1382 | 1527 | 1597 |
| CPU Multimedia | ||||||
| Integer MMX (it/s) | 4352 | 4911 | 2998 | 5491 | 6004 | 6271 |
| FP 3dNow (it/s) | 4998 | 5661 | 3452 | 6240 | 7442 | 7819 |
| Drive (NTFS) | 7630 | 8055 | ||||
| Memory Benchmark | ||||||
| ALU RAM (MB/S) | 398 | 436 | 427 | 401 | 456 | 482 |
| FPU RAM (MB/S) | 429 | 442 | 437 | 422 | 467 | 497 |
| 3DMark | ||||||
| CPU Speed | 85 | 89 |
As you can see. The 900 is, as expected, faster than the 800. The 1.0Ghz and 1.1Ghz Athlon numbers are courtesy of tone_man2001 (Tony).
Here are screen captures from the AMD CPU info utility (link to utility).
|
Before |
After |
On Batteries |
 |
 |
 |
If you really want to be jealous read on. One of our members through some really good personal contacts at AMD acquired a 1300+ Mobile Athlon processor. This processor is based on a newer core than the mobile Athlons currently available. To say the least the 1300+ is fast. Unfortunately this processor is not yet available to the rest of us.
| Benchmark | 900 Mhz Duron | 1300+ Athlon |
| CPU Benchmark | ||
| ALU (MIPS) | 2519 | 3594 |
| FPU (MFLOPS) | 1243 | 1800 |
| CPU Multimedia | ||
| Integer MMX (it/s) | 4911 | 7055 |
| FP 3dNow (it/s) | 5661 | 8142 |
Here is where things get really odd. The overriding reason that I bought the 900Mhz Duron processor was to gain PowerNow! support. I am happy to report that this new processor based on the Morgan core does support PowerNow!.
Before I delve into battery life with the new processor let me relate battery life with the original 800Mhz "Spitfire" processor. With the Original, R0102K5, bios I could get about 90 minutes of life out of the BP71 battery. When the WXP01K5 bios was installed my battery life jumped to about 120 minutes with the BP71 battery. Why I don't know. I think the new bios modified the cooling fan operating parameters. With the new processor and according to AMD's claims of 30% (link to AMD documentation) greater battery life I figured I should get at least 150 minutes on the BP71 battery. However as you will see that does not seem to be the case.
Under Win2K and WinME with the PowerNow! driver from here, this processor behaves, clocking up and down, just as the AMD documentation says it should (link to AMD whitepaper). With the PowerNow! driver installed there is a PowerNow! tab present in the Power Applet.
When running on batteries PowerNow! supports three (3) different options;
High Performance Mode -- No CPU Speed modification processor runs at max speed
Automatic Mode -- Processor speed varies based on load. Speed can vary from 550Mhz to 900Mhz on the 900Mhz Duron processor I have which agrees with AMD's specs for the 900Mhz processor.
Battery Saver Mode -- Processor runs at it's minimum speed which is 550Mhz.
So far so Good. Everything seems to be working great. I ran the speed benchmark numbers shown above and they seemed reasonable based on processor speed. Running AMD's CPUinfo and CPUid utilities showed that speed was indeed changing.
I made sure my BP71 battery was charged and began the first draw down test with PowerNow! enabled and in Automatic Mode. I got about 105 minutes (1 hour 45 minutes). That did not make sense. 105 minutes is less time than I got out of the old "Spitfire" Duron. I ran the test again and got about the same time. Now I was really confused. I decided to try a draw down with PowerNow! disabled to see what kind of time I would get.
With PowerNow! disabled I got 120 minutes (2 hours) on the BP71 battery. Something is very odd. I repeated the test 3 time and got about 120 minutes each time.
I then re-enabled PowerNow! support and repeated the exact same draw down tests with PowerNow! enabled (Using both Automatic and Battery Saver mode). Unfortunately my results did not improve. I still only got about 105 minutes out of the BP71 with PowerNow! enabled.
| 800 "Spitfire" | 800 "Spitfire" | 900 "Morgan" | 900 Powernow | |
| Bios | R0102K5 | WXP01K5 | WXP01K5 & R0107K5 | WXP01K5 & R0107K5 |
| Batter Draw Down Time (minutes) | 90 | 120 | 120 | 105 |
I am at a loss to explain this. Anyone got any ideas??
I have figured out the odd results. It turns out the the Battery Indicator is very inaccurate when PowerNow! is enabled. See Battery Results Revisited for revised battery draw down results.
I tried installing the FXA, R0107K5, bios to see if that made any difference with the PowerNow! support. It did not.
However, it did seem to solve a small stability problem that appeared with the WXP01K5 bios. With the WXP01K5 bios my machine was less stable (occasionally froze for no reason). With the R0107K5 bios I have not seen a freeze in 3 days of hard use. I would suggest that everyone who is having freeze problems move to the R0107K5 bios and see if that helps. It has definitely helped me.
As usual feel free to e-mail me with any comments additions or just to say hi.
Later,
Brad
