Earlier this week, we published an article comparing the Pentium M 760 to the Turion 64 ML-44. We have since discovered two problems with the tests, and we’ve revised the article to fix these issues.
First, due to an unusual confluence of events, the picCOLOR and WorldBench tests on the Pentium M 760 were inadvertently run without SpeedStep power management enabled. These tests have been rerun and the graphs updated with the new scores for the Pentium M. That means the revised article has new results for each of the WorldBench component tests, whether there has been significant change or not. The updates change the margin of victory in some cases, and in one case (the Worldbench Photoshop benchmark) the winning and losing positions swap between the two CPUs. Perhaps most notably, the WorldBench overall score moves from a tie to a three-point win for the Turion 64 ML-44. We have modified our benchmark commentary and the article’s conclusions based on these new results.
To ensure the accuracy of the other benchmark scores, we ran all of our benchmarks again on the Pentium M 760. No further abnormalities were found.
Second, it was discovered that the MSI motherboard used to test the Turion 64 ML-44 incorrectly configures the upper voltage range for that processor at 1.5V instead of 1.35V. As a result, the “load” power consumption numbers for the Turion 64 were five watts higher than necessary. We’ve updated that graph, as well.
We originally discovered these issues approximately 24 hours ago, but we wanted to make sure we had everything corrected and our facts straight before publishing an update. Our efforts were somewhat hampered by the fact that AMD has literally no technical documentation on its Turion 64 processors on its web site at the time of this writing, making it impossible for us to verify the maximum voltage of the ML-44. We wound up having to contact an AMD PR rep, who got back to us with the official word. Once we had the official number, we used RMClock to artificially lower the maximum voltage to 1.35V for testing, so we wouldn’t have to wait on an updated BIOS from MSI.
We strive for accuracy here at TR, and when our efforts in that area fail to meet our standards, we strive to correct our mistakes as quickly and completely as possible.