Charles Perreault had this to say about the incorrect voltages that are so common when using a desktop board with a Turion:
Hello,
In you article "AMD Turion 64 on the Desktop" (
http://www.silentpcreview.com/article300-page1.html), I noticed your hypothesis about the 1.3V of the voltage core of the turion MT where it should read 1.2V :
> At this point it is impossible to know exactly why the discrepancy occurred, and why it seems to be so widespread. There are a number of possibilities:
The MSI board is detecting the stock voltage correctly, and the voltages specified by the OPN are being interpreted incorrectly.
AMD has changed the stock voltage without updating the OPN for our processors.
The MSI board and many other boards are misinterpreting the CPUID string on the CPU, and thus applying the incorrect voltage.
None of these hypothesis is correct, or the 3rd is partly correct. The MSI board detects the voltage ID correctly (which is 10 to get 1.2V) using the CPUID integer (not the "AMD Turion(tm) MT-XX" but rather an hexadecimal ID) in a table called the Power-States Table located in the bios. When a motherboard does not seem to support the powernow! feature (cpu locked at maximum or minimum frequency, like the DFI Lancaster in your article), it's because this table (PSB/PST) is absent (which is common) or does not include the turion cpuids (which is even more common). This is also combined with a bad implementation of ACPI since the Turions/AMD64 frequency scaling should be done via ACPI nowadays, not via the PSB/PST legacy mode. Unfortunatelly, many laptops and almost every desktop motherboards have a bad ACPI implentation.
The voltage ID is then used by the bios or powernow! drivers to calculate the core voltage using the following formulaes :
1.550V - voltageID * 0.025V for desktop AMD64 cpus
1.450V - voltageId * 0.025V for mobile turions cpus
The fact is that the bios from every desktop board I saw is using the AMD64 formula for EVERY cpu, including turions. This results in core voltage always 0.1V higher than specified, at least for turions, exactly like the MSI board did when you tested it.
By the way, it is possible to get Powernow! working with every desktop board supporting turions (even the DFI one) under Linux. I created a patch of the powernow-k8 driver hardcoding the frequency / voltage ids for many turions that bypasses the bios. Also, patching the formula for calculating the core voltage can be done in the powernow-k8.c file of the linux kernel, resulting of perfect powernow! support with the MSI motherboard.
So, the answer to the bad core voltage is simple : the bios / powernow! driver uses the bad formula to calculate the core voltage even if it detects the right voltage ID.
Regards,
Charles Perreault
------------------------------------
M.Sc Computer Science
University of Sherbrooke