Atom N270 VID Documentation Discrepancy

[elesueur]

Hi,

I'm trying to ascertain what voltage the Atom N270 runs from at various frequencies.

The values read from the IA32_PERF_STATUS register do not seem to correlate with the values in the table on page 20 of document: 320032-001 - Mobile Intel Atom Processor N270 Single Core Datasheet.

More so, the data available at http://processorfinder.intel.com suggests that the N270 has a VID voltage range of 0.9-1.1V, but at the maximum frequency, the status register reads the VID as 0x24, which, according to the table in that document is 1.05V. At the minimum frequency, the status register reads the VID as 0x0F, which isn't even in the table.

Can anyone help me out with this?

[cdhooper]

Are you sure you are looking at the right bits in IA32_PERF_STATUS (and that they
are valid for the N270)? You might try to correlate the voltage you are seeing against
a board voltage sensor if one exists on your board. I looked at the VID table in
320032-001 and wow, what a pain to interpret! You might try searching for
the IMVP-6 specification.

You can tell just by looking at Intel's table that the larger the VID value, the
lower the voltage. After taking several data points, here's what I came up
with to convert from VID value to expected voltage:
Invert the bits, multiply by .0125 V and subtract .0875 V.

VID6 VID5 VID4 VID3 VID2 VID1 VID0
.80 .40 .20 .10 .05 .025 .0125

A few samples:
1.200 0011000 ^ 1111111 = 1100111 * 0.0125 = 1.2875 - 0.0875 = 1.200
0.550 1001100 ^ 1111111 = 0110011 * 0.0125 = 0.6375 - 0.0875 = 0.550
0.400 1011000 ^ 1111111 = 0100111 * 0.0125 = 0.4875 - 0.0875 = 0.400
0.300 1100000 ^ 1111111 = 0011111 * 0.0125 = 0.3875 - 0.0875 = 0.300

The value you are seeing, 0001111 is not likely:
0001111 ^ 1111111 = 0110000 * 0.0125 = 1.4000 - 0.0875 = 1.3125
In other words, off the chart.