tools/power turbostat: print Watts

tools/power turbostat: print Watts

Intel's Sandy Bridge and Ivy Bridge processor generations support RAPL (Run-Time-Average-Power-Limiting).
Per the Intel SDM (Intel® 64 and IA-32 Architectures Software Developer Manual)
RAPL provides hardware power information and control via MSRs (Model Specific Registers).
RAPL MSRs are designed primarily as a method to implement power capping.
However, even if power capping is not enabled, the RAPL regsiters
are useful for monitoring system power and operation.

Turbostat now displays the information provided by reading RAPL MSRs.
As always, turbostat never writes any MSRs.

turbostat's default display now includes Watts for hardware that
supports RAPL:

[root@sandy]# turbostat
cor CPU    %c0  GHz  TSC    %c1    %c3    %c6    %c7   %pc2   %pc3 %pc6   %pc7  Pkg_W  Cor_W GFX_W
          0.07 0.80 2.29   0.13   0.00   0.00  99.80   0.43   0.00 0.72  98.16   3.49   0.12  0.14
  0   0   0.14 0.80 2.29   0.12   0.00   0.00  99.74   0.43   0.00 0.72  98.16   3.49   0.12  0.14
  0   4   0.04 0.80 2.29   0.22
  1   1   0.06 0.80 2.29   0.08   0.00   0.00  99.86
  1   5   0.03 0.80 2.29   0.10
  2   2   0.17 0.80 2.29   0.14   0.00   0.00  99.69
  2   6   0.03 0.79 2.29   0.28
  3   3   0.03 0.80 2.29   0.07   0.00   0.00  99.90
  3   7   0.04 0.80 2.29   0.06

The Pkg_W column shows Watts for each package (socket) in the system.
On multi-socket systems, the system summary on the 1st row shows the total.

The Cor_W column shows Watts due to processors cores.
Core_W is included in Pkg_W.

The optional GFX_W column shows Watts due to the graphics "un-core".
GFX_W is included in Pkg_W.

The optional PKG_% column shows the % of time in the measurement interval that
RAPL power limiting is in effect.

Note that the RAPL energy counters have some limitations.

First hardware updates the countesr about once every milli-second.
This is fine for typical turbostat measurement intervals > 1 sec.
However, when turbostat is used to measure events that approach
1ms, the counters are less useful.

Second, the energy counters are 32-bits long and subject to wrapping.
For example, the counter increments in 15 micro-Joule units on my
local server, and the part could (in theory) consume energy at
its TDP specification of 130 Watts.  Here the 32-bit Joule counter
coult wrap as soon as 8 minutes.
Turbostat detects and handles up to 1 counter overflow per interval.
But when the measurement interval exceeds the guaranteed
counter range, we can't detect if more than 1 overflow occured.
So in this case turbostat indicates that the results are
in question by replacing the fractional part of the result
with "**":

Pkg_W  Cor_W GFX_W
  3**    0**   0**

Third, the RAPL counters are energy (Joule) counters -- they sum up
weighted events in the package to estimate energy consumed.  They are
not analong power (Watt) meters.  In practice, they tend to under-count
because they don't cover every possible use of energy in the package.
Also, the accuracy of the RAPL counters will vary between product generations,
and between SKU's in the same product generation.

turbostat's -v option now displays per-Package Thermal Design Power (TDP).
This is the specification for the part's maximum power consumption.
eg. on a 2-package SNB-Xeon system:

cpu0: 130.00 Watts Pkg Thermal Design Spec
cpu8: 130.00 Watts Pkg Thermal Design Spec

Finally, turbostat's -R option enables decoding and output of all RAPL registers
on turbostat startup.

Increment turbostat version number to 3.

Signed-off-by: Len Brown <len.brown@intel.com>