New system build - reboot loop when attempting to boot from SATA HDD

Char Jackson wrote:
On Tue, 18 Dec 2018 11:43:52 -0600, VanguardLH wrote:

I was looking through the data spec sheet at:

https://www.intel.com/content/dam/ww...heet-vol-1.pdf

to see what was the recommended normal operating voltage for Vcore or
Vcc. It says 1.52V maximum. Minimum is 0V (zero). Well, yeah, that
means unpowered. But you'd think they would list a recommended, normal,
typical, or optimal voltage.

At 1.1v, I guess I'm within that range. But yeah, why not just come
out with a recommended operating range so that a person would have a
starting point.

The processor has "multiplier min" and "multiplier max". On an
unlocked processor, the max isn't limited. Similarly, the processor
leaves the factory with a VID_min and a VID_max. The hardware
design basically "polices" controls that an enthusiast might
wish to hack. You can still override some of these things
(obviously on the K processors), but there aren't a lot of
"digital adder" implementations for exceeding the range
programmed into the processor.

It's not likely that "VID_max" = 1.5V in the VRM vid table.
It would not be like Intel to encourage knife-edge circuit
operation, for warranty support reasons.

Instead, an enthusiast motherboard may apply an offset, such
that the voltage is "VID_min + offset" to "VID_max + offset".
A tool which reads out only the VID field, doesn't tell
the whole story. You need an ADC converter, like in the
Hardware Monitor, to get some idea what the real value is,
as measured at the correct point (monitor pin) on the CPU.

On modern AMD systems, they have an entire measurement
subsystem, that measures stuff all over the processor
(dunno if tools exist to display this to users).

At one time, offsets were added manually with crappy little
home-grown circuits. Later, some VCore regulators were basically
"VRM non-compliant" in the sense that there were digital inputs
to specify an offset that might have run dangerously close
to the spec limit.

Some (compliant) VCore regulators actually have OVP, where
if they detect too much violation of the current regulated
voltage value, they latch off. When you manually offset those
by jamming current into the control node, jamming too much
turns them off. An enthusiast board isn't likely to be
designed around such a thing.

Enthusiasts usually deal in a "reliability limit", like
maybe noticing that long runs at 1.43 volts killed a certain
processor. And as a consequence, instead of dealing with
"absolute max suicide runs", most (poor people) users would
adhere to the herd wisdom on the "reliability limit" instead.

Reading some of these threads, suggests we're not getting
the best out of herd wisdom these days. Does anyone
still hang out at ExtremeSystems ? Over the years, sites
like that have had great primers on understanding
a generation of processors and their best
feeding and care. The people who destroy one $1000
processor after another, have done great service
for ordinary users :-) I hope those people haven't
passed on...

Paul