[Info-vax] Intel junk...Kernel-memory-leaking Intel processor design flaw forces Linux, Windows redesign

[Paul]

Rene Lamontagne wrote:
> On 01/06/2018 2:07 PM, nospam wrote:
>> In article <nga25dh9b1h75o62g6hr03b4e9oliovai5 at 4ax.com>, Doomsdrzej
>> <dre at do.om> wrote:
>>
>>>>> The biggest problem in even considering a Tesla is that I live in a
>>>>> very cold climate which, since mid-December, has seen its temperature
>>>>> go no lower than -25c. In such a climate, the already poor range of an
>>>>> electric car is even worse and there are good reasons to believe that
>>>>> it wouldn't even start.
>>>>
>>>> the batteries are heated in cold weather and the cars start just fine.
>>>
>>> Are they heated through the use of a block heater or is there some
>>> other solution I'm not aware of?
>>
>> the batteries are heated and shortly before leaving, you can preheat
>> the cabin via a smartphone app.
>>
>>>>> There's also the fact that the computers
>>>>> within it, something which is essentially problematic for all cars,
>>>>> tend to go crazy when the temperatures are too low so the car might
>>>>> effectively become useless.
>>>>
>>>> nonsense.
>>>
>>> Do you live in a climate where -28c temperatures are normal? My
>>> Infiniti started perfectly the other day at such a temperature but the
>>> continued exposure to the freezing temperatures caused the computer to
>>> go nuts and essentially all of the lights within the dashboard lit up
>>> and the system disabled everything from the power steering to the 4WD.
>>> Once things warmed up a few days later, all of the lights as well as
>>> the annoying check engine light turned off. To say the least, I
>>> wouldn't trust an electric car in such temperatures.
>>
>> based on that, you shouldn't trust a *gas* powered vehicle in such
>> temperatures.
>>
>> many gas powered vehicles have engine block heaters because they won't
>> start in extreme weather.
>>
> 
> Computer chips do not suffer from cold weather, The reverse is true and 
> heat will be more of a problem, You have some other problem in your 
> Infiniti.
> This none starting of an electric car in cold weather is pure BS.
> If you want to see some trending take a look at Tesla's Highway 
> tractors, That will surprise you.
> 
> Rene

You have to simulate and close timing, at the stated temperature.

Computer circuits work, as long as the temperature range is "expected".

In the case of your desktop computer, download the Intel data sheet
and check the temperature range. That will tell you what temps
are "valid" for the product. If the product says 0C to 70C,
there is no particular reason it has to remain "sane" at -20C.
The circuit will rapidly heat up and enter the "sane" range,
but that doesn't prevent problems during the first few seconds.
You might have to press reset twice or three times, until
it gets up to 0C inside the CPU. But then, what happens to
the (relatively cool) DRAM ? How many seconds does that
take to heat up ?

When you do chip design, you set the simulator for PVT (process,
voltage, temperature). You set the simulator conditions for
whatever the boss tells you :-) Maybe my chip was -5C to 105C
or so. At -5C, strong process, high voltage, the process is
"fast", and hold time can be violated at a DFF. CMOS circuits
work faster at low temps, but it can affect the ability for
one circuit island to talk to the next. The wider you set the
temp range, the harder it is to close timing at the two extremes.
(This statement is true for the "most naive" interfacing method.)

The automotive companies know all this stuff, and they'll
be careful to set the PVT on their chip designs, to easily
pass -20C or colder operation.

If I had to guess, it could have been a mechanical issue,
like tempco causing something to shrink or expand,
causing an intermittent in an electrical connection.
Some car console panel problems can be intermittent,
even at room temperature.

One other area that can be a bit flaky, is clock
generators. They use a quartz crystal. The person
designing one of those, has to know the expected
temperature range too, and those are a lot trickier
to get right.

    Commercial:   0°C to 85°C.
    Industrial: −40°C to 100°C.
    Automotive: −40°C to 125°C.  <=== some car electronics go to 150C
                                      for under the hood stuff
    Extended:   −40°C to 125°C.
    Military:   -55°C to 125°C.

The "regular" CMOS at our fab, was not long term
stable at much over 135C. That means after 100,000
hours at 135C, you might expect doping migration
on the silicon die, sufficient to slightly affect
circuit operation. That's why we don't push those
numbers even higher. Pure Bipolar circuits love
heat, and should be able to go higher than that.

    "... Useful bipolar transistor action over the
     temperature range from -195C to 550C has been
     demonstrated for heterojunction bipolar transistors
     in the GaP/AIGaP chemical system. This represents
     the highest temperature at which useful bipolar solid-
     state..."

Now, nobody builds complicated chips with these
exotic technologies for your home computer. Suffice
to say, space craft could benefit from stuff like that.
The faster exotic chips go at 100GHz, yet we
don't see any Intel 100GHz CPUs :-) A CPU built
from such things, would burn up from the power it
would use. It's fun to dream though.

    Paul