Not all that is happening is at the kernel level - there is more and more cruft 
being done at the 
microcode level (some due to threats of the folks in DC that want to be able to 
compromise any and 
all computers (like this will never blow up in their faces (idiots))).  This 
has the effect of 
making run time of even assembly code non determent.

Most likely powering down a peripheral chip will have a way to be disabled, but 
it needlessly 
increases complexity.  I suppose in the end the market place will punish those 
that do stupid things 
- sadly that does not apply to the wizards in the government.

A friend of mine was writing assembly code for an ARM chip where time of each 
cycle mattered. It 
turned out that if the code ran from the built in flash memory - sometimes the 
fetch would take 
longer than other - only moving the code to RAM solved the problem.

What this means is that most modern hardware can not really do what I call 
'determinant real time' - 
There are calls outside of user and system control that steal cycles and no way 
to predict the worst 
case.  The good news is the hardware is mostly fast enough that we can still 
respond in micro seconds.

I recently 'rooted' a droid device - and found that there was still code 
running that I did not have 
access to.  I dumped that one and bought a Chinese droid device that lacked 
this 'feature'. (I 
assume that all devices are compromised  - I don't think it prudent to ever 
store passwords on a 
connected device anymore.)



--------------------------------------------------------------------------------
Karl Schmidt                                  EMail [email protected]
Transtronics, Inc.                              WEB 
http://secure.transtronics.com
3209 West 9th Street                             Ph (785) 841-3089
Lawrence, KS 66049                              FAX (785) 841-0434

99% of lawyers give the rest a bad name.
--------------------------------------------------------------------------------

------------------------------------------------------------------------------
October Webinars: Code for Performance
Free Intel webinars can help you accelerate application performance.
Explore tips for MPI, OpenMP, advanced profiling, and more. Get the most from 
the latest Intel processors and coprocessors. See abstracts and register >
http://pubads.g.doubleclick.net/gampad/clk?id=60134071&iu=/4140/ostg.clktrk
_______________________________________________
Emc-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-developers

Reply via email to