[Info-vax] X86 first boot? - A Really Stupid Question

[Stephen Hoffman]

On 2019-03-29 22:30:30 +0000, Scott Dorsey said:

> It's worse than ever, because the new generation of video coprocessors 
> is so much more complex.

Yes, various high-end GPUs are often not openly documented.  OpenVMS 
hasn't supported high-end graphics controllers in decades, even when 
some of those high-end graphics controllers were made by DEC.  The 
Intel Graphics Technology (Intel HD Graphics, Intel Iris, Intel Iris 
Pro, etc) integrated graphics co-processors are openly documented.  As 
are various other lower-end graphics controllers.  And the IGPs are 
certainly comparable to what has been supported by OpenVMS, and quite 
probably substantially faster.

> There's a lot of this, and when it happens we figure it out.  We plug 
> the UPS into a Windows machine with the vendor's driver and look at 
> what is going down the serial line with a terminal Y-cabled off the 
> port.  Then we write a driver.

Or VSI or a third-party looks at open-source implementations, where 
that's available.

VSI will likely be tied up with core I/O and permutations for at least 
a little while.  Networking, USB, storage, serial, etc.

OpenVMS does not now support UPS control, though there is at least one 
third-party add-on package available for OpenVMS.

I'm accordingly thinking that any integrated UPS support from VSI will 
be a release or two down the road after the VSI OpenVMS x86-64 
production release and quite possibly later, and any of that will 
depend on which UPS models are then most popular.

As an example of available UPS documentation, Schneider/APC wasn't and 
probably still isn't offering their Microlink documentation, but did 
make the UPS-Link protocol documents available.  And various of the 
Microlink UPS controllers can be upgraded (cross-graded?) to support a 
MODBUS control protocol and which is documented, and other UPS models 
can be retrofit with an interface for UPS-Link.  Eaton was reportedly 
open about their control interface documentation, so there's that.  And 
I'm expecting that many of the folks interested in UPSs will be 
interested in using a network-based interface into the UPS, and not 
USB.  And use of apcupsd or analogous may (will?) be an option for 
some.  And it wouldn't surprise me to learn that the existing 
third-party UPS interface package will be ported to OpenVMS x86-64.

Keeping track of what works and what doesn't work, and what's been 
tested and on what versions, is no small effort; testing and tracking 
and documentation.  And hardware support retirement plans, too.  
Keeping track of all of this is arguably as if not more important than 
the testing and implementation work.  The hardware support database is 
a big deal.

But then this is all several years down the road...


-- 
Pure Personal Opinion | HoffmanLabs LLC