Nil wrote:
There were reports a while back about support for Windows 7 on Intel
Skylake processors going away. Truthfully, I don't really understand
the implications of that. However, this article,

http://www.channelpro.co.uk/advice/9...-vs-skylake-14

says that Microsoft has changed their minds and will support that
CPU until mid-2018 at least.

Is there a reason for me to avoid building a Windows 7 system based
on a Skylake processor?

There's the XHCI-only USB ports. The Win7 media doesn't support
XHCI at startup. (USB3 drivers are standard on Win8 or Win10, but
the drivers are provided by the hardware maker [Intel] for Win7
as far as I know.)

http://www.tomshardware.com/answers/...-platform.html

If a desktop, can you install Win7 with a SATA optical drive ?
I hope so. Just not a USB optical drive or a USB flash drive,
without some slipstreaming. And if the hardware is XHCI only,
you're going to want a PS/2 port on there, as you might have
no keyboard/mouse on first boot. Slipstreaming in the XHCI driver
sounds like a good plan.

You have to ask yourself what Intel was thinking, when they made
radical changes to a hardware block like that, removing
backward compatibility. After all, Intel supported IDE mode
forever on their disk ports, solving at least the ability to
get disk drivers to work for Win98. So even if your OS didn't
have video drivers, the disk side of things still worked. Why
Intel mucked about with USB, is unclear, when they could have
just copied the previous generation of intellectual property
design for the USB port. In this day and age, it would be
pretty difficult to accept some "gate count" excuse. I'm sure the
PCH type silicon is so small, it's pad limited, and if you
save gate count, it just leaves undoped silicon in all
the open areas created.

*******

This has nothing to do with any "Microsoft" support story.
Which is an entirely different issue.

An example of Microsoft support, would be the two patches released
to improve things like Bulldozer support on older OSes. The
performance difference wasn't great, but at least it shows
that Microsoft had to spend the time in the lab proving that
scheduler changes would work. So that's a support cost,
associated with a novel architecture showing up. I don't
see how the Intel architectures particularly trigger this
kind of patching and labwork. Intel has a large shared L3,
small L1 and l2, cache coherency, and relatively easy
scheduling work (not much to worry about). There may be
various flavors of internal ring busses, but the scheduler
doesn't care about hardware at that level of granularity.

So I guess at some level, I don't understand Microsoft
whining. It's AMD they have to watch (is Zen different?).

*******

Use standard practice. If buying hardware in any generation,
visit the motherboard page on Newegg. If customers are having
trouble installing OSes on their new purchase, perhaps they
will document it. You can also try the pages of Amazon for this.

Never buy hardware, a week after it ships. Leave it to
the early adopters to get their hands dirty. If the
installer DVDs don't work, the customer reviews should
contain enough reports after a three month period.

And I really don't expect much of Intel. It would
be shabby chipset driver packages I'd be watching for.
Microsoft has their part to play, but at this point
I seem to be missing what their big concern might
be with regard to Intel. ACPI C-states maybe ?
Hardware that runs hotter than it has to ?

Paul