Improved performance

My wife and I each have several Win10 machines.
The other day, I decided to check out the tower she's used without issue.

I noticed excess hard-drive activity, yet saw no obvious problem.


I decided to poke around in Device Manager and saw the SATA controller
was a 14 year old, generic Windows driver. I decided to try the "update
driver" option and it found an actual Intel driver about ten years newer.

Problem solved.


Looks like I lucked out there. For some reason I thought that Windows
Update automatically looked for hardware driver updates as well.
I know after a fresh installation, the video drivers are often updated.

philo wrote:
My wife and I each have several Win10 machines.
The other day, I decided to check out the tower she's used without issue.

I noticed excess hard-drive activity, yet saw no obvious problem.


I decided to poke around in Device Manager and saw the SATA controller
was a 14 year old, generic Windows driver. I decided to try the "update
driver" option and it found an actual Intel driver about ten years newer.

Problem solved.


Looks like I lucked out there. For some reason I thought that Windows
Update automatically looked for hardware driver updates as well.
I know after a fresh installation, the video drivers are often updated.

The disk controller in the Southbridge has several
operating modes.

Compatible mode, offered a Win98-like set of resources,
in I/O space. In Native mode this might move to a PCI
bar and use a PCI interrupt signal. Microsoft has standard
drivers for both in Windows 10, and you'd be surprised
at how old those drivers are (yes, they still work).

There are also AHCI and RAID options.

Microsoft has a "class driver" called MSAHCI, and that
will work with hardware claiming to be AHCI compatible.
Perhaps a BIOS declaration helps identify when a port
could be covered by this (the "CC" value may hint at
a property the class driver could use). These standard
drivers is how the OS can be installed! If there weren't
non-Intel drivers for bringup, half the attempts
at installation would fail.

Intel makes an AHCI driver too.

If you're the curious type, you would look at the
Intel INF file and see if it does #include MSAHCI
or similar. Some Intel drivers "call" the Microsoft
subsystem. The Intel driver provides an "Intel label"
for Device Manager, but some of the heavy lifting is
done with Microsoft materials.

Intel also has various flavors of RAID, with the RST
driver and the like.

I think Windows has things like "IASTOR", which may
be suitable for Windows 10 to "bring up" a RAID, without
a visit to the Intel site. The name of that one,
could have changed once or twice since that string
was used.

Your description of a "feeling of inadequacy" with
your drivers is misplaced. The disk activity might
well be some Search Indexer or Windows Defender
background scan activity. The disk activity is not
likely to be some "oh jeez, I got behind and look
at me, I'm still doing disk I/O ten minutes later".
If Windows does get behind (because the disk itself
is slow), the LED will just stay continuously lit
while the cache is unloaded out to disk. Windows
has a System Write Cache which is charged against
main memory, and the Task Manager "Gigabytes Used",
the line will start to drop down a bit, if the
System Write Cache is "draining" to disk. The OS
has a maximum percentage size it seems to use
for the Write Cache. It will not use the entire
System RAM as a Write Cache.

And yes, the Write Cache is dangerous. It's possible,
with some effort and bad luck, to start two processes
running, which "bump heads" and the OS seizes up.
I had this happen one day, and as I was watching
the machine misbehave a bit, by the time I'd formed
a mental picture of the mistake I'd made ("setting
up two running processes to bump heads and get
into a resource fight"), it was too late and the
machine just froze for lack of RAM. When it does that,
there's still furious activity inside to "gimme RAM,
gimme RAM", but you can only squeeze a lemon so flat,
and there's no longer lemon juice. And so the system
is "dynamically frozen" in a death loop. Power button
comes next... You can't start Task Manager without RAM.
Not on Windows 10 at least.

OK, you've installed a driver, and yes, there might
be some difference in CrystalDiskInfo, but the
driver that was there previously was likely perfectly
functional.

Microsoft class drivers with old date stamps, are
not to be feared. They stabilized long ago.

Paul

The reason I made the assumption it was the driver was because the excess disk activity had been going on for days. If I just let the machine sit doing nothing the activity would eventualkt stop...but one movement of the mouse was enough to get it going again.



Anyway...for now, it looks ok.

I also have gotten bad drivers but not anytime within recent memory...probably back in the Win98 era.

Anyway...it's my wife's #3 machine so not terribly critical.

philo wrote:
The reason I made the assumption it was the driver was because the
excess disk activity had been going on for days. If I just let the
machine sit doing nothing the activity would eventualkt stop...but one
movement of the mouse was enough to get it going again.



Anyway...for now, it looks ok.

There are ways to track it down.

You can try Resource Monitor (a thing to click in Task Manager).

Even Task Manager itself, if you do an "Add Column" and add
disk read and disk write to the columns, that'll give a hint
which process is doing it.

Or, you can use Sysinternals Process Monitor to
track disk I/O, and see what process or Service Host
is doing it.

Sysinternals Process Explorer, if run as administrator,
will show which services are inside a SVCHOST, narrowing
down which service it might be (you get the PID from Process Monitor,
them look for more info about that PID in Process Explorer).

For Windows Defender, there is an EXE doing it so it's a
bit more obvious what is scanning in that case. The
Search Indexer daemon is likely inside a SVCHOST, so
requires the "Windows 2-step" to figure out.

Paul

Thank you Paul.

You gave me some good, additional info.

I will probably take another look at the machine .

On 6/15/2020 1:42 PM, philo wrote:
My wife and I each have several Win10 machines.
The other day, I decided to check out the tower she's used without issue.

I noticed excess hard-drive activity, yet saw no obvious problem.


I decided to poke around in Device Manager and saw the SATA controller
was a 14 year old, generic Windows driver. I decided to try the "update
driver" option and it found an actual Intel driver about ten years newer.

Problem solved.


Looks like I lucked out there. For some reason I thought that Windows
Update automatically looked for hardware driver updates as well.
I know after a fresh installation, the video drivers are often updated.



Here is what I have done.
Paul , I took your advice and have been poking around with the utilities
in Sysinternals. Nothing looks unusual.

After the driver update, though disk activity at idle has now
stopped...the act of just opening Firefox initially pegs HD usage.


On my own machine (same specs. Quad core CPU & 16gigs RAM) there is zero
disk activity when I open Firefox so something is clearly something very
much wrong.

philo wrote:
On 6/15/2020 1:42 PM, philo wrote:
My wife and I each have several Win10 machines.
The other day, I decided to check out the tower she's used without issue.

I noticed excess hard-drive activity, yet saw no obvious problem.


I decided to poke around in Device Manager and saw the SATA controller
was a 14 year old, generic Windows driver. I decided to try the
"update driver" option and it found an actual Intel driver about ten
years newer.

Problem solved.


Looks like I lucked out there. For some reason I thought that Windows
Update automatically looked for hardware driver updates as well.
I know after a fresh installation, the video drivers are often updated.



Here is what I have done.
Paul , I took your advice and have been poking around with the utilities
in Sysinternals. Nothing looks unusual.

After the driver update, though disk activity at idle has now
stopped...the act of just opening Firefox initially pegs HD usage.


On my own machine (same specs. Quad core CPU & 16gigs RAM) there is zero
disk activity when I open Firefox so something is clearly something very
much wrong.

How does the drive look on an HDTune benchmark ?

http://www.hdtune.com/files/hdtune_255.exe

Is performance normal ? Or a flat line at 5MB/sec (PIO mode) ?

Paul

Thanks Paul.
According to HD tune, average transfer rate is 80MB/sec. Which I believe is normal.


Also ran msconfig and took everything out of startup and stopped all services.
Made no difference.

Were it not for the HD activity which I can hear...I doubt if I would have had a concern as the machine seems to be running ok.

philo wrote:
Thanks Paul.
According to HD tune, average transfer rate is 80MB/sec. Which I believe
is normal.


Also ran msconfig and took everything out of startup and stopped all
services.
Made no difference.

Were it not for the HD activity which I can hear...I doubt if I would
have had a concern as the machine seems to be running ok.

Process Monitor

https://docs.microsoft.com/en-us/sys...nloads/procmon

can log all CreateFile, ReadFile, WriteFile operations.

Double-click it to start it running, dismiss the filter
dialog and it collects a trace.

To stop the trace, you can go to the File menu and untick
the tick that's there.

Edit should have a "clear" for a stopped trace.

You can simply scroll through the trace, to see
what's doing reads and writes.

Applying filters using the filter dialog is nice, and
that's how you reduce the "noise" in the trace to a
decent level.

Paul

philo wrote:
Thanks Paul.
According to HD tune, average transfer rate is 80MB/sec. Which I believe
is normal.


Also ran msconfig and took everything out of startup and stopped all
services.
Made no difference.

Were it not for the HD activity which I can hear...I doubt if I would
have had a concern as the machine seems to be running ok.

You should also set the time zone and sync the clock on your
machine, so the timestamps on your posts will be more indicative
of when your post came in.

Right-click Start and run "control.exe" or just "control"
to find the traditional control panels. There should be
a Date one in there, and a bit of fiddling should find the
time zone and sync to time.windows.com or whatever.

Paul

In article m, philo
wrote:

Thanks Paul.
According to HD tune, average transfer rate is 80MB/sec. Which I believe is
normal.

for a modern drive, that's slow.

As to my time stamp.
I am posting frim my phone now but I also post from my dual boot computer which runs win10 and Ubuntu.

I specifically set my time so that no matter which OS I boot to the machine displays the correct local time.

I never thought to check to see what the time of post is.

Anyway I'm thinking maybe I should just put an SSD in my wife's machine and then I won't be bothered by the sound of the HD.

On 6/18/2020 10:29 AM, philo wrote:
As to my time stamp.
I am posting frim my phone now but I also post from my dual boot
computer which runs win10 and Ubuntu.

I specifically set my time so that no matter which OS I boot to the
machine displays the correct local time.

I never thought to check to see what the time of post is.

Anyway I'm thinking maybe I should just put an SSD in my wife's machine
and then I won't be bothered by the sound of the HD.



That post was from my phone and I see the wrong time stamp. this on is
from my Win10 machine

philo wrote:
On 6/18/2020 10:29 AM, philo wrote:
As to my time stamp.
I am posting frim my phone now but I also post from my dual boot
computer which runs win10 and Ubuntu.

I specifically set my time so that no matter which OS I boot to the
machine displays the correct local time.

I never thought to check to see what the time of post is.

Anyway I'm thinking maybe I should just put an SSD in my wife's
machine and then I won't be bothered by the sound of the HD.



That post was from my phone and I see the wrong time stamp. this on is
from my Win10 machine

You would think, with all the tech in a phone, they
could manage to keep a clock of some sort. Even the
incoming CallerID information, I think it has a
timestamp, so that a receiving device could even
set their clock by it.

Things like GPS provide time (not every phone will
have GPS), but the time in that case is UTC and
there's no time zone info.

It sounds like maybe the phone needs a timezone,
to go from the "native" UTC to the "display" timezoned
value.

Paul