k, first for the stuff that *is* relevant to this part of the thread g

(the last time I had IE6 installed was last summer)

Symptoms; with IE6 installed, large file deletes causes WE to hang; also
file operations can cause mouse/GUI to stop until completion.


WinME completely patched
p3-667, 512MB
VIA Apollo+ Chipset (Asus CuV4X MB) 4-in-1 drivers v4.37
Matrox G400
Creative SB0100 (SB Live)
SoHoWare NIC
Partition layout similar to yours.
DMA2 on the drive.


cquirke (MVP Windows shell/user) wrote:
On Tue, 05 Apr 2005 12:46:23 -0400, Rick T

cquirke (MVP Windows shell/user) wrote:

On Mon, 04 Apr 2005 22:05:02 -0400, Rick T


Is WinME OK with this, in your experience?


hasn't complained yet g... no, I just recently put Win2K on to see
what it's like (different), the problems I experienced (apparently) with
IE6 were on WinME.


Are these "the" problem, or others? I'm fishin' for info to pin down
the scope of "the" problem, within Win95/SR1/SR2/98/98SE/ME and IExx


I note where "mouse or keyboard sticks during file operations" is listed
as not being related... IME that occurred also with the upgrade to IE6
and I'm in for an uninstall if it isn't fixed.


What typically would happen is during heavy disk usage the mouse would
stop responding.


That does not reassure me. Normal disk operations and general
software overhead don't stick the mouse or delay keystrokes, as the
interrupt routines for those services usually cut right through.
Blocking them suggests a delay within similarly-low-level code, such
as the interrupts that serve the HD itself.


HD's fine though I occasionally get that first issue (occasionally as in
once every couple months for a minute)


I'm even less reassured! On my table this week we
- HD that passes diags but reports "write error" fixing the FAT (!)
- HD with patchy increased latency on DOS mode surface scan

bad spacing/temperature calibration algorithm? (user turned it off for
some reason at some point in time)

- HD that died outright and isn't coming back

The last was a bitch. User's quite savvy, and I taked them through
"the prelim" before the system came in; MemTest and DOS mode surface
scan. Those passed OK, reportedly. System comes in and I do a formal
virus scan; fixes a RAT. Then I leave it doing a DOS mode surface
scan while doing other stuff. C: and D: pass OK, but then I hear
clanking noises and see E:'s cluster progress counter has stopped.

Power off, proceed to evacuate data. But HD makes clanking noises
while delaying POST for ages; BIOS never sees HD. Shroud time.

Only *then* does user tell me "oh yes it's been making clanking noises
every now and then". The scary thing is, without that info and even
with a best-practice approach to such matters, I'd still have had no
reason to evacuate HD before doing anything else.

So this sort of data tragedy could indeed happen again.

could, hasn't happened in the last few months, though for me it'd be
just an inconvenience.


As to the cause, my magic 8-ball says IE6 WE relies on the NT threading
model which causes 9x to go nicely pear-shaped (and explains the
stickiness too).

More on that?

pure speculation on my part; NT(/2K/XP) use a more advanced threading
model; my theory is IE6 (specifically those 2 files) was writ with that
in mind and it just barely works on 9x.

Hm - too foggy to focus on, but I'd expect a "more advanced" threading
model to be less likely to fall on it's ass ;-)

well the chair it's expecting to sit on could be non-existent.

Or, bluntly, there's too many forking file processes floating around
tripping over each other for 9x to handle properly.

I think there's some wrap-up overhead that spirals out of control or
fails in some deadly embrace until some supervising timeout kicks in.

yup; dueling file-locks, etc.

Hmmm... wonder if anybody with that problem uses SCSI.

Dunno.

because SCSI has an ordering mechanism in place.

Win2K is nice and is more robust than WinME (only one crash and that was
on a non-cert driver) but booting takes a long time... WinME + license
vs. Win2K without (or Win2K + $$$ to get one) is starting to get
weighted towards WinME.

Win2000 is a different OS (NT 5.0 vs. Win9x 4.2 or whatever). I'd
expect more robustness, faster if more RAM but slower in low RAM, and
various compatibility tradeoffs.

Much slower booting, a little faster all-round and speedier shutdown; a
bit rougher 'round the edges user-friendly-wise than WinME.

I also have occasional mouse-sticking problems (on the original IE) with
2K, but that isn't restricted to file operations and feels more like a
tasking priority issue; didn't with WinME + original IE.

OK. Scoping those on the basis of which HD was used may be revealing.

?

It's a Quantum Bigfoot TS (last of the 5.25s) with specs comparable to a
5400... it ain't the drive.


Rick




---------- ----- ---- --- -- - - - -

Gone to bloggery: http://cquirke.blogspot.com

---------- ----- ---- --- -- - - - -