On Sat, 4 Aug 2007 16:16:05 -0700, JJ wrote:
>I thought that this problem went away when Windows 95 was released. But it's
>real with Vista and it's upon me.
It's the same damn problem, too!
>I would have a few apps open - my normal work routine. They include Windows
>Media Player, Windows Explorer, a text editor called Crimson Editor from
>www.crimsoneditor.com, a few tabs open in Internet Explorer 7, the sidebar
>with 6 gadgets. I found that IE is a huge resource hog - with just one tab
>open right now it's gobbling up 90MB+ of memory according to Task Manager.
>But my problem is that when my memory consumption reaches someover 55%, I
>start to experience problems like not being able to open dialog boxes,
>context menus not working, etc.
In spite of the melt-down of Win3.yuk, and the difficulties that Win9x
faced when weaning software off legacy 16-bit heaps, Windows STILL
uses finite global heap stores for unbounded per-instance data.
The 32-bit heaps would allow massive heap size, but these are still
preset to a fixed limit in XP and Vista. The fix is a manual setting
(registry edit, I suppose) that boosts one of the heaps beyond the
current 3M or so limit.
Did MS miss Scalability 101 ?
Let's see if the screaming has started yet...
Google( Vista heap size )
http://blogs.msdn.com/ntdebugging/archive/2007/07/05/desktop-heap-part-2.aspx
http://www.lockergnome.com/nexus/theoracle/2007/07/27/vista-showing-ho...icrosof
....chatter...
http://henbo.spaces.live.com/blog/cns!2E073207A544E12!162.entry
Uh-oh:
http://blogs.technet.com/askperf/archive/2007/07/24/sessions-desktops-...-window
Note the caveat in red:
"Please do not modify these values on a whim. Changing
the second or third value too high can put you in a no-boot
situation due to the kernel not being able to allocate memory
properly to even get Session 0 set up."
I think we're going to see some bad advice as this issue emerges (it's
been bugging me in XP SP2 with IE 7 and Irfan View for about a year
already) and this caveat will smash a lot of advice-takers.
Looks like I'll have to blog this, after all...
If MS do have a good technical reason why they can't use 386-era
address relocation to create the illusion of a contiguous yet
dynamically-scalable heap (e.g. risk of exploitable race conditions),
they should get that information out asap, as the temptation to bash
MS for being hopelessly unable to learn from experience is strong.
>------------ ----- ---- --- -- - - - -
The most accurate diagnostic instrument
in medicine is the Retrospectoscope
>------------ ----- ---- --- -- - - - -