IE7 beta 2 preview refresh - released march 20th

[Son Goku]

Well, I loaded it on along with an update to the other browser while BOINC was shutdown following the WU backup... Initially there was a 50+ KB kernel paged pool usage that had me concerned, as it seemed to be going up rapidly, and I have had XP blue screen when that got too large. However, following a cleaning of the pre-fetch folder (well if upgrading software, makes sense to get rid of older copies that might get loaded from pre-fetch) and a reboot, all seemed well.

Definitely looks different then IE 6, and it took me a sec to find the bookmarks this morning Other then having to acustom myself to a new interface, all seemed well so far...

[SavagePaladin]

Prefetch actually keeps track of the state of your EXEs when it stores the data, so if they change, it changes. No real driving need to clean it out, from what I know.

[Sazar]

Doesn't really affect my performance one way or the other, if I clean out the prefetch folder or not.

It may have something to do with my system essentially being pretty much clutter free.

[rewt]

Opera still pwnz!! w00t

[Son Goku]

Yeah, probably doesn't make much of a perf difference. Main reason why I did that when I saw the kernel paged pool go up, was well some bad experiences I've had of late, when that starts happening, and hopefully to remove any components that might have been loaded with it. A reboot probably helped a lot too.

Lets just say that the 32-bit version of Windows XP, given the limits in a 32-bit addressable memory space, imposes some hard limits on various system components, including paged pool memory. In the past, I had seen that spiral out of control for whatever reason, not when I first got my A64, but sometime after. When this hits about 128 MB (my bad, I meant over 50 MB not KB), system responsiveness begins to slow and grind to a hault, as various system resources start hitting the maximum that the system is internally allowed to allocate. The cause isn't so much pre-fetch, or any particular thing that gets loaded; it's excessive kernel memory allocation. AKA at those times the system kernel was hitting about 192 MB or so...

Just the other day, when my comp crashed and trashed my entire BOINC cache, while Kaspersky 6.0 release candidate did a scan that locked up the comp, it sent kernel paged pool to 160 MB, and at that moment anything trying to begin on the comp just got a

Unable to allocate system resources

or

System resources maxed out

error message. Usually, the kernel paged pool getting that large preceeded a BSOD, and checking the specific STOP message off Microsoft's website, gave reference to this limit in paged pool memory, and that limit being exceeded.

Other then the beta antivirus crashing, things have been more stable over the last few months, so not entirely sure why my comp was crapping out after exceeding memory allocation limits set in the system for certain things. Has resulted in me keeping an eye, as I rather do not like blue screens of death, or losing WUs that take 400+ or 1,800+ hours of CPU time itself to crunch on an A64 to complete

[SavagePaladin]

Opera does rock. Especially Opera 9.

But IE7 will have a 64 bit version. And I'll use that.

I never had any problems with the limitations in XP. You either use some wierd programs or some buggy ones...

__
(or are a developer or somesuch, they use a lot deeper tools)

[Son Goku]

Nah, I seriously doubt it is either... Quite frankly my programs did not change before/after like that... So my software does not represent a change. Something perhaps was going on, but it wasn't because I loaded any new software...

The only thing I did last year was move my data folder to the d: drive, and repartition e: as a Linux drive, then install Linux. And that itself would hold no logical correlation. Other then that, software hadn't changed at all, but that limit did come up. I can absolutely rule out software changes as being even a possible cause however, but beyond that can't say more...

BTW, and just as a note, one should not be so quick to equate personal experience with what can and can not be flat out, or to assume that because someone never saw something themself "it has to be your software"... Many of the bugs which were explicitly fixed in Service Pack 3 for winNT 4.0 under the heading of "Service Pack 2 causes..." I had not personally experienced. However, MS did provide a fix for them in their following service pack. One of them, the SP2 causes RAS disconnection problem I never did see myself until I tried to download SP3 itself incidently. That was on an old Pentium system, back when Service Pack 3 for Windows NT 4.0 was first released... Since then, there have also been all kinds of fixes introduced in Service Packs and software patches for things I've never experienced personally...

One thing about a 64-bit version however... Not sure if this will be remedied latter on, but with a 64-bit version of Mozilla as available under Linux-64; not all plugins (such as Macromedia Flash as I remember) will load. Some plugins do as of yet refuse to play with a 64-bit web browser it would seem. There's gonna be a new for greater 64-bit software support, though expectation can be that when Vista is out, we might see more of this. Some users of winXP-64 had also mentioned some compatibility issues in the past, and at least on the Linux side, I had seen them in action wrt browser plugins and the like...

Moral of the story there, was one had to have a 32-bit browser to run side by side with the native included 64-bit browser to use things like flash...

[j0j081]

Originally Posted by SavagePaladin

Opera does rock. Especially Opera 9.

But IE7 will have a 64 bit version. And I'll use that.

I never had any problems with the limitations in XP. You either use some wierd programs or some buggy ones...

__
(or are a developer or somesuch, they use a lot deeper tools)

hopefully they get the 64 bit flash to go with it.

[SavagePaladin]

Although NT4 has very little relevance, I do see your point.

At any rate, since I don't really want to go into a debate right now, I will move on to this...

Flash and other plugins will come eventually. Yeah, I'll load up a 32 bit version when I need Flash, but in general I'm going to be using the 64 bit version. Personally, I think if Opera or Firefox were to release official x64 builds, it would prod the plugin makers into it. The pressure involved would grow.

On the other hand, I seriously hope IE7 64 doesn't bother you if a site needs a 32 bit plugin like IE6 64 does today. That's just retarded.

[Son Goku]

Originally Posted by SavagePaladin

Although NT4 has very little relevance, I do see your point.

At any rate, since I don't really want to go into a debate right now, I will move on to this...

That there have been service packs that have broken things in the past is actually quite relavent, but anyhow... Sometimes in the process of fixing one codeing error another crops up. But the point really comes to the fact that in the end we're running a system, and it really isn't as easy as simply pointing a finger and saying "such and such must be at fault". Sometimes nothing in particular is at fault, it's just one of those things... I've seen this sorta thing before, and it's enough to give people headaches.

CPDN (a distributed science project that's involved with running climate models to help better understand how the weather and atmospheric changes and the like) has problems with sulpher units on some configs but not others. And there's no clean cut picture... It happens, and reason things get multiply QA tested over a wide variety of configs... And they're matters that can drive many a developer nuts.

Even on the CPU front, given different core steppings, 2 seperate steppings of a given CPU (say an A64, a Pentium III, or whatever) can have seperate errata, and yet every CPU no doubt has some... Because they don't just recall CPUs with newer models, it's many times deemed easier to address this in software, so companies like Intel release the errata sheets and allow programmers to know the issues with example code on how to avoid it, and then releasing such example code and the like leave it to the software engineers to implement the "fixes".

On the other hand, there can in fact be software faults/bugs that don't come up under all configurations/circumstances, and only come up under certain situations. Reason in part an app could run quite well for 5+ years, and then on some system or under certain circumstances throw up a general protection fault (essentially an incorrect attempt to access some memory, that had been de-allocated already, perhaps due to an incorrect pointer or something)...

Which also brings up the ever important matter of error handling where exceptions can be trapped, and ideally some exception handling code can be run without the end user ever having to know the exception occured in the first place. When possible, the code being designed to essentially "self correct". It's not always possible though (for instance invalid user input), so some errors do result in the user having to be inconvenienced. But given exception handling, the comp can many times experience many more internal errors then the user would ever notice, other then a few extra CPU cycles to run exception handling/garbage clean up code, which run in nano-seconds time, the user never notices anyhow...

It's as Fred once told me however (Fred having been a computer engineer who worked at NASA among other places for about 45+ years of his life when he said this); many people are just used to pointing fingers and blaming. You call Intel and it's invariably a Microsoft problem. You call Microsoft and it's an Intel problem (and on the Intel developer forums many years back I saw cases where people looking for support were jockeyed between MS and Intel support by both companies almost continually)

His contention as an engineer himself being that one gets the people who just look at the hardware, and people who just look at the software, but how many look at the complete system? But it's the complete system people do run. And of course the user really doesn't care about "assigning blame", or saving face for this or that particular vendor

Flash and other plugins will come eventually. Yeah, I'll load up a 32 bit version when I need Flash, but in general I'm going to be using the 64 bit version. Personally, I think if Opera or Firefox were to release official x64 builds, it would prod the plugin makers into it. The pressure involved would grow.

Yeah... Lets just say there's a reason I don't use winXP-64 (not wanting to buy another version of XP only to get Vista latter) being part of it. By the time Vista is released, I rather expect that Vista will see greater software support then winXP-64 has seen. Vista's release is the time I'm rather expecting to make the switch to a 64-bit OS. Until then however...

[Son Goku]

Well, after having checked, I thought I'd alert people to a potential matter as of yet. Went to go to Windows Update the other night and after having put IE 7 on... For now, Windows Update won't work, insisting that it requires IE 5 or greater. Of course IE 7 beta 2 is < IE 5, but it's not being recognized...

Of course, this is a matter that will no doubt be addressed by the time IE 7 is out of beta and released; but for now testing the newer version of IE does leave it unrecognized by Windows Update itself... It is of course possible to get updates otherwise; but people will need to follow the links to gather them one's self. Thought some peeps who might be interested in trying IE 7, might like to know this ahead of time...

[NightFire]

Son Goku, this is very interesting, as I have also loaded IE7, and am downloading updates from Windows Update right now! In the Beta 1 build, WU wouldn't work, but since Beta 2, everything's been going fine for me.