Will Windows 7 32bit support more than 4 gigs of ram?

csnv said:
i can't even get 4g out of my 4g, i get like 2.75gig out of 4gig. Anyone know why?
Click to expand...
Assuming you have a 64 bit OS installed, check your BIOS for a setting called Memory Mapping, or something to that effect. If you have it, make sure it is enabled.
 
pallesen said:
AWE has very little to do with PAE (nothing, actually). All programs will benefit from the extra memory without special coding.

And ... PAE is not that slow (most users have PAE enabled already).
Click to expand...

They do, but Windows 32 bit still doesn't throw out the TLB due to the 2GB Kernel/2GB User. If you start using 4GB per process, your performance will go down, because you're going to thrash your TLB. (How much, different people say different things...)


(Note, I haven't worked on this stuff in a really long time... I'm going off memory here.)

This posting is provided "AS IS" with no warranties, and confers no rights.
 
Ranma_Sao said:
Geoff, have you looked at the performance penalties of PAE Mode?
Click to expand...

Only since you ask me directly...

I've looked at it no more than said in the article. I haven't looked hard because general arguments lead to an expectation of no substantial penalty (and the only people who do talk of substantial penalties plainly haven't understood Intel's literature, or Microsoft's for that matter). Of course, general arguments can be blown away by particularities that one hasn't yet accounted for.

I can't see anyone disputing that 32-bit with PAE is slower than 32-bit without PAE, not to a substantial degree but certainly to a measurable one. If you didn't have memory above 4GB and weren't chasing the security benefits of DEP (which requires PAE), then you wouldn't enable PAE. I expect the comparison that matters in practice is between 32-bit with PAE and 64-bit. This comparison is much less clear: there are performance factors in favour of each. I have no idea which wins, but I suspect that neither wins reliably over a wide range of real-world configurations (especially allowing a wide range in what people actually do end up testing).

Either way, the effects are nothing like as large as some say. PAE doesn't switch memory in and out from alternate banks. It provides as direct an access to physical RAM as anyone ever has in these days where we take paging as granted, i.e., that the linear addresses we load into registers for instructions that read or write memory are not physical addresses.

PAE is not a magic bullet, it takes a lot of work for the application developer to support. I seriously doubt most client devices would take the work to develop for PAE Mode.
Click to expand...

No third-party software at any level has to support PAE explicitly. Even among kernel-mode components, only relatively few drivers are affected even indirectly. These work with physical memory addresses and do need to be coded with an awareness that physical memory addresses can exceed 4GB. This same awareness is needed by 64-bit drivers. For instance, if a device can do DMA only for the first 4GB, then both its 32-bit and 64-bit drivers need to arrange for double-buffering.

It's true that 32-bit drivers for client devices were sometimes (or even often) not coded with this awareness. The only examples I know of that compile without warning yet wouldn't need to be fixed in porting to 64 bits are highly contrived.

AWE shows the amount of work using PAE to support programming above the 4GB barrier.
Click to expand...

AWE is a whole different matter. It operates at a (much) higher level. It provides that what's seen at a given linear address varies according to which "address window" you view through. Where PAE allows the system to use physical memory above 4GB to support everyone's need for (separate) linear address spaces, AWE is a technology by which each process can use more memory, over time, than fits in its linear address space.

Obviously, AWE requires its user (typically an application working with very large amounts of data) to be explicitly aware and in control. As you found, a lot of coding can be required. AWE works very much better if PAE is enabled so that all the memory you want to use can always be supplied immediately as physical memory. But AWE does not require PAE, nor vice versa, and to count AWE's overheads as PAE's would be entirely groundless.

I don't see why people are so excited about PAE, you want 64 bit address spaces, go 64 bit.
Click to expand...

Again it's not something I see any responsibility to comment on except that you've put it to me. I thought I was always clear in my article that appraising 32-bit with PAE against 64-bit is not the subject, but as more and more people abused me for wanting to retard progress, I did strengthen the point so that it is very explicit. The article is in no way a recommendation of PAE. It is an archivist's record that, contrary to what many people think, 32-bit Windows Vista has all the necessary code for using physical memory above 4GB but such use is constrained by tamper-protected licensing data. Indeed, Windows Vista is the first version that has both the code and the constraint.

Whether anyone should want to use memory above 4GB with 32-bit Windows is a whole other question. Obviously, if you were buying a new computer, you'd get as much memory in excess of 4GB as you can and you'd put 64-bit Windows on it. I can't imagine anyone would even think to dispute that.
 
mikeblas said:
Either you're just using this extreme to try and further your own argument, or you don't actually understand the compromise that you're insisting that you do understand. Since you say that it's a fair description, then you've already satisfied your own plea for an explanation of your misunderstanding.
Click to expand...

Oh, so this "extreme" you keep talking about is my summary phrasing of "free-for-all blue screens" to describe the susceptibility of ordinary users to blue screens from bad drivers?

It is a fair description. A free-for-all is a fight that's open to all comers. It says nothing of proportion, just of exposure. But I've elaborated the term already. For you to seize on it as showing that I don't understand the costs of even a small risk scaled over many users is ridiculous. It's obvious that even a small risk of ordinary users with ordinary configurations running into blue screens from bad drivers has to be eliminated. Oh, but I've said that before, too.

How do you get from this obvious driver of compromise to what Microsoft actually did? As I keep saying, and you keep ignoring, Microsoft had ways to eliminate that risk without reaching for the blunt instrument of licensing.

Basically, you present general principles that everyone will agree with and you say those principles explain Microsoft's decision, but you do not actually _connect_ those principles to what Microsoft actually did. I say those principles do not themselves explain what was done, and I give my reasoning.

If you had something to point to by way of refuting that reasoning, you'd have pointed to it by now, but you don't look like you mean to try. I can't make you. For my purposes, I'm satisfied now with writing you off as a big fish who's happy in his pond where he gets away with little attention to logical development of arguments. You have nothing and you'll get no further attention from me.

Are you going to post the version of the page which was current when I made that statement?
Click to expand...

That's a good point, on the face of it. Like Microsoft's documents, mine change from time to time. Though I think you're clutching and I'm not going to bother with you further, here's something for the record.

You made "that statement" on 27th August. The page hasn't been touched at all since 1st September. The main change between those dates was to reduce the chance of my bandwidth allowance getting blown away again: I changed to smaller pictures and added scripts so that clicking on those pictures would get the high-resolution ones. There will have been some editing of text before I finalised it, but really, once the article was viewed several tens of thousands of times in a day, which was the day of your statement, I pretty much regarded it as set in stone.

I don't believe, but I can't disprove, that there can have been any editing in those few days to cover any previous assumption that "Microsoft supports only new machines". Even the draft in January cannot plausibly have been written with any such notion. It's unthinkable. No analysis of Windows can sensibly even be attempted without considering backwards compatibility.

Again, if you had good ground for your accusation, you'd have found a way to present it, even with some remark about not remembering such and such being in the article at the time of your statement. So, given that you've shown no substance on other points, I'm satisfied you have nothing on this one either.
 
As an Amazon Associate, HardForum may earn from qualifying purchases.
Tawnos said:
Could it be that he was actually there when some of these decisions were being made? *gasp*!
Click to expand...

How would this matter? If anything, if he was there and should have an insight into why one of several apparently good solutions was preferred, then it's just the more remarkable that he avoids presenting that insight. You might think he could very easily dismiss the alternatives. Perhaps he could but prefers to play games.

Whichever. It's not my concern any more. I've only revisited to meet his request for a version of the article from the day he wanted, but I must say I should be fascinated if he actually does find that the article on that day does make the assumption he alleges and the final article from a few days later doesn't!
 
Hmm...first I've noticed this. It's actually pretty simple, there's no 32 bit version of Server 2008 R2, therefor there was no need to actually work on PAE support for this kernel.
 
Question, why is it a "marketing ruse" to get people to go 64-bit? MS makes no more money on 64-bit than 32-bit Windows, so in what way is this a marketing ruse? The sooner users are on 64-bit the better for them, MS benefits indirectly by users using the better system and thus having a better experience overall and better ecosystem, but it does not rake in dough from 64-bit versions it does not get from 32-bit versions. That's just one reason that article is stupid, but it's a pretty big reason and easily overlooked.
 
Geoff Chappell said:
How would this matter? If anything, if he was there and should have an insight into why one of several apparently good solutions was preferred, then it's just the more remarkable that he avoids presenting that insight. You might think he could very easily dismiss the alternatives. Perhaps he could but prefers to play games.

Whichever. It's not my concern any more. I've only revisited to meet his request for a version of the article from the day he wanted, but I must say I should be fascinated if he actually does find that the article on that day does make the assumption he alleges and the final article from a few days later doesn't!
Click to expand...

You asked "how do you get ... what Microsoft actually did?" I showed a way that could have happened. Is it really that hard to get, or are you just butthurt that someone disagrees with your article, assumptions, and conclusions, and want to eFight about it?
 
Geoff Chappell said:
A free-for-all is a fight that's open to all comers.
Click to expand...
Another definition is "a chaotic situation", which would imply a high population of users exepriencing the problem. It also describes a fight, which doesn't really seem applicable. And this is the problem with your writing: you're so tied up in colorful rhetoric and making a tempest in a teacup that the signal is lost in all of your own noise. At the end of the day, most people don't know what the hell you're talking about. The people who do agree with your tone (rather than your content) find it interesting and agreeable; the people who don't agree with your tone are put off and don't read much further. The people who care about your content are overwhelmed by your appeal to emotion and induction.

Another definition is a brawl; a fight, as you indicate above. There's no brawl here; a large company made a complicated decision that was influenced by many factors, only a few of which you investigate. You wrote an emotionally charged piece on the decision which neglects most of the other factors, yet insists that you know better and that the company reacted using tools blunter than you would have liked.

Geoff Chappell said:
If you had something to point to by way of refuting that reasoning, you'd have pointed to it by now, but you don't look like you mean to try.
Click to expand...
No, I don't mean to try -- at least, not vigorously. I participate in forums as I have time; I don't have much time right now. Sorry to disappoint you, but entertaining your rhetoric and responding to your meta-arguments (that is, not only must I prove my point, but I must do it precisely the way you'd like me to do so) just isn't too high on my list of priorities.

Since your analysis was so obviously incomplete, why would anyone care what you conclude? Since you can't divorce the facts from your emotions, you make it that much harder to discuss the issue. Given that combination, for most people, it's just not worth the effort.

Even if we stipulated that your analysis was complete, and based on a more complete set of facts and circumstance than you probably have, of what consequence is your diatribe? The decision is the decision, and that's that. You can cry about it, posting bleating treatsies about it; or you can make do with what you have and move on to productive work.

Geoff Chappell said:
I can't make you. For my purposes, I'm satisfied now with writing you off as a big fish who's happy in his pond where he gets away with little attention to logical development of arguments. You have nothing and you'll get no further attention from me.
Click to expand...
You're announcing that you're taking away something that is of no value, and backing it up with an ad homenim attack. Really, why bother?
 
devil22 said:
Question, why is it a "marketing ruse" to get people to go 64-bit? MS makes no more money on 64-bit than 32-bit Windows, so in what way is this a marketing ruse?
Click to expand...

To charge more for 64-bit Windows than for 32-bit Windows would be only an obvious and immediate gain from marketing, but it is in no way the only possible gain.

The article itself identifies the bringing forward of a cycle of upgrades as customers with 64-bit Windows think that their old applications, which they might otherwise have persisted with, look long in the tooth, being 32-bit on 64-bit.

But any gain suffices, even if it has no readily quantifiable dollar value. That was one point to my quoting Mark Russinovich. I might just as well have quoted you:

The sooner users are on 64-bit the better for them, MS benefits indirectly by users using the better system and thus having a better experience overall and better ecosystem, ...
Click to expand...

Indeed, many of my critics make my point for me, evidently without wanting to and apparently without realising it.

... but it does not rake in dough from 64-bit versions it does not get from 32-bit versions.
Click to expand...

I don't say it does and my argument doesn't depend on it even implicitly. Any desirable movement in a market can be the target of marketing, wouldn't you think?

In the paragraphs before I _suggest_ a marketing ruse, I establish that the technical reasons presented by Microsoft do not themselves explain the implementation and indeed that some of Microsoft's technical reasons are absurd. It's hardly unreasonable that I then ask what else can have informed Microsoft's implementation.

Commentators are easily found who say that Microsoft has done well to limit 32-bit Windows Vista to 4GB by licensing, else the market would not move to 64-bit Windows as fast as is desirable for one reason or another. Indeed, they typically make this case for a marketing ruse more strongly than I do. Is your objection to marketing or to ruse?
 
Tawnos said:
You asked "how do you get ... what Microsoft actually did?" I showed a way that could have happened.
Click to expand...

Surely the context was plainly of how he gets from A to B in what he's saying. That he may have been present when others got from A to B is no use to anyone if he's not showing any reasoning to support how A implies B.

Indeed, what you've presented is a classic ad hominem argument to support someone's assertion not by adding to his reasoning but by appealing to their person, in this case, it seems, to his authority from his possibly having been present at an event that anyway wouldn't tell us how A implies B.

Is it really that hard to get, or are you just butthurt that someone disagrees with your article, assumptions, and conclusions, and want to eFight about it?
Click to expand...

I'm not the slightest bit troubled that someone disagrees. As I've written in other messages, I stand to learn. If nothing else, I may see that text needs to be revised to avoid misinterpretation. It's always possible that the criticism may actually be well-founded.

Against that, if someone disagrees with what he says is an assumption but I don't think I've made anything like that assumption, then yes, it hurts if it's trouble to get out of them a logical argument in support of their criticism. Supporting your criticisms, and making time to do so, is just minimal responsibility that you acquire when you write critically. You may disagree with that, but it's nothing that I don't expect of myself when a target of my criticism writes to me to ask me to support what I've written about him.
 
mikeblas said:
You're announcing that you're taking away something that is of no value, and backing it up with an ad homenim attack. Really, why bother?
Click to expand...

I'd happily let you have the last word but you go too far. I doubt I have made an ad hominem attack in my adult life. An ad hominem attack occurs where another's argument is rebutted not by examining his reasoning but by playing to the man, most notably by asking after his authority or motivation). Mike, you have given no reasoning: where do you think is the ad hominem attack?

Indeed, though you talk of my rhetoric, I have been wondering all along at how your only use for words is to avoid giving your reasoning.

You criticise me for an assumption that I surely don't make, but despite several invitations, you never did explain where I make your alleged assumption. You criticise me for not examining compromises, yet I have at the very least started that examination by looking at alternatives to see if what Microsoft did actually is implied by what Microsoft says - and all your support for your criticism comes down to assertions and talk of such things as unnamed other factors. I know I've got carried away in hoping against hope that you may actually offer some reasoning, but the record is plain that you haven't offered any.
 
Ok, it's a market ruse. It's MS grand evil plan to get users off limited, shitty 32-bit and on to something modern and appropiate for this decade. you got me. Anyway, I also take exception with your technical arguments. (I assume your the same geoff that wrote the article) You said you could not get certain versions of nvidia graphics drivers to work with the 4GB+ hack, before that you said a driver writer would basically have to break this on purpose because it's no different than the normal situation to use over 4GBs when running PAE. So then, either Nvidia broke their driver on purpose :)rolleyes:) or you are wrong and it is not something a driver writer would have to break on purpose. Additionally, if you managed to find a common device manufacturer's drivers wouldn't run in your limited tests, what do you think MS found in their tests of thousands of device drivers and other software? You also said that later versions of Nvidia drivers worked, ok, well, maybe at the time MS made this decision, there were no nvidia drivers that worked with over 4GBs in PAE mode? That would mean right off the bat, 30% of users would get a crashed computer if they tried this. All this, and you don't think even a little bit, "hey maybe PAE 32-bit for over 4GBs of ram isn't the best idea..." ?
 
***DISCLAIMER*** I didn't read past the 1st page. Doing some work and just wanted to answer.

Some people don't take in to consideration the following:

* Most computers aren't 64bit ready
* People don't have the cash to shell out for a 64bit ready system
* People do not want to shell out more cash for the OS

I'm lucky enough to have a 64bit ready system that I got from Dell almost 3 years ago (as I went with the XPS series). I'm also lucky that I get this software for free through my occupation. So always keep in mind, there aren't others as fortunate.
 
Geoff Chappell said:
I'd happily let you have the last word but you go too far. I doubt I have made an ad hominem attack in my adult life. An ad hominem attack occurs where another's argument is rebutted not by examining his reasoning but by playing to the man, most notably by asking after his authority or motivation). Mike, you have given no reasoning: where do you think is the ad hominem attack?
Click to expand...
I quoted it above; you're writing me off as a big fish in a small pond, where logical formulation of arguments doesn't matter.

Geoff Chappell said:
Indeed, though you talk of my rhetoric, I have been wondering all along at how your only use for words is to avoid giving your reasoning.
Click to expand...
I've provided my reasoning at least once, but you've refused to accept it. Since you're so positive in the pat that you've reasoned out for yourself, it only follows that you wouldn't entertain the possibility of factors which are unknown to you.
 
Geoff Chappell said:
Surely the context was plainly of how he gets from A to B in what he's saying. That he may have been present when others got from A to B is no use to anyone if he's not showing any reasoning to support how A implies B.

Indeed, what you've presented is a classic ad hominem argument to support someone's assertion not by adding to his reasoning but by appealing to their person, in this case, it seems, to his authority from his possibly having been present at an event that anyway wouldn't tell us how A implies B.
Click to expand...
Actually, the logical fallacy you'd be looking for is appeal to authority, not ad hominem (or ad hominem tu quoque). Even with what you seek, though, you fail to present a valid case. I stated that he might get what MS actually did because he was there. He was part of Microsoft. Unlike the appeal to authority fallacy, his presence and choices within the company indicate direct knowledge of how something occurred., it's not simply a citation because he's an authority in the field.

blah blah blah ...
But any gain suffices, even if it has no readily quantifiable dollar value.
Click to expand...
Ultimately, the only gain that matters is the financial. Feel-good doesn't put food on the table. The product must produce enough customer value that they are enticed to spend their hard earned dollars on the product. Without that monetary impact, no other gain has meaning.
 
Duvi said:
***DISCLAIMER*** I didn't read past the 1st page. Doing some work and just wanted to answer.

Some people don't take in to consideration the following:

* Most computers aren't 64bit ready
* People don't have the cash to shell out for a 64bit ready system
* People do not want to shell out more cash for the OS

I'm lucky enough to have a 64bit ready system that I got from Dell almost 3 years ago (as I went with the XPS series). I'm also lucky that I get this software for free through my occupation. So always keep in mind, there aren't others as fortunate.
Click to expand...

Most computers that aren't 64-bit ready, probably also can't take more than 4GBs of Ram. Between early bioses and motherboards, this adds up.

A 64-bit OS does not cost any money (save a $10 shipping fee for Vista) if you're using Vista or Win 7. I just don't think there's going to be many people with P4s running XP that want/need 8GBs of memory.

Additionally, MS has to get all drivers certified for operation above 4GBs in 32-bit PAE mode, and what does MS do? If they require this, that costs a lot of money and resources to IHVs, if they don't require it, nobody gives a damn and nothing changes.

Also it would have killed the superior solution, 64-bit, and we'd be stuck with the 32-bit PAE limit for a long time (64GBs or so) - instead, with 64-bit, it's easy to build a server with 2TBs of ram and almost all devices will work with it (assuming you have the wad of cash for all that ram), it's a much better situation. Let it die already.
 
You must log in or register to reply here.
Back
Top