Talk:Physical Address Extension

From Wikipedia, the free encyclopedia

 This article is part of WikiProject Microsoft Windows, a WikiProject devoted to maintaining and improving the informative value and quality of Wikipedia's many Microsoft Windows articles.
??? This article has not yet received a rating on the assessment scale.
??? This article has not yet received a rating on WikiProject Microsoft Windows's importance scale.

Contents

[edit] Plural or singular?

Shouldn't this article be at Physical Address Extensions? Documents I've seen from both AMD and Intel always seem to refer to it in plural. JulesH 17:39, 8 October 2006 (UTC)

Microsoft seems to use the singular: http://www.microsoft.com/whdc/system/platform/server/PAE/PAEdrv.mspx 198.20.50.34 14:50, 20 August 2007 (UTC)
Intel seems to use the singular: http://www.intel.com/design/processor/manuals/253668.pdf Guy Harris 01:29, 21 August 2007 (UTC)

[edit] Page Size Extension

Shouldn't there be a section contrasting and comparing Page Size Extension to Physical Address Extension? Dsf7183 10:22, 15 November 2007 (PST)

Probably. Be bold! Jeh (talk) 00:07, 10 January 2008 (UTC)

[edit] Windows: paragraph re. Address Windowsing Extensions removed

I just removed this graf:

Access to physical memory greater than 4 GiB by 32-bit applications is by means of Address Windowing Extensions (AWE), which works something like Expanded memory (EMS) and Extended Memory (XMS) on 16-bit systems.

Reason - AWE really is unrelated to the PAE mechanism. You can use AWE APIs, and they will work (if you don't ask for much RAM), on a system not running in PAE mode. Hence it is not limited to accessing "physical memory greater than 4 GiB." It is just that only on systems with a LOT of physical memory, and will an app have much reason to use AWE. Jeh (talk) 00:05, 10 January 2008 (UTC)

[edit] Workaround for the XP SP2 32-bit physical address limit? No, sorry.

The Windows section previously contained

If PAE is enabled in the normal way, this compatiblity hack is disabled. See [1].

The link had been there for some time but the sentence preceding it ("If PAE is enabled...") was new. Anyway, I'm afraid that both are incorrect. As someone at the end of that very blog page stated, you can try all the ways and combinations of ways to enable PAE you like; you still can't access RAM above the 4 GiB address boundary. And because of conflicts with PCI device space below the 4 GiB point, you can't access all of 4 GiB of RAM on these systems either. Jeh (talk) 03:17, 10 January 2008 (UTC)

From [2]:
"If you add the /PAE switch, you get the normal PAE behavior, and all bets are off. Of course, this is exactly what you want in the server space; after all, you got that extra RAM for a reason, right? Also note that the properties aren’t transitive. While both /PAE and /NoExecute use the same kernel file (ntkrnlpa.exe or ntkrpamp.exe) and address translation mechanism, you need both switches in place to enable both features." - Yuhong (talk) 21:15, 12 January 2008 (UTC)
Sorry but IT DOES NOT WORK AS THAT PAGE CLAIMS. I've tried it myself -- it doesn't work. Jeh (talk) 21:56, 12 January 2008 (UTC)
BTW, not all chipsets and BIOSes support more than 32-bit physical address space. If it doesn't work on 64-bit OSes that must be the cause. —Preceding unsigned comment added by 207.34.170.129 (talk) 21:59, 12 January 2008 (UTC)
Sorry, but no. I have here a machine with 8 GiB RAM installed, and all is available on a 64 bit OS. This hack does not work to allow Vista x86 or XP SP2 x86 to see more than about 3.2 GiB. Jeh (talk) 22:03, 12 January 2008 (UTC)
How about on Server 2003 SP1 Enterprise x86? - Yuhong (talk) 22:05, 12 January 2008 (UTC)
Server was never subject to the "cripple"! Only the client OSs. If /pae is necessary to see RAM addrs > 4 GB on Server 2003, even if /noexecute=(not-disable) is already there, that is an interesting point, but it deosn't help the folks running client OSs who can't see all their RAM. Jeh (talk) 22:12, 12 January 2008 (UTC)
Re your most recent change to the page... this I have not tested (yet), but I don't believe the /PAE switch is needed on Server as long as /noexecute is there with anything other than /noexecute=disable . Do you have a reference other than your own blog page? Jeh (talk) 22:20, 12 January 2008 (UTC)
I don't think that is true. Please test. - Yuhong (talk) 19:22, 13 January 2008 (UTC)
I will. It won't be today, but likely within the week! Jeh (talk) 22:30, 13 January 2008 (UTC)

[edit] Not windows specific...

Why on earth is this "WikiProject Microsoft Windows", PAE is used by Linux as well and several other operating systems, it is hardware feature not something windows specific. --83.181.53.188 (talk) 22:38, 3 March 2008 (UTC)

It can be part of other projects too. Nothing says an article has to be part of just one project! Jeh (talk) 04:23, 4 March 2008 (UTC)

[edit] Illustrations for PAE

I made some drawings illustrating how PAE (and non-PAE paging on x86) works. Perhaps you find them useful for the article?

--81.27.124.124 (talk) 20:32, 26 March 2008 (UTC) (RokerHRO)

Good work. Could you fix this one, it says 4KB instead of 2MB? --Kubanczyk (talk) 09:31, 27 March 2008 (UTC)
I would suggest that low addresses or offsets or index values should go at the top of the page and high at the bottom. This is the way e.g. debuggers display memory, and is also the way e.g. structures are defined in languages such as C: Offsets start at 0 at the top of the page, or output, or the first line of the struct definition, and increase in value as you go "down" te page, or etc. Jeh (talk) 15:14, 27 March 2008 (UTC)

[edit] the footnotes are ambiguous

"1 GB = 1024 MB" is an ambiguous statement. If you click on the link you find that MB has three different definitions.

Not ambiguous and you made the edit on May 5 to not use IEC. The refs make the disambiguation unambiguous because they include GB MB and KB. DavidPaulHamilton (talk) 16:02, 8 May 2008 (UTC)
In the interests of trying to stop the numerous reverts I added extra disambiguation for 1 GB = 1024 MB to say "1 GB = 1024 MB ; 1 MB = 1024 KB ; 1 KB = 1024 B" and also added similar for MB. KB and PB don't need extra disambiguation as it is already precisely clear what they mean from the footnote. This article does not currently mention any file sizes using these prefixes so it is unambiguous and precise to not include any other explanatory text in the footnotes regarding binary or decimal use. I think it would be too bloated to include "This article uses prefixes in a binary sense" in each footnote since the exact quantities are already disambiguated and the article doesn't change between binary and decimal use. However, if you want to add that text Thunderbird2 then you can. Fnagaton 16:48, 8 May 2008 (UTC)