Myth: In order to use AWE, you must enable PAE

re: Myth: In order to use AWE, you must enable PAE

Paul Wilson — Thu, 19 Aug 2004 15:46:00 GMT

Its amazing how many people spend more time researching what others say instead of just testing things for themselves. :)

re: Myth: In order to use AWE, you must enable PAE

Chris Walker — Thu, 19 Aug 2004 16:35:00 GMT

Sidney Marshall of Dartmouth Timesharing (DTSS) fame used to say "Computers is an experimental science".

re: Myth: In order to use AWE, you must enable PAE

Miles Archer — Thu, 19 Aug 2004 17:19:00 GMT

Caveat: I don't know anything about PAE and AWE. And I believe you when you say that they are it's not required.

However, proving that one test case works does not prove that it works for the general case or that there are subtle problems that will occur over time.

re: Myth: In order to use AWE, you must enable PAE

Raymond Chen — Thu, 19 Aug 2004 17:32:00 GMT

True, you can't use an experiment to prove that something will always happen, but you can use it to disprove that something *never* happens or *requires* something else.

re: Myth: In order to use AWE, you must enable PAE

Jerry Dennany — Thu, 19 Aug 2004 18:00:00 GMT

Well, it is documented that AWE does require PAE. "Without PAE, AWE cannot reserve memory in excess of 4 GB." [1]

So, I wouldn't call it a myth.

[1] http://support.microsoft.com/?kbid=283037

re: Myth: In order to use AWE, you must enable PAE

Raymond Chen — Thu, 19 Aug 2004 18:09:00 GMT

It doesn't say that AWE requires PAE. It says that AWE requires PAE in order to allocate more than 4GB of physical memory. But that's just stating the obvious. You need PAE in order to access more than 4GB of physical memory, AWE or no AWE.

re: Myth: In order to use AWE, you must enable PAE

Greg Page — Thu, 19 Aug 2004 19:02:00 GMT

I wrote that article, so yeah, it should agree with exactly with what Raymon says. If it doesn't let me know, and it will be fixed post haste. and I blame any inaccuracies on editors who change things I write after the fact.

re: Myth: In order to use AWE, you must enable PAE

Jerry Dennany — Thu, 19 Aug 2004 19:22:00 GMT

And is there a valid use for AWE without PAE?

re: Myth: In order to use AWE, you must enable PAE

Greg Page — Thu, 19 Aug 2004 19:48:00 GMT

The same uses as with it, I would imagine.

And one of the most benficial things is, developers can test AWE apps without having to enable PAE.

re: Myth: In order to use AWE, you must enable PAE

Raymond Chen — Thu, 19 Aug 2004 19:54:00 GMT

One use is if you want to allocate non-swappable memory.

re: Myth: In order to use AWE, you must enable PAE

Peter Bindels — Thu, 19 Aug 2004 20:59:00 GMT

Can't it use PSE to access beyond 4GB using 4M pages above that area? I thought PSE was meant for that, but didn't succeed because the 4M pages were not as usable as Intel had hoped. The top few bits were in the bottom of the address part that wasn't used in the 4M page entries.

re: Myth: In order to use AWE, you must enable PAE

RJ — Thu, 19 Aug 2004 23:49:00 GMT

I think giving these functions a fancy name (AWE) leads to confusion. I've used VirtualAllocEx several times to inject code into other processes. Never knew it was part of AWE.

re: Myth: In order to use AWE, you must enable PAE

Raymond Chen — Thu, 19 Aug 2004 23:52:00 GMT

Please take another look at the documentation. VirtualAllocEx is AWE only if you pass the MEM_PHYSICAL flag.

Ah, that /3GB switch again?

You Had Me At EHLO... — Fri, 20 Aug 2004 20:02:00 GMT

The story of the 3GB server switch

Savage Nomads — Fri, 20 Aug 2004 20:44:00 GMT

  As Evan already mentioned on his blog, Raymond Chen has a great series on /3GB switch on his blog. What is really cool is that Raymond takes on some myths about the /3GB switch and  the fact that he...