Actually, the whole thing seems to be really exciting - I cannot recall any “LP64 vs LLP64” flame war in this NG in so far…
Anton Bassov
Actually, the whole thing seems to be really exciting - I cannot recall any “LP64 vs LLP64” flame war in this NG in so far…
Anton Bassov
>
Actually, the whole thing seems to be really exciting - I cannot
recall any
“LP64 vs LLP64” flame war in this NG in so far…
Neither can I. I think the question was asked a while ago, and it went
something like:
OP: Is Windows LP64 or LLP64?
Respondant: It’s LLP64
OP: OK Thanks.
James
“App compat shimming works to a certian degree, but it is not an absolute fix.”
Doron,
I suggest that Windows should only expose new functionality of version x (like user interface/graphics goodies) and behaviors to applications marked as “Version x”. At the same time, for an application marked “Version X”, any compat shims for previous versions should be disabled. Thus, if an ISV wants new goodies, they will need to get their shit together.
We have that already, it is called a manifest file
d
-----Original Message-----
From: xxxxx@lists.osr.com [mailto:xxxxx@lists.osr.com] On Behalf Of xxxxx@broadcom.com
Sent: Wednesday, February 17, 2010 2:44 PM
To: Windows System Software Devs Interest List
Subject: RE:[ntdev] Is Windows LP64?
“App compat shimming works to a certian degree, but it is not an absolute fix.”
Doron,
I suggest that Windows should only expose new functionality of version x (like user interface/graphics goodies) and behaviors to applications marked as “Version x”. At the same time, for an application marked “Version X”, any compat shims for previous versions should be disabled. Thus, if an ISV wants new goodies, they will need to get their shit together.
NTDEV is sponsored by OSR
For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars
To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer
Doron Holan wrote:
We have that already, it is called a manifest file
I have a rather lengthy grumpy speech about those, too, but it’s
probably not good for my blood pressure to rant twice in two days.
–
Tim Roberts, xxxxx@probo.com
Providenza & Boekelheide, Inc.
I never said it was a great solution, just that there was an attempt to address the problem that exists today
d
-----Original Message-----
From: xxxxx@lists.osr.com [mailto:xxxxx@lists.osr.com] On Behalf Of Tim Roberts
Sent: Wednesday, February 17, 2010 2:53 PM
To: Windows System Software Devs Interest List
Subject: Re: [ntdev] Is Windows LP64?
Doron Holan wrote:
We have that already, it is called a manifest file
I have a rather lengthy grumpy speech about those, too, but it’s
probably not good for my blood pressure to rant twice in two days.
–
Tim Roberts, xxxxx@probo.com
Providenza & Boekelheide, Inc.
NTDEV is sponsored by OSR
For our schedule of WDF, WDM, debugging and other seminars visit:
http://www.osr.com/seminars
To unsubscribe, visit the List Server section of OSR Online at http://www.osronline.com/page.cfm?name=ListServer
> You could ask the Itanium designers who upped the page size many
years ago.
Windows uses 8K pages on Itanium. From what I heard, larger sizes
were tried at one point and they didn’t provide much benefit - in fact,
the overall effect was negative due to extra IO costs.
Things might be different today but I would be surprised if switching
from 4K pages to say 16K significantly improved things, except for
very specific workloads that could benefit from reduced TLB miss
rate… But those workloads (huge database servers etc) can already
use large pages, and even with 16 MB large pages on IA64 the typical
gains are rather small (e.g. a few percentage points on various
TPC-style benchmarks, from what I’ve seen).
And the large page size add on you alluded to is further
proof 4K is too small, but we are stuck with them.
Large pages are not a general purpose replacement for regular pages,
so the fact that x64 has 2 MB large pages doesn’t necessarily mean
that the regular 4K pages are too small .
–
Pavel Lebedinsky/Windows Fundamentals Test
This posting is provided “AS IS” with no warranties, and confers no rights.
“Pavel Lebedinsky” wrote in message
news:xxxxx@ntdev…
>> You could ask the Itanium designers who upped the page size many
>> years ago.
>
> Windows uses 8K pages on Itanium. From what I heard, larger sizes
> were tried at one point and they didn’t provide much benefit - in fact,
> the overall effect was negative due to extra IO costs.
>
> Things might be different today but I would be surprised if switching
> from 4K pages to say 16K significantly improved things, …
Those Digital designers were geniuses, hands down. Even 30 years later
it’s hard to improve their directions.
– pa
Tim Roberts wrote:
xxxxx@osr.com wrote:
> [quote]
> Tim says that Windows is ILP64 and Peter says LLP64.
> [/quote]
>No, Tim never said Windows was ILP64. Those acronyms are attributes of
a COMPILER, not of an operating system. Tim said Visual C++ was ILP64.> It’s LLP64.
>Yes. It used to be ILP64, because Visual C++ was late in getting “long
long” support. Now, it’s LLP64.
Which release of 64-bit VC++ had ‘int’ and ‘long’ as 64-bit types? I’m
surprised not to have heard of that before, especially if they
subsequently changed them back to 32-bit. I don’t think ILP64 ever got
into any of the DDK compilers.
I thought 64-bit VC++ went from IL32P64 to IL32LLP64, with some
compiler-specific types also providing 64-bit integers in both cases.
J. J. Farrell wrote:
Tim Roberts wrote:
>
> Yes. It used to be ILP64, because Visual C++ was late in getting “long
> long” support. Now, it’s LLP64.Which release of 64-bit VC++ had ‘int’ and ‘long’ as 64-bit types? I’m
surprised not to have heard of that before, especially if they
subsequently changed them back to 32-bit. I don’t think ILP64 ever got
into any of the DDK compilers.I thought 64-bit VC++ went from IL32P64 to IL32LLP64, with some
compiler-specific types also providing 64-bit integers in both cases.
Well, bend me over and spank me silly. I misread the chart. I’m
completely wrong.
Everyone else is right. 64-bit VC++ has always been LLP64, with the one
exception that the “long long” type had to be spelled “__int64” until
the 21st Century. It never has been ILP64, and neither has Unix.
–
Tim Roberts, xxxxx@probo.com
Providenza & Boekelheide, Inc.