Re: IA32 execution layer

Re: IA32 execution layer

[Matt Domsch]

> Duraid> Hi all, It seems that an IA32 execution layer is going to make
> Duraid> its way onto upcoming ia64 windows systems. Is anyone thinking
> Duraid> of adding support for such a layer to linux-ia64?
> 
> It's already there.  If you have the appropriate libraries, IA32 linux
> programs run perfectly well.

Popular I/T press is touting some new work by Intel to emulate x86 
instructions on IPF, which (per the articles) is faster than running it on 
the x86 hardware in the IPF.  I've gotten this query internally too.

-- 
Matt Domsch
Sr. Software Engineer
Dell Linux Solutions www.dell.com/linux
Linux on Dell mailing lists @ http://lists.us.dell.com

IA32 execution layer

[Duraid Madina]

Hi all,

	It seems that an IA32 execution layer is going to make its way onto 
upcoming ia64 windows systems. Is anyone thinking of adding support for 
such a layer to linux-ia64?

	Just curious,

	Duraid

IA32 execution layer

[Peter Chubb]

>>>>> "Duraid" = Duraid Madina <duraid@octopus.com.au> writes:

Duraid> Hi all, It seems that an IA32 execution layer is going to make
Duraid> its way onto upcoming ia64 windows systems. Is anyone thinking
Duraid> of adding support for such a layer to linux-ia64?

It's already there.  If you have the appropriate libraries, IA32 linux
programs run perfectly well.

Peter C

Re: IA32 execution layer

[Jeff Woods]

Matt Domsch wrote:
>>>Hi all, It seems that an IA32 execution layer is going to make its way 
>>>onto upcoming ia64 windows systems. Is anyone thinking adding support 
>>>for such a layer to linux-ia64?
>
>Popular I/T press is touting some new work by Intel to emulate x86 
>instructions on IPF, which (per the articles) is faster than running it on 
>the x86 hardware in the IPF.  I've gotten this query internally too.

Intel made some press releases in April 2003 about something they were 
calling "BTRANS" which is designed to replace/enhance the IA32 hardware on 
Itanium processors by emulating some of the IA32 instructions in 
software.  It is supposed to run faster than the "300MHz Pentium Pro" 
effectively in the McKinley processors plus be able to execute instructions 
(SSE, etc.) that the embedded IA32 hardware doesn't support.

At LinuxWorld/SF last month I cornered techs from Intel, Red Hat and SuSE 
specifically to ask about BTRANS.  I discovered that it's now called "IA-32 
Execution Layer" and the story from both Intel and SuSE was that it's being 
incorporated into the "next release" of major Linux distributions for IA64 
(e.g., SLES 8.1 or 9.0).  The Red Hat guy said something about it "isn't 
open source" so they weren't planning to include it.  After relating Red 
Hat's comments to the Intel folks, they simply said they would see that the 
appropriate folks at Intel talk to Red Hat about it.

http://www.intel.com/pressroom/archive/releases/20030630comp.htm
[or: http://tinyurl.com/mtxw ]
>>In addition, later this year Intel plans to introduce [...] the IA-32 
>>Execution Layer, a software technology that will enhance the performance 
>>of 32-bit applications running on the Itanium processor family-based 
>>platforms. Prior to the IA-32 Execution Layer release, Microsoft will 
>>offer a preview of the technology to end users who participate in the 
>>Windows Server 2003 Service Pack 1 beta program.

http://www.intel.com/ebusiness/pdf/prod/itanium/dp_qrg.pdf
[or: http://tinyurl.com/mtyy ]
>>IA-32 Application Support: Today, like all Itanium 2 processors, Itanium 
>>2 processor 1.40 GHz with 1.5M L3 cache and Low Voltage Itanium 2 
>>processor offer support for IA-32 applications. To further enhance this 
>>capability Intel will soon introduce a new technology called IA-32 
>>Execution Layer.††

>>†† Intel and Microsoft plan to introduce the IA-32 Execution Layer with 
>>Windows Server 2003 Service Pack 1

http://www.intel.com/ebusiness/pdf/prod/itanium/ds022001.pdf
[or: http://tinyurl.com/mtz1 ]
>>IA-32 Application Support: Today, all Itanium 2 processors offer support 
>>for IA-32 applications; a new technology called IA-32 Execution Layer 
>>will further enhance this capability in 2H ’03.

Of course, maximum performance requires native execution.  Compiling 
performance-critical code with an optimizing native compiler is the proper 
solution.

--
Jeff Woods <kazrak+kernel@cesmail.net>