Text Size

K10: Barcelona, Shanghai, Quad-Core Opteron, Phenom

Intel/AMD settlement and Intel software

AMD's latest line-up of CPUs including Shanghai Opteron, Phenom and Phenom II X4 & X3, Athlon X2.

Intel/AMD settlement and Intel software

Postby agner » Mon Dec 21, 2009 11:09 am

For years, Intel have made compilers and function libraries that work poorly on AMD and VIA processors. Apparently, the new Intel/AMD settlement requires that Intel change this practice.

First, the technical background:
Intel produces some of the best softare function libraries available for many technical and scientific applications. These function libraries have multiple versions of each function, each optimized for a processor and instruction set, for example SSE2, SSE3, etc. The system includes a so-called CPU dispatcher: A piece of code that detects which type of CPU it is running on and choosing the optimal version of the function for that CPU. However, the Intel CPU dispatcher does not only check which instruction set is supported by the CPU - it also checks if the vendor ID string says 'GenuineIntel'. If the CPU is not from Intel then, in many cases, it will run the slowest possible version of the function, even if the CPU is fully compatible with a better version. The same applies to code built with an Intel compiler. If the programmer chooses support for a specific instruction set or some form of automatic CPU dispatching then the compiled program will work suboptimally or not at all on AMD and VIA processors.

This behavior is not transparent to the programmer. A program that has been built with an Intel compiler or using an Intel function library may, unbeknownst to the programmer, work less than optimally on non-Intel processors.
In my C++ manual, I have shown how it is possible to replace the Intel CPU dispatcher and proven that the Intel software can run faster on an AMD processor if you circumvent this "cripple AMD" feature (http://www.agner.org/optimize/#manual_cpp). Others have shown that a common benchmarking program shows better performance on VIA chips if you change the vendor ID string (http://arstechnica.com/hardware/reviews/2008/07/atom-nano-review.ars/6).

Apparently, AMD have put a feature to manipulate the vendor ID string into their next CPUs for the same purpose (http://forums.amd.com/devforum/messageview.cfm?catid=203&threadid=95754). [Correction 2009-12-28: No they haven't. It was just a proposal.]

Now, back to the Intel/AMD settlement:
2.3 TECHNICAL PRACTICES

Intel shall not include any Artificial Performance Impairment in any Intel product or require any Third Party to include an Artificial Performance Impairment in the Third Party’s product. As used in this Section 2.3, “Artificial Performance Impairment” means an affirmative engineering or design action by Intel (but not a failure to act) that (i) degrades the performance or operation of a Specified AMD product, (ii) is not a consequence of an Intel Product Benefit and (iii) is made intentionally to degrade the performance or operation of a Specified AMD Product. For purposes of this Section 2.3, “Product Benefit” shall mean any benefit, advantage, or improvement in terms of performance, operation, price, cost, manufacturability, reliability, compatibility, or ability to operate or enhance the operation of another product.

In no circumstances shall this Section 2.3 impose or be construed to impose any obligation on Intel to (i) take any act that would provide a Product Benefit to any AMD or other non-Intel product, either when such AMD or non-Intel product is used alone or in combination with any other product, (ii) optimize any products for Specified AMD Products, or (iii) provide any technical information, documents, or know how to AMD.

Was this clause written in order to prevent Intel from deliberately making software that works poorly on AMD processors? It looks like "Intel product" here refers to Intel compilers and function libraries, and "any Third Party" refers to programmers who use these compilers and function libraries.

Am I right that Section 2.3 refers to software? I don't see that it would make much sense in relation to hardware because Intel hardware can't be used in connection with AMD chips anyway.

For years I have criticized Intel for making software with hidden "discrimination" features and they have tried to explain away what they were doing. AMD have never made any comment on this issue to my knowledge, but it appears that they have won a victory here. (Or have they? There seems to be many loopholes in Section 2.3).

I will certainly check the next version of Intel's compiler and function libraries.
Last edited by agner on Mon Dec 28, 2009 4:33 pm, edited 1 time in total.
agner
 
Posts: 22
Joined: Fri Dec 11, 2009 8:47 am

Re: Intel/AMD settlement and Intel software

Postby xigh » Mon Dec 21, 2009 11:47 am

agner wrote:Apparently, AMD have put a feature to manipulate the vendor ID string into their next CPUs for the same purpose (http://forums.amd.com/devforum/messageview.cfm?catid=203&threadid=95754).


There is no need to add such feature as you can trap cpuid instruction with AMD-V. What a waste of die space ... :(
xigh
 
Posts: 17
Joined: Fri May 02, 2008 9:14 pm

Re: Intel/AMD settlement and Intel software

Postby muziqaz » Mon Dec 21, 2009 12:03 pm

agner, depending on AMDs programmers efficiency, next version of Intel compiler should be shit free ;)
this version is still terrible for amd when it comes to SSEx instruction deployment.

I just cannot wait for the release of 'fixed' compiler and someone doing the comparison with benchmarks compiled with older version and newer. Sure that someone should be not from those stupid Intel paid sites, but someone who wants to see the difference. And the the difference should be big :)
Proud member of Folding@Home project.
http://folding.stanford.edu
Image
User avatar
muziqaz
K8 Athlon 64 X2 (Manchester) Elite Boarder
K8 Athlon 64 X2 (Manchester) Elite Boarder
 
Posts: 2823
Joined: Sun Sep 02, 2007 3:34 am
Location: London UK

Re: Intel/AMD settlement and Intel software

Postby agner » Mon Dec 21, 2009 12:42 pm

muziqaz wrote:I just cannot wait for the release of 'fixed' compiler and someone doing the comparison with benchmarks compiled with older version and newer. Sure that someone should be not from those stupid Intel paid sites, but someone who wants to see the difference. And the the difference should be big :)

You should never trust a benchmarking program unless it is open source and compiled with a known compiler. It is too easy to fill it with hand-picked instructions that you know run faster on a particular CPU than on its competitor.

If the benchmark program is open source then you can compile it with different compilers. And for the Intel compiler, you can compile it with and without the workaround CPU dispatcher published in my C++ manual (http://www.agner.org/optimize/#manual_cpp).
agner
 
Posts: 22
Joined: Fri Dec 11, 2009 8:47 am

Re: Intel/AMD settlement and Intel software

Postby muziqaz » Mon Dec 21, 2009 12:44 pm

That is why I am saying, that the comparison should be done by someone who wants to find the difference which will be there I am sure of that :)
Proud member of Folding@Home project.
http://folding.stanford.edu
Image
User avatar
muziqaz
K8 Athlon 64 X2 (Manchester) Elite Boarder
K8 Athlon 64 X2 (Manchester) Elite Boarder
 
Posts: 2823
Joined: Sun Sep 02, 2007 3:34 am
Location: London UK

Re: Intel/AMD settlement and Intel software

Postby muziqaz » Mon Dec 21, 2009 12:45 pm

P.S. in open source community we already see that amd is very strong, because there is no shit in compilers :)
Proud member of Folding@Home project.
http://folding.stanford.edu
Image
User avatar
muziqaz
K8 Athlon 64 X2 (Manchester) Elite Boarder
K8 Athlon 64 X2 (Manchester) Elite Boarder
 
Posts: 2823
Joined: Sun Sep 02, 2007 3:34 am
Location: London UK

FTC: Federal Trusted Compiler

Postby Boundless » Mon Dec 21, 2009 12:54 pm

agner: > For years, Intel have made compilers and function libraries
> that work poorly on AMD and VIA processors. Apparently, the
> new Intel/AMD settlement requires that Intel change this practice.


Not just. The FTC is on spIntel's case about this too:
Proposed Remedy #7 in the Intel Complaint.

7. Requiring that, with respect to those Intel customers that
purchased from Intel a software compiler that had or has the
design or effect of impairing the actual or apparent performance
of microprocessors not manufactured by Intel (“Defective Compiler”),
as described in the Complaint:
a. Intel provide them, at no additional charge, a substitute compiler
that is not a Defective Compiler;
b. Intel compensate them for the cost of recompiling the software they
had compiled on the Defective Compiler and of substituting, and
distributing to their own customers, the recompiled software for
software compiled on a Defective Compiler; and
c. Intel give public notice and warning, in a manner likely to be
communicated to persons that have purchased software compiled
on Defective Compilers purchased from Intel, of the possible need
to replace that software.

_______
"Defective" is an interesting choice of words.
User avatar
Boundless
K8 Athlon 64 (Orleans) Expert Boarder
K8 Athlon 64 (Orleans) Expert Boarder
 
Posts: 2000
Joined: Mon Mar 01, 2004 4:40 pm

Re: Intel/AMD settlement and Intel software

Postby gruffi » Mon Dec 21, 2009 1:04 pm

agner wrote:Apparently, AMD have put a feature to manipulate the vendor ID string into their next CPUs for the same purpose (http://forums.amd.com/devforum/messageview.cfm?catid=203&threadid=95754).

That would be a nice feature. 8)
Tradition is not holding the ashes but passing the flame.
User avatar
gruffi
K6-III Fresh Boarder
K6-III Fresh Boarder
 
Posts: 252
Joined: Sun Jun 21, 2009 12:39 pm

Re: Intel/AMD settlement and Intel software

Postby muziqaz » Mon Dec 21, 2009 1:14 pm

is it true, that crysis uses only 3dnow instructions on my 965BE ? :D
imagine if it used sse3, amd chips would murder intel chips easily.
Proud member of Folding@Home project.
http://folding.stanford.edu
Image
User avatar
muziqaz
K8 Athlon 64 X2 (Manchester) Elite Boarder
K8 Athlon 64 X2 (Manchester) Elite Boarder
 
Posts: 2823
Joined: Sun Sep 02, 2007 3:34 am
Location: London UK

Re: Intel/AMD settlement and Intel software

Postby The_Ghost » Tue Dec 22, 2009 3:22 am

interesting topic :mrgreen:
Image
The_Ghost
K10 Opteron (Barcelona) Administrator
K10 Opteron (Barcelona) Administrator
 
Posts: 25675
Joined: Sun Feb 29, 2004 5:06 pm

Re: Intel/AMD settlement and Intel software

Postby wuttz » Tue Dec 22, 2009 4:27 am

muziqaz wrote:is it true, that crysis uses only 3dnow instructions on my 965BE ? :D
imagine if it used sse3, amd chips would murder intel chips easily.


someone profile and dismantle that benchmark yet? :wink: :wink: :wink:
wuttz
K8 Athlon 64 X2 (Manchester) Elite Boarder
K8 Athlon 64 X2 (Manchester) Elite Boarder
 
Posts: 2988
Joined: Sat Aug 08, 2009 6:48 pm
Location: Pearland, Texas

Re: Intel/AMD settlement and Intel software

Postby muziqaz » Tue Dec 22, 2009 8:22 am

wuttz wrote:
muziqaz wrote:is it true, that crysis uses only 3dnow instructions on my 965BE ? :D
imagine if it used sse3, amd chips would murder intel chips easily.


someone profile and dismantle that benchmark yet? :wink: :wink: :wink:


nah, I don't know, just read someone mentioning it in amd dev forum linked here :)
Proud member of Folding@Home project.
http://folding.stanford.edu
Image
User avatar
muziqaz
K8 Athlon 64 X2 (Manchester) Elite Boarder
K8 Athlon 64 X2 (Manchester) Elite Boarder
 
Posts: 2823
Joined: Sun Sep 02, 2007 3:34 am
Location: London UK

Re: FTC: Federal Trusted Compiler

Postby agner » Tue Dec 22, 2009 8:36 am

Boundless wrote:The FTC is on spIntel's case about this too:
Proposed Remedy #7 in the Intel Complaint.

Thanks for the reference to the FTC complaint. Very interesting. How embarrassing for Intel if they have to make a public statement that their software is "defective" and they have to pay for replacement of all software that was compiled with their compiler and libraries.

I wonder if we have to wait years for the FTC case to be settled before we can get a non-defective compiler from Intel, or they will make it as a consequence of the AMD/Intel settlement. A lot of programmers need the Intel compiler and especially the function libraries.
agner
 
Posts: 22
Joined: Fri Dec 11, 2009 8:47 am

Re: FTC: Federal Trusted Compiler

Postby Boundless » Tue Dec 22, 2009 1:01 pm

agner: > Thanks for the reference to the FTC complaint. Very interesting.

And, as I mentioned in another thread, widely unreported
by ad-supported sites and biased enthusiast sites, for
some reason :).

> How embarrassing for Intel if they have to make a public
> statement that their software is "defective" and they have
> to pay for replacement of all software that was compiled
> with their compiler and libraries.


Based on their only statement so far, I expect them to
stonewall on this, claiming that it is not the case and/or
that they had to code the compilers the way they did.

> I wonder if we have to wait years for the FTC case to be settled
> before we can get a non-defective compiler from Intel, ...


They have already said that the AMD settlement changes nothing.
Expect them to change nothing.
________
Focus, as always, on Plan B,
since Plan A never works.
User avatar
Boundless
K8 Athlon 64 (Orleans) Expert Boarder
K8 Athlon 64 (Orleans) Expert Boarder
 
Posts: 2000
Joined: Mon Mar 01, 2004 4:40 pm

Re: FTC: Federal Trusted Compiler

Postby MKruer » Wed Dec 23, 2009 12:25 am

agner wrote:
Boundless wrote:The FTC is on spIntel's case about this too:
Proposed Remedy #7 in the Intel Complaint.

Thanks for the reference to the FTC complaint. Very interesting. How embarrassing for Intel if they have to make a public statement that their software is "defective" and they have to pay for replacement of all software that was compiled with their compiler and libraries.

I wonder if we have to wait years for the FTC case to be settled before we can get a non-defective compiler from Intel, or they will make it as a consequence of the AMD/Intel settlement. A lot of programmers need the Intel compiler and especially the function libraries.


Ethnically any review site should discard all benchmarks that were compiled using the Intel Compilers. But what will end up happening is that places like Anand and Toms will try to justify keeping the existing benchmarks and then Blaming AMD because Intel excluded then from the compiler purposefully. :roll:

I am sure that someone can and will figure out the real damages. and my best guess is that the real damages will exceed that of the total company. Serious if you think of all the software for all the companies that used to Intel compilers will need to recompile and then install. You are talking the lost of billions of dollars in lost hours to correct this "error"

IMNSHO this is justifiable grounds to bust up Intel into two different companies one for Software the other for Hardware.
Lian-Li PC-V2000 Plus Aluminum Case; Seasonic S12 Energy+ 550 PSU; Asus M4A785TD-V EVO; Phenom II X4 965 Black Edition C3 @ 4.0Ghz ; Thermalright Ultra-120 eXtreme Rev.C; 8GB OCZ AMD Black Edition @ 1333Mhz; Sapphire Radeon HD 7870
User avatar
MKruer
K10 Opteron (Barcelona) Administrator
K10 Opteron (Barcelona) Administrator
 
Posts: 2439
Joined: Mon Mar 01, 2004 4:21 am
Location: I am not paid to do this, I don't even like to do this, I wonder why am I still doing this?

Re: FTC: Federal Trusted Compiler

Postby abinstein » Wed Dec 23, 2009 12:51 am

MKruer wrote:IMNSHO this is justifiable grounds to bust up Intel into two different companies one for Software the other for Hardware.

I also believe that it is perfectly justified from the competitiveness point of view to separate the software part into an independent entity.

However, from a technical point of view, the reverse is happening. Software and hardware co-development is the future. Intel will have a very good "reason" to keep the software part till the end of time. ;)

At the least, though, the regulator should force Intel to open up the ICC to allow 3rd party to plug-in their own optimization backends. Just like Microsoft is forced to open up the Windows API and required to offer support to ISVs.
Anandtech -- a site which every visit makes me regret my time spent! It is a living testimony of Einstein's quote:"Only two things are infinite, the universe and human stupidity, and I'm not sure about the former."
abinstein
K8 Opteron (SledgeHammer) Moderator
K8 Opteron (SledgeHammer) Moderator
 
Posts: 7170
Joined: Sat Oct 30, 2004 9:49 pm

Re: FTC: Federal Trusted Compiler

Postby BaronMatrix » Wed Dec 23, 2009 1:27 am

MKruer wrote:
agner wrote:
Boundless wrote:The FTC is on spIntel's case about this too:
Proposed Remedy #7 in the Intel Complaint.

Thanks for the reference to the FTC complaint. Very interesting. How embarrassing for Intel if they have to make a public statement that their software is "defective" and they have to pay for replacement of all software that was compiled with their compiler and libraries.

I wonder if we have to wait years for the FTC case to be settled before we can get a non-defective compiler from Intel, or they will make it as a consequence of the AMD/Intel settlement. A lot of programmers need the Intel compiler and especially the function libraries.


Ethnically any review site should discard all benchmarks that were compiled using the Intel Compilers. But what will end up happening is that places like Anand and Toms will try to justify keeping the existing benchmarks and then Blaming AMD because Intel excluded then from the compiler purposefully. :roll:

I am sure that someone can and will figure out the real damages. and my best guess is that the real damages will exceed that of the total company. Serious if you think of all the software for all the companies that used to Intel compilers will need to recompile and then install. You are talking the lost of billions of dollars in lost hours to correct this "error"

IMNSHO this is justifiable grounds to bust up Intel into two different companies one for Software the other for Hardware.



The bigger problem is chipsets and CPU. Now they are moving in with SSDs and want to do GPUs. The hardware is the issue as it allows them to basically control the supply chain. Two companies. CPU and all other HW.
MSI 790GX X4 940 - @ 3GHz - AC64 HSF - HD 4870 512MB - 8GB DDR2 - 1TB HDD - BluRay - TX650 W - Antec 300 - 2X ACER 23"
HP DV7-3065 - M600 - 17" LCD - BD-Rom - HDMI\eSATA - .5TB HDD - HD4200
User avatar
BaronMatrix
K8 Athlon 64 (San Diego) Expert Boarder
K8 Athlon 64 (San Diego) Expert Boarder
 
Posts: 1949
Joined: Mon Jan 29, 2007 3:13 am

Re: Intel/AMD settlement and Intel software

Postby mdk777 » Wed Dec 23, 2009 1:56 am

On the technical side, this core is able to take advantage of most modern CPU optimizations (SSE2, SSE3, SSSE3, SSE4.1 and SSE4.2), however, a few compatibility issues are still present on AMD processors, resulting in the core only using SSE2 on these chips. This should change in the not-too-distant future when the issues have been worked out.


It will take time to undo the damage that INTEL has done. This new folding core is just another example. :evil:
Image

Transparency and Accountability, the necessary foundation of any great endeavor!
User avatar
mdk777
Shiny Badge of Honor Member
 
Posts: 1768
Joined: Mon Jul 07, 2008 3:32 pm

Re: FTC: Federal Trusted Compiler

Postby hyc » Wed Dec 23, 2009 3:32 am

abinstein wrote:
MKruer wrote:IMNSHO this is justifiable grounds to bust up Intel into two different companies one for Software the other for Hardware.

I also believe that it is perfectly justified from the competitiveness point of view to separate the software part into an independent entity.

However, from a technical point of view, the reverse is happening. Software and hardware co-development is the future. Intel will have a very good "reason" to keep the software part till the end of time. ;)

At the least, though, the regulator should force Intel to open up the ICC to allow 3rd party to plug-in their own optimization backends. Just like Microsoft is forced to open up the Windows API and required to offer support to ISVs.


Optional pluggable optimizers won't solve anything; most ISVs will not obtain anything optional. The compiler has to work correctly from the get-go for this situation to be solved.

Or, as I said before, do an end-run around Intel by doing binary recompilation. That's really the only guaranteed way to know you've got optimal code. And of course, that's just a half-step back to where things ought to have been all along - full open source, no closed-source binary-only software at all. Which is exactly how things worked in the world before Microsoft came along.
hyc
K8 Athlon 64 (Winchester) Expert Boarder
K8 Athlon 64 (Winchester) Expert Boarder
 
Posts: 1349
Joined: Mon Jan 16, 2006 4:38 pm

Re: Intel/AMD settlement and Intel software

Postby agner » Wed Dec 23, 2009 8:59 am

MKruer wrote:IMNSHO this is justifiable grounds to bust up Intel into two different companies one for Software the other for Hardware.

The Intel compiler is optimizing far better than the Microsoft and Borland compilers according to my tests. And many of Intel's function libraries have no good alternatives. They sell this at a very low price and the support is excellent. If you put this into a separate company it would surely not be profitable :wink:
agner
 
Posts: 22
Joined: Fri Dec 11, 2009 8:47 am

Re: Intel/AMD settlement and Intel software

Postby Эльбрус » Wed Dec 23, 2009 4:24 pm

Hi Agner,

I asked JF-AMD (John Fruehe) already about the compiler topic, he stated, that future ICC revisions should produce better code for AMD chips. The answer is here:

viewtopic.php?f=52&t=137095&p=171746#p171695

Happy holidays
User avatar
Эльбрус
K7 Athlon XP (Palomino) Junior Boarder
K7 Athlon XP (Palomino) Junior Boarder
 
Posts: 463
Joined: Sat May 02, 2009 7:13 pm

Re: Intel/AMD settlement and Intel software

Postby JF-AMD » Fri Dec 25, 2009 2:38 pm

"Should" is the operative word.
While I work for AMD, my posts are my own opinions.

http://blogs.amd.com/work/author/jfruehe/

Follow AMD Opteron on Twitter: @JF_AMD
User avatar
JF-AMD
XIP
 
Posts: 1832
Joined: Thu Apr 23, 2009 7:27 am

Re: Intel/AMD settlement and Intel software

Postby mdk777 » Fri Dec 25, 2009 4:46 pm

or perhaps "eventually"

As we all know the blinding speed of compliance after litigation..... :wink:
Image

Transparency and Accountability, the necessary foundation of any great endeavor!
User avatar
mdk777
Shiny Badge of Honor Member
 
Posts: 1768
Joined: Mon Jul 07, 2008 3:32 pm

Re: Intel/AMD settlement and Intel software

Postby Эльбрус » Fri Dec 25, 2009 6:47 pm

Merry Christmas everybody :)

I already got the presents, all the relatives are gone, too, so I have some free time.

I found the pdf for the Intel 11 compiler. Seems like the Follow-Up to the QxO option. The only difference is a better code-path handling and explicit refering to AMD, example:
Automatic Processor Dispatch. Generates specialized code for the corresponding Intel processors while also generating a default code path. Multiple values, separated by commas, may be used to tune for additional processors in the same executable, e,g. /QaxSSE4.1,SSE3. The default code path may be modified by using in addition a /Qx (-x) or /arch (-m) switch.†
For example, for best performance on the Intel® 45nm Hi-k next generation Intel Core™ microarchitecture and also good performance
on an AMD processor that supports only SSE3, use /QaxSSE4.1
/arch:SSE3 (-axsse4.1 –msse3 on Linux*).
This will produce binaries with two code paths, using automatic processor dispatch technology. One code path will take full advantage of the Intel® 45nm Hi-k next generation Intel Core™ microarchitecture. The other code path will still run well on both Intel and non-Intel processors that support SSE3 but not SSE4. At runtime, the application automatically identifies the Intel processor on which it is running and selects the appropriate code path, either specialized or default.
(...)
† The /arch (-m) option values SSE3, SSE2, and IA32 produce binaries that should run on non-Intel processors that implement the same capabilities as the corresponding Intel processors. The corresponding /Qx (-x) option values perform additional optimizations that are not enabled by /arch (-m), but will run only on Intel processors.


http://cache-www.intel.com/cd/00/00/22/ ... 222300.pdf

I guess AMD has to proof that the intel-only optimizations are not architectur specific, if they want to get the full stuff. The formulation "will run well" is somehow ironic ...

Furthermore I wonder what will happen in the future. There are no compile flags for non-intel SSE4 CPUs, thus Bulldozer will be out of the game in 2011 and score suboptimally in the usual (SSE4 enabled) benchmarks.
Another (earlier) victim will be probably VIA, their Nano 3000 CPU will be launched earlier.

Happy Holidays
User avatar
Эльбрус
K7 Athlon XP (Palomino) Junior Boarder
K7 Athlon XP (Palomino) Junior Boarder
 
Posts: 463
Joined: Sat May 02, 2009 7:13 pm

Re: Intel/AMD settlement and Intel software

Postby Boundless » Fri Dec 25, 2009 7:50 pm

Эльбрус: > Furthermore I wonder what will happen in the future.

There was never any reason for spIntel to make any
allowances for non-spIntel CPUs in their compiler.
I haven't studied the claims they made for it (evidence
of which may now be evaporating from their web), but
my guess is:
  • They wanted ISVs to use their compiler.
  • Therefore the code had to at least run on non-spIntel.
  • But not run too well.
Had they claimed nothing about support for non-spIntel, the
compiler caper wouldn't be mentioned in the settlement or
the FTC complaint.

Had they claimed non-spIntel support, but restrained themselves
from what appears to be active sabotage of non-spIntel codepaths,
we likewise would not be at today's crossroads.

One easily possible outcome is that the spIntel compiler will
fall silent on non-spIntel CPUs, and perhaps even stop making
any allowances for them.
________
If I were writing performance-sensitive apps, I'd use 3rd-party
compilers, and as in days of yore, profile the code, then hand-
code the critical routines in assembly.
User avatar
Boundless
K8 Athlon 64 (Orleans) Expert Boarder
K8 Athlon 64 (Orleans) Expert Boarder
 
Posts: 2000
Joined: Mon Mar 01, 2004 4:40 pm

Next

Return to K10: Barcelona, Shanghai, Quad-Core Opteron, Phenom

Who is online

Users browsing this forum: No registered users and 3 guests