Therefore only x86 processors can include SSE2 edit

This phrase is wrong as SSE2 is for instance supported on later Intel Itanium processors which are definatly not x86 or x86-64 architecture. — Preceding unsigned comment added by 82.66.245.121 (talk) 18:05, 1 March 2015 (UTC)Reply

I do not believe Itanium processors had SSE2 instructions sets. Searching through Intel's product specification website, Itanium-series processors are not shown to have any SSE instruction set extensions. Instead, IA-64 architectures had 2 SIMD registers to allow parallel computations, but they were not SSEx registers. Bansalsi1 (talk) 01:19, 23 April 2024 (UTC)Reply


Untitled edit

It may be useful to maintain a list of software packages that require SSE2 until many of the processors listed that do not support it become less common.

- Adobe Premiere Pro 2.0 - Adobe Encore DVD 2.0


Can we get a reference to a page that actually lists and defines the SSE2 instructions?

sse2 instrucion set for amd athlon processor edit

I bought a computer with amd athlon processor in 2005 Now I want to install Abode Premier 2 .0 Pro How to get sse2 Instruction set

ramamohanrajup@yahoo.com  —Preceding unsigned comment added by 117.195.172.76 (talk) 07:27, 31 August 2008 (UTC)Reply 

You'll need a new processor,as instruction sets are hardware, not software, features. Sorry.Bettering the Wiki (talk) 16:45, 31 August 2008 (UTC)Reply

BTW, you're using a talk page as a forum, which is not permitted, per policy. I have warned you as such.Bettering the Wiki (talk) 16:53, 31 August 2008 (UTC)Reply

Differences between x87 FPU and SSE2: Example code edit

Why is the example coded in Fortran? Seriously. Setting aside the issues of Fortran's datedness as a programming language, or relevance to modern instruction-set extensions, I think it's a valid concern that Fortran is no longer a widely known language among modern programmers. I would suggest C, Java, or any form of BASIC in its place. 75.26.177.53 (talk) 13:45, 13 February 2009 (UTC)Reply

Simplicity of the language (similarity to Basic, Matlab, Octave, Scilab) compared to C/Java/C++? Compiler that can target x87 & SSE2? (compared to Basic)? Expectation that most programmers who care about numerical issues will have been exposed to Fortran? Bendel boy (talk) 08:11, 8 October 2009 (UTC)Reply

What exactly is this example supposed to show? That for SSE -(a/b) == (-a)/b but not for 387? One could get the impression that SSE calculates more exactly, which certainly is not the case. The difference is that it does exactly the same rounding when calculating y and z, so naturally their difference is “exactly” zero. --134.102.204.124 (talk) 14:16, 7 March 2011 (UTC)Reply

number of SSE2 registers edit

All the newer SIMD articles seem to miss the exact number of xmm registers. SSE1 seems to have had 8, but in the literature people often see xmm8..15. It is not clear from these articles which extension upped the register count. (I suspect SSE2 since it established SSE as a standalone execution unit, which is why I post this comment here).

Some quick browsing seems to indicate that SSE2 in x86_64 has 16 regs, and x86 8 ? 88.159.74.100 (talk) 10:22, 4 June 2009 (UTC)Reply

instruction set: a list of the instructions would be helpful edit

what's missing here is a commented exhaustive list of the sse2 instructions or a link to a page with such a list —Preceding unsigned comment added by 217.91.127.94 (talk) 11:46, 2 July 2009 (UTC)Reply

agree I've added X86_instruction_listings#SSE2_instructions to the see also section, but really a table of instructions and short explanation is required. C xong (talk) 01:58, 16 March 2010 (UTC)Reply

"The Intel C++ Compiler can automatically generate SSE4/SSSE3/SSE3/SSE2 and/or SSE-code without the use of hand-coded assembly." edit

It should be noted, that the code generated by the Intel C++ Compiler is only executed, if a "GenuineIntel" CPU is detected at runtime. All CPUs from other vendors will execute slow x87 FPU code, even if they support SSE/SSE2/SSE3/SSE4. --92.224.142.234 (talk) 18:41, 11 June 2013 (UTC)Reply