SSE instruction set
The SSE (streaming SIMD extensions, single-instruction multi-data stream extension) Instruction Set was first launched by intel in the Pentium III processor. In fact, before piII was officially launched, Intel once published the so-called kni (Katmai new instruction) instruction set through various channels. This instruction set is also the predecessor of SSE instruction set, it was once called the next version of MMX instruction set by many media, namely the mmx2 instruction set. It turns out that the "kni" Instruction Set was the first Instruction Set Name that Intel named for its next-generation chip, the so-called "mmx2" is completely the evaluation of "kni" by the hardware critics and the media. Intel has never officially published any message about mmx2.
The final SSE instruction set is the so-called "Internet SSE" instruction set. The SSE instruction set includes 70 commands, including 50 simds that increase the efficiency of 3D graphics operations (single command multi-data technology) floating Point operation commands, 12 MMX Integer Operation enhancement commands, and 8 optimized continuous data block transmission commands in the memory. Theoretically, these commands play an all-round enhancement role in popular multimedia applications such as image processing, floating point operations, 3D operations, video processing, and audio processing. S se command and 3 dnow! Commands are not compatible with each other, but SSE contains 3 dnow! Most of the functions of the technology are implemented in different ways. SSE is compatible with MMX commands. It can process multiple floating point data in parallel through SIMD and a single clock period to effectively improve the floating point operation speed.
Later, Intel began to respond to AMD's 3 dnow! + Instruction set, sse2 was developed based on SSE, and some instructions were added to greatly improve the performance of its P4 processor. By the end of the P4 design, Intel has added a set of sse2 instruction sets including 144 new commands. Like the earliest SIMD extended instruction set, sse2 immediately executes a single command (namely, SIMD, the best way to calculate low-level industrial control is to execute more work per command ). The most important thing is that sse2 can handle 128-bit and double-precision floating point mathematical operations. The ability to process more precise floating point numbers makes sse2 a basic configuration for accelerating multimedia programs, 3D processing projects, and workstation tasks. But it is important that the software can be properly optimized and utilized.
3D now! Instruction Set
3 dnow proposed by AMD! Instruction Set should appear before SSE instruction set and be widely used by AMD on its K6-2, K6-3 and athlon (K7) processor. 3 dnow! Instruction Set technology is actually an Extended Instruction Set of 21 machine codes.
Unlike Intel's MMX technology, which focuses on integer calculation, 3 dnow! The Instruction Set is mainly used for 3D modeling, coordinate transformation, effect rendering, and other 3D applications. With the help of the software, the 3D processing performance can be greatly improved. Later, enhanced 3 dnow! was developed on athlon !. These amd standard SIMD commands have the same performance as Intel's SSE. Due to the impact of Intel's business and Pentium III success, the software supported SSE compared to 3 dnow! More common. Enhanced 3 dnow! AMD continues to add up to 52 instructions, including some SSE codes, so it can achieve better performance in software optimized for SSE.
The SSE instruction set includes 70 commands, including 50 simds that increase the efficiency of 3D graphics operations (single command multi-data technology) floating Point operation commands, 12 MMX Integer Operation enhancement commands, and 8 optimized continuous data block transmission commands in the memory. Theoretically, these commands play a comprehensive enhancement role in popular multimedia applications such as image processing, floating point operations, 3D operations, video processing, and audio processing.
In particular, the SSE command also contains eight commands to optimize the continuous data block transmission in the memory.