AMD Confidential

User Manual

November 21st, 2008

A.6.7 Extension to the 3DNow! Instruction Set

This section describes the five new DSP instructions added to the 3DNow! Instruction set.

 

 

 

 

 

 

Instruction

 

 

 

Supported

 

 

Mnemonic

 

 

Opcode

 

 

 

Description

 

 

 

 

 

 

 

 

 

 

 

 

 

PF2IW mmreg1,mmreg2/m64

 

 

0F

0F

/1C

 

 

Packed

floating-point to

integer

 

 

 

 

 

word conversion with sign extend.

 

 

 

 

 

 

 

 

 

 

 

 

 

PFNACC mmreg1,mmreg2/m64

 

 

0F

0F

/8A

 

 

Packed

floating-point

negative

 

 

 

 

 

accumulate.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PFPNACC mmreg1,mmreg2/m64

 

 

0F

0F

/8E

 

 

Packed

floating-point

mixed

 

 

 

 

 

positive-negative accumulate.

 

 

 

 

 

 

 

 

 

 

 

 

 

PI2FW mmreg1,mmreg2/m64

 

 

0F

0F

/0C

 

 

Packed 16-bit integer to floating-

 

 

 

 

 

point conversion.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PSWAPD mmreg1,mmreg2/m64

 

 

0F

0F

/BB

Packed swap double word.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Table 15-11: Extension to 3DNow! Instruction Reference

A.6.8 Prescott New Instructions

Prescott New Instruction technology for the x64 architecture is a set of 13 new instructions that accelerate performance of Streaming SIMD Extension technology, Streaming SIMD Extension 2 technology, and x87-FP math capabilities. The new technology is compatible with existing software and should run correctly, without modification. The thirteen new instructions are summarized in the following section. For detailed information on each instruction refer to a complete Instruction Set Reference.

 

 

 

 

 

 

 

Instruction

 

 

 

 

 

 

 

Supported

 

 

Mnemonic

 

 

 

Opcode

 

 

 

 

Description

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Add/Subtract packed double-precision

 

ADDSUBPD xmm1,xmm2/m128

 

66

0F

D0

/r

 

 

floating-point

number

from

XMM2/Mem

 

 

 

 

 

 

 

 

 

 

 

to XMM1.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Add/Subtract packed single-precision

 

ADDSUBPS xmm1,xmm2/m128

 

 

F2

0F

D0

/r

 

 

floating-point

number

from

XMM2/Mem

 

 

 

 

 

 

 

 

 

 

 

to XMM1.

 

 

 

 

 

 

 

 

FISTTP m16int

 

 

DF /1

 

 

 

 

Store

ST

as

a

signed

integer

 

 

 

 

 

 

 

(truncate) in m16int and pop ST.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FISTTP m32int

 

 

DB /1

 

 

 

 

Store

ST

as

a

signed

integer

 

 

 

 

 

 

 

(truncate) in m32int and pop ST.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

FISTTP m64int

 

 

DD /1

 

 

 

 

Store

ST

as

a

signed

integer

 

 

 

 

 

 

 

(truncate) in m16int and pop ST.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Add

horizontally

packed

double-

 

HADDPD xmm1,xmm2/m128

 

66

0F

7C

/r

 

 

precision

floating-point

numbers

 

 

 

 

 

 

 

 

 

 

 

from XMM2/Mem to XMM1.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Add

horizontally

packed

single-

 

HADDPS xmm1,xmm2/m128

 

 

F2

0F

7C

/r

 

 

precision

floating-point

numbers

 

 

 

 

 

 

 

 

 

 

 

from XMM2/Mem to XMM1.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Subtract horizontally packed double-

 

HSUBPD xmm1,xmm2/m128

 

66

0F

7D

/r

 

 

precision

floating-point

numbers

 

 

 

 

 

 

 

 

 

 

 

from XMM2/Mem to XMM1.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Subtract horizontally packed single-

 

HSUBPS xmm1,xmm2/m128

 

 

F2

0F

7D

/r

 

 

precision

floating-point

numbers

 

 

 

 

 

 

 

 

 

 

 

from XMM2/Mem to XMM1.

 

 

 

 

 

LDDQU xmm,m128

 

 

F2

0F

F0

/r

 

 

Load

128

bits

from

 

Memory

to XMM

 

 

 

 

 

register.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Appendix A

225

Page 237
Image 237
AMD 4.4.5 user manual Extension to the 3DNow! Instruction Set, Prescott New Instructions, 225

4.4.5 specifications

AMD 4.4.5 is a robust version of the AMD software ecosystem that focuses on enhanced performance, stability, and efficiency for users relying on AMD processors and graphics cards. This iteration brings various features and technologies designed to optimize gaming, professional applications, and general computing tasks.

One of the standout enhancements in AMD 4.4.5 is the integration of Smart Access Memory technology, which allows the CPU to access the full graphics memory directly. This feature enhances data flow between the CPU and GPU, leading to improved frame rates and overall performance, especially in memory-intensive games. Users can experience a marked increase in gaming performance without the need for additional hardware upgrades.

Another significant feature introduced is the updated Radeon Software Adrenalin Edition. This software update encompasses optimizations for various games, ensuring that gamers can enjoy a seamless experience with the latest titles. The Adrenalin interface also provides features such as Radeon Chill, which helps reduce power consumption during less demanding scenes, and Radeon Anti-Lag, designed to minimize input lag and enhance responsiveness in competitive gaming.

Additionally, AMD 4.4.5 brings the latest drivers that include important performance enhancements and bug fixes that allow for greater system stability. Regular updates ensure that users have the best experience with their AMD hardware, and it significantly bolsters compatibility with the latest gaming titles and software applications.

AMD's Enhanced Sync technology offers users the ability to eliminate screen tearing and stuttering while delivering smooth gameplay. It dynamically synchronizes the refresh rate of the display with the frame rates produced by the GPU, improving the visual experience significantly.

With the introduction of FidelityFX Super Resolution, AMD continues to bolster its suite of technologies that enhance graphics quality and performance. This feature allows lower-resolution images to be upscaled effectively, making it easier for users to enjoy high-end visuals without taxing their hardware excessively.

In summary, AMD 4.4.5 represents a comprehensive update for users by integrating cutting-edge technologies that cater to a wide range of applications from gaming to content creation. With enhancements across the board, AMD reaffirms its commitment to providing high-performance solutions that keep pace with the ever-evolving technology landscape. Whether for gaming aficionados or professional creators, the capabilities offered in this version make it a significant milestone in the AMD software ecosystem.