Enhanced 3DNow!Technology AthlonProcessor Enabling Superior Visua
|
|
|
Top Searches for this datasheet
Enhanced 3DNow!Technology AthlonProcessor
Enabling Superior Visual Computing Experience Next-Generation Computing Platforms
ADVANCED MICRO DEVICES, INC. Place Sunnyvale, 94088
Page
Enhanced 3DNow! Technology
August 2000
Introduction: Taking 3DNow!Technology Next Level
3DNow!technology proven significant addition architecture that drives today's computing platforms. Introduced feature AMD-K6®-2 processor 1998, 3DNow! technology growing worldwide installed base more than million mid-1999). With introduction 3DNow! technology, became first company deliver enhancement instruction platforms, well first introduce superscalar SIMD floating point technology processors. Intel delivered SIMD floating point solution (SSE) nine months after advent 3DNow! technology. SIMD, which stands Single Instruction Multiple Data, enables single instruction operate parallel multiple pieces data. original 3DNow! technology consists instructions that enhance personal computing experience using SIMD floating point techniques open processing bottlenecks floating point-intensive multimedia applications. 3DNow! instructions designed improve performance single-precision floating point operations, enabling four these operations execute single clock cycle. 3DNow! technology-optimized hardware software applications used create more productive, visually realistic, entertaining platforms. 3DNow! technology enables faster frame rates high-resolution video scenes, better physical modeling real-world environments, sharper more detailed imaging, smoother video playback, theater-quality audio. result more satisfying visual computing experience. With introduction Athlonprocessor, taken 3DNow! technology next level performance functionality. Athlon processor features enhanced version 3DNow! technology that adds instructions existing original 3DNow! instructions. These additional instructions include:
instructions that improve multimedia-enhanced integer math calculations used such applications speech recognition video processing instructions that accelerate data movement more detailed graphics functionality Internet browser plug-ins other streaming applications, enabling richer Internet experience instructions that enhance performance communications applications, including soft modems, soft ADSL, MP3, Dolby Digital surround sound processing. This functionality unique Athlon processor supported Intel's Pentium® processor.
Page
Enhanced 3DNow! Technology
August 2000
enhancing 3DNow! technology, kept instruction design simple, powerful. AMD's goal designing these additional 3DNow! instructions provide powerful SIMD performance least comparable while enabling ease implementation software developers. relatively instructions enhanced 3DNow! technology total instructions versus SSE's 71), well full support various compilers debuggers, allow developers adopt this technology optimize their applications quickly. addition additional 3DNow! instructions, Athlon processor implements microarchitecture enhancements that enable more efficient operation these instructions, resulting fewer coding restrictions greater programming efficiencies. Table below compares instructions implemented 3DNow! technology SSE. 3DNow! technology superior solution because first market continues have tremendous momentum with large installed base. also offers latest SIMD extensions, well functionality available SSE, enables enhanced 3D/multimedia performance that planned provide tangible end-user benefits years come. Table Enhanced 3DNow!Technology Instruction Comparison
Functionality
SIMD floating point functionality first)
3DNow!(original 3DNow! instructions)
Conclusion
Comparable functionality: Both technologies support operations clock. 3DNow! technology simpler implement. many more instructions because Intel's architecture requires MMXcontrol functionality duplicated requires instructions extension: SIMD another scalar operations. Comparable functionality: Both have instructions cache streaming controls. Unique functionality: advances SIMD with extensions soft modems, soft ADSL, complex math, MP3, Dolby Digital. Advantage AMD: Enhanced 3DNow! technology greater functionality than SSE. When combined with superior engine Athlon processor, result best FP/multimedia performance processors.
MMX(integer) augmentation data movement DSP/communication extensions Total number instructions
(additional instructions) (additional instructions)
Page
Enhanced 3DNow! Technology
August 2000
3DNow!Instructions Added AthlonProcessor
described introduction, multimedia instructions added enhanced 3DNow! technology implemented Athlon processor fall into three categories:
extensions augment MMXinstruction more accurate speech recognition software high-quality video encoding/decoding applications (see Table Appendix data movement/streaming instructions enhanced performance many software applications ability move large amounts multimedia data into processor, enabling improved data movement Internet plug-ins other streaming applications (see Table Appendix DSP/communications extensions faster processing communications-based applications, such soft modems, cable modems, ADSL modems, well advanced audio applications, such Dolby Digital surround sound (see Table Appendix
Superior Implementation Enhancement
Athlon processor's seventh-generation microarchitecture well suited multimedia extensions implemented processor, well original 3DNow! instruction set. developing original 3DNow! technology, kept instruction design sparse, powerful. AMD's goal provide powerful SIMD performance while enabling ease implementation software developers. relatively instructions 3DNow! technology allow developers adopt this technology optimize their applications quickly easily. Likewise, multimedia enhancement, data movement, extensions implemented Athlon processor designed handle computation requirements complex multimedia applications, total number instructions still manageable enough allow easy, efficient programming.
More Efficient Multimedia Execution with 2-Way SIMD
SIMD technique implemented enhanced 3DNow! technology involves instruction acting upon multiple data values. 3DNow! instructions operate 32-bit values time. This 2-way SIMD approach matches 64-bit execution unit size Athlon processor, which enables very efficient multimedia execution resources.
Page
Enhanced 3DNow! Technology
August 2000
Athlon processor generate four 32-bit floating point results (adds, subtracts, multiplies) clock cycle. Each instruction only operates singleprecision floating point operands, Athlon processor microarchitecture also execute 3DNow! instructions clock cycle through execution pipelines. flexible microarchitecture allow future Athlon processor implementations augmented with additional pipelines. Athlon processor also separate Multiply hardware, which enables perform simultaneous floating point multiply calculations. multimedia units Athlon processor fully pipelined execute most additional multimedia enhancement 3DNow! technology extensions single cycle. Other implementations (such Pentium III) whose 4-way SIMD approach 64-bit execution unit size match, require 2-pass technique execute some their multimedia instructions. Athlon processor's 2-way SIMD approach saves programmers from having painstakingly restructure their data arrays match 4-way SIMD architectures used Pentium III). Exploitation 4-way SIMD requires total rearrangement data structures, which have wide-ranging impact entire software project impact non-program-related content. Software developers continue their existing data structures, which more effectively exploited with 2-way SIMD technology than with 4-way SIMD, enabling them focus more their time providing richer content. When processing vectors with lengths four more values, Athlon processor fully exploit superior execution data movement resources, enabling outperform previous-generation processors, such Pentium III. addition, Direct3D application programming interface (API) Microsoft's DirectX contains data structures that provide better match 2-way SIMD approach enhanced 3DNow! technology. short, 2-way SIMD involves much less shuffling data than 4-way SIMD, with performance degradation.
AthlonProcessor Architecture Benefits Data Movement
Many multimedia algorithms require movement large data sets into processor manipulation then processor video card display. realize benefits added data movement instructions, processor must offer necessary microarchitecture features bandwidth. Athlon processor's deep internal buffering pipelines enable very high operating frequencies.
Page
Enhanced 3DNow! Technology
August 2000
Athlon processor's extremely large 128KB cache holds much more data than smaller (32KB) cache Pentium III, enabling Athlon processor access data more rapidly than competing processors. addition, Athlon processor brings data 64-byte cache blocks-twice size Pentium III. Athlon processor also features on-die 256KB cache that enables cache same frequency processor.1 cache organized 16-way associative design, allowing improved ratio sophisticated data structures. addition this extremely high-bandwidth interface, Athlon processor uses "exclusive" cache design which improves efficiency excluding data from cache. Most caches utilize "inclusive" design, forcing cache include store redundant data cache already storage. This redundancy storage reduces total amount information cache store other important data. Athlon processor total 128KB+256KB=384KB full-speed on-die cache-a total system cache more than 95KB larger than that Pentium III. enhanced 3DNow! technology, 200MHz system bus, full-speed on-die cache, superscalar floating point engine, other seventh-generation architectural features Athlon processor contribute processor's ability move data faster more efficiently than previous-generation implementations, such Pentium III.
Zero Switching Overhead
Athlon processor imposes switching overhead between executing (floating point) 3DNow! instructions. This allows developers port 3DNow! technology without fear performance degradation when they switch back required code. Because there switching overhead between instructions, 3DNow! instructions easily added existing applications.
Operating System Support Needed
Enhanced 3DNow! technology allows 3DNow! registers shared. Therefore, enhanced 3DNow! instruction does require operating system support, which enables compatibility with Microsoft® Windows® many other popular operating systems, such Linux Solaris. Adding registers contributes
While Athlon processors currently production feature 256K full-speed, on-die cache memory, previous versions without ondie cache still commercially available.
Page
Enhanced 3DNow! Technology
August 2000
complexity maintaining another processor state subsequent context switching overhead. Sharing 3DNow! registers also provides benefit being able instructions load/store, shuffle, compare 3DNow! instruction data. example, programmer instructions quickly extract exponents mantissas functions directly from 3DNow! instruction data.
Powerful Instruction Decoders
Athlon processor high instruction decode bandwidth enable efficient operation 3DNow! instructions. Athlon processor contains three powerful instruction decoders, which instruction cache four times larger than Pentium instruction cache. Each decoder decode type instruction, enabling decode three 3DNow! instructions cycle. Pentium processor only decode 4-way type instruction cycle because only complex decoder. combination sophisticated instruction decoders dual load/store pipes Athlon processor enables 3DNow! instructions that access memory execute without affecting peak throughput other 3DNow! instructions. This much more efficient more scalable alternative adding registers register space.
Instructions Unique AthlonProcessor
majority digital signal processing (DSP) instructions represented four algorithms: complex arithmetic multiplies, filter evaluation, FFT/DFT, ReedSolomon error correction code. five extensions added 3DNow! technology greatly accelerate execution these algorithms, enabling low-cost implementations communications products, such ADSL, cable, V.90 modems, well advanced audio applications, such Dolby Digital surround sound. Athlon processor currently only processor that offers instructions specifically designed enhance DSP/communications applications.
More Efficient Approach Prefetching
addition supporting various flavors streaming PREFETCH instructions, Athlon processor implements 3DNow! technology's PREFETCHW instruction intent-to-modify purposes. When programmer wants modify
Page
Enhanced 3DNow! Technology
August 2000
prefetched data, other existing processors would have prefetch instruction then issue write operation. This write operation changes cache state "Modified," which causes additional cycles consumed. using PREFETCHW instruction Athlon processor, data brought simultaneously marked "Modified." This efficient approach prefetching ultimately enhances multimedia performance saving cycles over less-efficient technique handling cache state change after issuing write. These saved cycles mean program faster, which ultimately helps enhance user's computing experience.
Summary: Making Powerful Technology Even Better
enhancing 3DNow! technology Athlon processor architecture, made powerful technology even better. additional features instructions added Athlon processor will allow even greater numbers software programs (and therefore users) benefit from enhanced performance enabled added instructions. addition, more types programs these instructions, therefore emerging technologies such voice-recognition, digital video, enhanced Internet audio video streaming, digital communication will become more commonly available broader audiences. 3DNow! technology's high performance operating system friendliness enabled garner widespread software support while providing tangible benefits many existing customers processors with 3DNow! technology. Support 3DNow! technology exists today leading industry-standard APIs, including Microsoft's DirectX SGI's OpenGL. addition, numerous applications from leading independent software vendors (ISVs) worldwide have been process being optimized 3DNow! technology. proliferation Athlon processor-based platforms planned enhance growth vitality 3DNow! technology installed base opening door wider range markets, including small business commercial enterprise segments. processors with 3DNow! technology planned span complete range computing platforms, from sub-$1,000 laptops based AMD-K6-2 processor high-end desktop systems, workstations, servers powered Athlon processor. future, Athlon processor with enhanced 3DNow! technology planned deliver superior visual computing experience next generation computing platforms.
Page
Enhanced 3DNow! Technology
August 2000
Overview
(NYSE: AMD) global supplier integrated circuits personal networked computer communications markets. produces processors, flash memories, products communications networking applications. world's second-leading supplier Windows compatible processors, shipped more than million microprocessors, including more than million Windows compatible CPUs. Founded 1969 based Sunnyvale, California, sales marketing offices worldwide manufacturing facilities Sunnyvale; Austin, Texas; Dresden, Germany; Bangkok, Thailand; Penang, Malaysia; Singapore; Aizu-Wakamatsu, Japan. revenues $2.9 billion 1999.
Cautionary Statement
This White Paper includes forward-looking statements that made pursuant safe harbor provisions Private Securities Litigation Reform 1995. Forward-looking statements generally preceded words such "expects," "plans," "believes," "anticipates," "intends." Investors cautioned that forward-looking statements this white paper involve risks uncertainties that could cause actual results differ from current expectations. Forward-looking statements this white paper about Athlon processor involve risk that will able produce processor volume required customers timely basis; that successful developing infrastructure support processor; that third parties provide infrastructure solutions support processor; that processor will achieve customer market acceptance; that software applications will optimized with processor. urge investors review detail risk uncertainties company's Securities Exchange Commission filings, including most recently filed Form-10K.
AMD, logo, Athlon combinations thereof, 3DNow! trademarks, AMD-K6 registered trademark Advanced Micro Devices, Inc. Microsoft Windows registered trademarks Microsoft Corporation. trademark, Pentium registered trademark Intel Corporation. Other product names used this publication identification purposes only trademarks their respective companies.
Page
Enhanced 3DNow! Technology
August 2000
Appendix
Instructions Enhanced 3DNow!Technology
Table Multimedia Extensions
Instruction
PAVGB PAVGW PEXTRW PINSRW PMAXSW PMAXUB PMINSW PMINUB PMOVMSKB PMULHUW PSADBW PSHUFW
Function
Packed average unsigned byte Packed average unsigned word Extract word into integer register Insert word from integer register Packed maximum signed word Packed maximum unsigned byte Packed minimum signed word Packed minimum unsigned byte Move mask integer register Packed multiply high unsigned word Packed absolute byte differences Packed shuffle word
Table Data Movement/Streaming Instructions
Instruction
MASKMOVQ MOVNTQ PREFETCHNTA PREFETCHT0 PREFETCHT1 PREFETCHT2 SFENCE Streaming store Prefetch non-temporal access Prefetch cache levels Prefetch cache levels except level Prefetch cache levels except level Store fence
Function
Streaming store using byte mask
Table DSP/Communications Extensions
Instruction
PF2IW PFNACC PFPNACC PI2FW PSWAPD
Function
Packed floating point integer word conversion with sign extend Packed floating point negative accumulate Packed floating point mixed positive-negative accumulate Packed integer word floating point conversion Packed swap double word
Page
Enhanced 3DNow! Technology
August 2000
Appendix Original Instructions 3DNow!Technology
Table Floating Point Instructions
Instruction
PAVGUSB PFADD PFSUB PFSUBR PFACC PFCMPGE PFCMPGT PFCMPEQ PFMIN PFMAX PI2FD PF2ID PFRCP PFRSQRT PFMUL PFRCPIT1 PFRSQIT1 PFRCPIT2 PMULHRW
Function
Packed 8-bit unsigned integer averaging Packed floating point addition Packed floating point subtraction Packed floating point reverse subtraction Packed floating point accumulate Packed floating point comparison, greater equal Packed floating point comparison, greater Packed floating point comparison, equal Packed floating point minimum Packed floating point maximum Packed 32-bit integer floating point conversion Packed floating point 32-bit integer Packed floating point reciprocal approximation Packed floating point reciprocal square root approximation Packed floating point multiplication Packed floating point reciprocal first iteration step Packed floating point reciprocal square root first iteration step Packed floating point reciprocal/reciprocal square root second iteration step Packed 16-bit integer multiply with rounding
Table Performance-Enhancement Instructions
Instruction
FEMMS PREFETCH
Function
Faster entry/exit MMXtechnology floating point state Prefetch least 32-byte line into data cache (Dcache)
Page
Enhanced 3DNow! Technology
August 2000
Other recent searches
UT54LVDS031LV - UT54LVDS031LV UT54LVDS031LV Datasheet
RB751CS-40 - RB751CS-40 RB751CS-40 Datasheet
NSCG455AT - NSCG455AT NSCG455AT Datasheet
BAS40T - BAS40T BAS40T Datasheet
BAS40-04T - BAS40-04T BAS40-04T Datasheet
2SD2118 - 2SD2118 2SD2118 Datasheet
Privacy Policy | Disclaimer
© 2012 Datasheet Archive
