Kĩ thuật viễn thông - Các bộ vi xử lý trên thực tế

h end consumer Fixed PointSeries continued; typical app.: GSM, set-top box, audio1.8-5V, max. 160 MHz, 144 pin, .15mm (1999), 0.32mW/MIPS for the coreSpecialized on-chip unit: will occur more often in futuree.g. C5420: dual core + 2x100 MW on-chip SRAMe.g. C5402: 5$ for 100 MIPSLoop controllerBuffered seriallinks, timers, ...6 channelDMA controllerFixed ALU32+32->40Fixed Add32+32->40Fixed multiply17x17->34ViterbiPROMDual accessXRAMYRAMaddressdata3217addressdata1616ACUACUI/O83Texas Instruments TMS320C5510High end consumer Fixed PointSeries continued; typical app.: UMTS handheld1.6V, 200 MHz, .15mm (2000), 400 MIPS,0.05mW/MIPS (core), power management per unit and per cycleSpecialized on-chip unit: will occur more often in futurePower MgmentBuffered seriallinks, timers, ...6 channelDMA controllerFixed ALU32+32->40Fixed Add32+32->40Fixed multiply17x17->34ViterbiPROM32 KByteDual accessXRAM (256 Kbyte)YRAM(64 Kbyte)addressdata3224addressdata1616ACUACUI/OFixed multiply17x17->34P-cache24 KByte84Texas Instruments TMS320C8xFixed Point VideoSeries discontinued; typical app.: video phone, video conferencing, multimedia workstationsIntroduced: 1995, 50 MHz, 305 pinMultiprocessor-on-a-chip; sub-word SIMD for each DSPDSP processor 1DSP processor 2DSP processor 3DSP processor 4General purposeRISC processorTransfercontrollerdataaddress3264Video controller2 Kbyte RAM12 Kbyte RAM162 Kbyte I-cache12 Kbyte I-cache44 Kbyte D-cache2 KByte RAM4 KByte I-cacheX-bar85Texas Instruments TMS320C6201High end Fixed PointSeries continued; typical app.: modems, multimedia1997, 0.25 mm, 5ML, 352 pin, 200 MHz, 2.5V, 1.9W, $85Super scalar (8 Instr./cycle), 1600 MIPSVLIW: 256 bit instruction wordfixed MUL16x16->32fixed MUL16x16->32fixed ALU32+32->40fixed ALU32+32->40fixed ALU/branch32+32->40fixed ALU/branch32+32->40integer ACU32+32integer ACU32+3216KByte D-SRAM16KByte D-SRAM16KByte D-SRAM16KByte D-SRAM64KByteP-SRAM/cacheJTAG / clock pump4 channel DMA2 Serial ports2 TimersExt. memoryinterfacedataaddress1716Host interfacedataaddress2332External memory86Texas Instruments TMS320C6202High end Fixed PointSeries continued; typical app.: modems, multimedia1999, 0.18 mm, 5ML, 352 pin, 250 MHz, 1.8V, 1.9W, $130Super scalar (8 Instr./cycle), 2000 MIPS, scales well till 700 MHz (6000 MIPS)Optimum choice when all data fits in on-chip memoryfixed MUL16x16->32fixed MUL16x16->32fixed ALU32+32->40fixed ALU32+32->40fixed ALU/branch32+32->40fixed ALU/branch32+32->40integer ACU32+32integer ACU32+322x16KByte D-RAM(Shadow load)2x16KByte D-RAM(Shadow load)2x16KByte D-RAM(Shadow load)2x16KByte D-RAM(Shadow load)2x128KB P-RAM(Shadow load)JTAG / clock pump4 channel DMA2 Serial ports2 TimersExt. memoryinterfacedataaddress17?32Expansion busdataaddress23?32External memory87Texas Instruments TMS320C6203High end Fixed PointSeries continued; typical app.: base stations2000, 0.15 mm, 5ML, 18 mm2 package size, 300 MHz, 1.5V, 1.5WSuper scalar (8 Instr./cycle), 2400 MIPSOptimum choice when all data fits in on-chip memoryfixed MUL16x16->32fixed MUL16x16->32fixed ALU32+32->40fixed ALU32+32->40fixed ALU/branch32+32->40fixed ALU/branch32+32->40integer ACU32+32integer ACU32+322x64KByte D-RAM(Shadow load)2x64KByte D-RAM(Shadow load)2x64KByte D-RAM(Shadow load)2x64KByte D-RAM(Shadow load)256KByte P-RAM128KB P-cache/RAMJTAG / clock pump4 channel DMA2 Serial ports2 TimersExt. memoryinterfacedataaddress17?32Expansion busdataaddress23?32External memory88Texas Instruments TMS320C6211High end Fixed PointSeries continued; typical app.: modems, multimedia1999, 0.18 mm, 5ML, 256 pin, 150 MHz, 1.8V, 1.5W, $25VLIW, 1.2 GIPS; cheap (25$ in ‘99, 5$ in ‘01)Optimum for random access to large memory space80% of performance of C6x with infinite on-chip memoryfixed MUL16x16->32fixed MUL16x16->32fixed ALU32+32->40fixed ALU32+32->40fixed ALU/branch32+32->40fixed ALU/branch32+32->40integer ACU32+32integer ACU32+324KByte L1 Dcache(2 way set assoc.)4KByte L1 Pcache(2 way set assoc.)4x16KByte L2cache (direct map)JTAG / clock pump16 channel DMA2 Serial ports2 TimersExt. memoryinterfacedataaddress1716Host portdataaddress3032External memory89Texas Instruments TMS320C6416 High end Fixed PointSamples June 2001, 0.12 mm, 6 LM, 532 pin, 400 MHz-600 MHz, 1.2V, starts at 95$ in volumeSuper scalar (8 Instr./cycle), 3200-4800 MIPSSub-word (8bit or 16bit) parallelismSpecialized instr.: Galois Field Mult, bit manipulationfixed MUL16x16->32fixed MUL16x16->32fixed ALU32+32->40fixed ALU32+32->40fixed ALU/branch32+32->40fixed ALU/branch32+32->40integer ACU32+32integer ACU32+32JTAG / clock pump64 channel DMA3 Serial ports3 Timers16 Kbyte L1Pdirect mapped16 Kbyte L1D2way dual access1 Mbyte RAM/L24wayDual EMIF & HPI &PCI & Utopiadataaddress?32HPIdataaddress3064External memorydataaddress3016Viterbi decoderacceleratorTurbo decoderaccelerator90Texas Instruments TMS320C6701High end Floating PointSeries continued; typical app.: video compressionIntroduced: 1998, 0.18 mm, 5ML, 352 pin, 167 MHz, 1.8VSuper scalar (8 Instr./cycle); VLIW; 1 GFLOPForeseen for ‘00: 50$ (cf. C6211) & 3 GFLOP (cf. C6202)Fixed/Float MUL32x32/64x64Fixed/Float MUL32x32/64x64Fixed/Float ALU32+32/64+64Fixed/Float ALU32+32/64+64Fixed ALU/BranchFloat 1/x & xFixed ALU/BranchFloat 1/x & xinteger ACU32+32integer ACU32+3216KByte D-SRAM16KByte D-SRAM16KByte D-SRAM16KByte D-SRAM64KByteP-SRAM/cacheJTAG / clock pump4 channel DMASerial interface2 TimersExt. memoryinterfacedataaddress1716Host interfacedataaddress2332External memory91Texas Instruments TMS320C6711High end Floating PointSeries continued; typical app.: video compression2000, 0.18 mm, 5ML, 256 pin, 100 MHz, 1.8V, 2W, $20VLIW, 600 MFlopsOptimum for random access to large memory space80% of performance of C6x with infinite on-chip memoryFixed/Float MUL32x32/64x64Fixed/Float MUL32x32/64x64Fixed/Float ALU32+32/64+64Fixed/Float ALU32+32/64+64Fixed ALU/BranchFloat 1/x & xFixed ALU/BranchFloat 1/x & xinteger ACU32+32integer ACU32+32JTAG / clock pump4 channel DMASerial interface2 TimersExt. memoryinterfacedataaddress1716Host interfacedataaddress2332External memory4KByte L1 Dcache(2 way set assoc.)4KByte L1 Pcache(2 way set assoc.)4x16KByte L2cache (direct map)92Texas InstrumentsTMS320C541 (1995)93Texas InstrumentsTMS320C545 (1995)94Texas InstrumentsTMS320C80 (1994)95Chương 7: Các bộ vi xử lý trên thực tếGeneral purpose microprocessorsIntel 80x86Xu hướng phát triểnMicrocontrollersVi điều khiển của MotorolaHọ vi điều khiển 8051Họ vi điều khiển AVRPSOCXu hướng phát triểnDigital signal processorsTexas InstrumentsMotorolaPhilipsXu hướng phát triển96Motorola MC56xxxAudio Fixed Point24 bit for audio: 16 bit data + overflow16 or 24 bitinteger CPULoop controllerSelection ofperipherals:ADC, DAC, comm.,timers, PIO, ...ACUPRAMXRAMYRAMaddressdata2418ACU97Motorola MC5600298Motorola MC5616699Chương 7: Các bộ vi xử lý trên thực tếGeneral purpose microprocessorsIntel 80x86Xu hướng phát triểnMicrocontrollersVi điều khiển của MotorolaHọ vi điều khiển 8051Họ vi điều khiển AVRPSOCXu hướng phát triểnDigital signal processorsTexas InstrumentsMotorolaPhilipsXu hướng phát triển100Philips VSP-1Fixed Point Video12 bit for video: 8 bit data + overflowClock Frequency: 27 MHz1 instruction per sample period for HDTV,2 instructions per sample period for TV12 bitinteger ALU12 bitinteger ALU512x12 bitMemory element512x12 bitMemory element12 bitinteger ALU10x18 cross-bar121210101Philips VSP-1Fixed Point VideoALUALUALUMEMEOutputsInputs102Philips VSP-1Fixed Point VideoALUMemoryElementOutputFIFOs206K Transistors1.1W dissipation27 MHz clock176 pinIntroduced in 1991103Philips VSP-2Fixed Point Video12 bit for video: 8 bit data + overflowClock Frequency: 54 MHz2 instructions per sample period for HDTV,4 instructions per sample period for TV22x50 cross-bar22121212 bitinteger ALU112 bitinteger ALU2512x12 bitMemory element1512x12 bitMemory element212 bitinteger ALU12512x12 bitMemory element4104Philips VSP-2Fixed Point Video1.15 M Transistors5W dissipation54 MHz clock frequency208 pinIntroduced in 1994105Sony Graphics EnginePlaystation 3Status: prototype in 2001287.5 MTOR256 Mbit on-chip embedded DRAM2000-bit wide internal bus462 mm2180 nm CMOS106Chương 7: Các bộ vi xử lý trên thực tếGeneral purpose microprocessorsIntel 80x86Xu hướng phát triểnMicrocontrollersVi điều khiển của MotorolaHọ vi điều khiển 8051Họ vi điều khiển AVRPSOCXu hướng phát triểnDigital signal processorsTexas InstrumentsMotorolaPhilipsXu hướng phát triển107Trends for DSP processorsNo new generations that replace old generations, but multiple co-existing architecture linesWord length application dependent Automotive: 16-bit fixed point (e.g. C2x)Speech: 32-bit floating point (e.g. C30)Audio: 24-bit fixed point (e.g. MC56K)Telecommunications: 16-32 bit fixed point (e.g. C5x, C6x)Video: 12-32 bit fixed point (e.g. C8x)Single architecture line is whole familydifferent memory & on-chip peripheralsfor embedded applications (cf. microcontrollers)108Trends for DSP processorsDeterministic behaviorno caches, no virtual memory, but on-chip RAM banksno out-of-order executiondelayed branch predictionIncreasing address space: 12 -> 32Multiple functions on single chip: CPU, FPU, multiple RAM banks, ACUs, loop controller, ADC, DAC, PWM, serial interfaces, Often provisions for parallel processing109