NOTE: The information provided below is for a Cyrix product that is no longer in production.

The 5x86™ processor utilizes efficient fifth-generation architectural features to significantly improve performance while minimizing transistor count. It achieves this performance using a superpipelined architecture in the integer unit combined with data forwarding, branch prediction, a 16-KByte unified write-back cache, single-cycle instruction decode, and single-cycle execution. The processor's built-in power-saving features automatically power down the Floating Point Unit (FPU) and other idle internal circuits, while the System Management Mode (SMM) conserves power flowing to system peripherals.

• Features and Benefits
• Architectural Overview
• Technical Specifications
• Performance Benchmarks

Features and Benefits

The Cyrix 5x86™ processor, formerly called the M1sc, is first in a family of processors for desktop and mobile systems. With its fifth-generation architectural core, the 5x86 processor gives users an affordable performance alternative to the Pentium® processor with a market-ready socket solution.

Fifth-Generation Architecture
The new 5x86 processor family rivals the performance of Pentium® processors to achieve compelling system performance while consuming only half the power of competing alternatives. Fifth-generation architectural features were carefully evaluated and selected for their contribution toward maximum efficiency, performance, and simplicity of design. Features such as a 64-bit internal architecture, branch prediction, data forwarding, and multiple operations issued per clock (made possible by a decoupled load/store unit) are combined with an 80-bit floating point unit (FPU) and 16K unified write-back cache. Aggressive power management features conserve power within the processor as well as power flowing to system peripherals.

Minimal Power Consumption
The Cyrix 5x86 architecture was engineered with power-saving intelligence to track, monitor, and automatically power down the floating point unit and other internal circuits when not in use. It features Cyrix's proven system management mode (SMM) to control power flowing to system peripherals. At 100 MHz @ 3.3 volts, the 5x86 processor consumes less than 3.5 watts of power, which minimizes heat dissipation and makes the 5x86 processor the ideal choice for power-sensitive mobile systems.

Package
The Cyrix 5x86 processor is an example of Cyrix's strategy to design next-generation processor architectures that leverage existing designs. It is initially available in a 168-pin PGA or a 208-pin QFP package. This socket solution offers easy design-in with minimal board space requirements for maximum integration flexibility.

In designing the Cyrix 5x86 processor, Cyrix engineers analyzed the performance features of the M1 processor. The goal was to identify those features that could increase the performance of a single-execution pipeline with minimum added transistor count and power consumption.

Two facts were fundamental in identifying features for the 5x86: the 32-bit architectural standard of x86 technology, and the average instruction length for existing 8/16-bit and 32-bit code. These facts enabled Cyrix to reduce the bus width required to handle most data and code transactions to 32 bits. To exploit the inherent parallelism, the 5x86 utilizes decoupled units interconnected with multiple 32-bit, split-transaction buses.

The 5x86 processor employs a dedicated branch unit including a branch target buffer, a 16-KByte unified write-back cache, a Floating Point Unit, and an instruction fetch and instruction decode unit. The Memory Management Unit contains a 32-entry translation lookaside buffer, a load/store unit capable of managing concurrent operations, and an address calculation unit. The 5x86 functional units are interconnected by two 32-bit buses that permit non-blocking operation of the units. A 128-bit instruction fetch bus feeds 16 bytes of code per cycle to a three-line-deep buffer in the instruction decode unit.

Integer Unit
The superpipelined Integer Unit fetches, decodes, and executes x86 instructions through the use of a six-stage integer pipeline.

1. The Instruction Fetch Stage generates a continuous, high-speed instruction stream from the on-chip cache. Up to 128 bits of code are read during a single clock cycle.
2. The Instruction Decode Stage evaluates the code stream provided by the instruction fetch stage and determines the number of bytes in each instruction and the instruction type. Instructions are processed and decoded at a maximum rate of one instruction per clock.
3. The Address Calculation function is superpipelined to contain two stages -- AC1 and AC2. If the instruction refers to a memory operand, AC1 calculates a linear memory address for the instruction. AC2 performs any required memory management functions, cache accesses, and register file accesses. If a floating point instruction is detected, AC2 sends it to the FPU for processing.
4. The Execution Stage, under control of microcode, executes instructions using the operands provided by the address calculation stage.
5. Write-Back updates the register file within the integer unit, or writes to the load/store unit within the Memory Management Unit.

Floating Point Unit (FPU)
The 5x86 FPU is based on the same core as that found in Cyrix's sixth-generation M1 processor. The FPU interfaces with the integer unit and the cache unit through a 64-bit interface. It is x87 instruction-set compatible (including the extended 80-bit format) and adheres to the IEEE-754 standard. Since most applications contain FPU instructions mixed with integer instructions, the 5x86 achieves high performance by completing integer and FPU operations in parallel.

Write-Back Cache
The 5x86 implements a 16-KByte, four-way set associative unified instruction/data cache that can operate in either write-back or write-through mode. It has a dedicated 128-bit port for transferring instructions to the IF unit, and a 64-bit wide data port that can be split into two 32-bit data paths. The cache is arranged as four sets of 256 lines per set with 16 bytes per line. Cache buffers allow an entire cache line to be read or written in a single clock cycle to maximize cache bandwidth. Since the 5x86 is scalar and implements these buffers, it alleviates the need for more sophisticated cache banking techniques for concurrent accesses.

Memory Management Unit (MMU)
The 5x86 MMU contains the load/store unit, the 32-entry translation lookaside buffer (TLB), and the address calculation (AC) unit. The AC unit performs all address calculations, maintains instruction pointers for each pipeline stage, and initiates load and store transfers. The advanced load/store unit is capable of managing concurrent operations and processing loads and stores out of order while maintaining a three-deep load queue and four-deep store queue.

The Bus Interface Unit
The 5x86 64-bit internal bus is tapered down to a 32-bit external bus to allow the processor to fit existing designs, a strategy that minimizes customers' development cycles. The 5x86/100 MHz core speed option can operate with a bus speed of either 33 MHz or 50 MHz. The 120 MHz core speed option operates with a bus speed of 40 MHz. Eight buffers allow sufficient buffering of write activity to maintain bandwidth for read operations, thus reducing pipeline stalls. The bus protocol is standard except for an optional higher-performance linear burst mode, which can be implemented instead of the Cyrix "one-plus-four" mode. The one-plus-four mode is compatible with all existing 32-bit chipsets.

Power Management
The 5x86 was engineered with advanced power management features. The processor monitors and automatically powers down the FPU and other idle internal circuits. Each 32-bit section of the 64-bit internal data bus is driven only when needed. The core design of the 5x86 is completely static to allow for easy clock manipulation, a feature commonly used to adjust processor power consumption. Additionally, the System Management Mode (SMM) software model implemented is compatible with past and future Cyrix processors and can be used to perform processor and system power conservation tactics.

System-Level Performance Results
Windows®-based benchmarks execute scripts that run actual Windows applications such as word processing, desktop publishing, database management, computer-aided design (CAD), and scientific/engineering (MathCAD) software. These benchmarks generally weight the results of each application (based on the application's unit market share) and derive a composite performance score.

Processor-Intensive Performance Results
DOS-based benchmarks illustrate 32-bit and 16-bit performance of the Cyrix 5x86 processor. Some of these benchmarks are small enough to fit in the 16-KByte on-chip cache, avoiding any dependency on the design or the second-level cache. CPUmark₁₆ is a large-format benchmark that does not fit in the internal cache. It demonstrates the performance of the combined CPU and memory subsystem, including the external second-level cache.

*System Configurations
Cyrix reference systems: 120 MHz 5x86 CPU, Chicony 471A or ECS UP 8810-AIO Motherboard, 16MB DRAM, 256K L2 cache, Diamond Stealth VLB - Chicony Platform and Diamond Stealth 64 PCI - ECS platform, 64-bit VRAM, Conner CFA 540A, 524MB hard disk, IDE (on board I/F). Windows for WorkGroups 3.11, 20MB SWAP file, 32 bit file and disk access.
Gateway P90: 90 MHz Pentium® processor, 16MB DRAM, no L2 cache, Diamond Stealth 64 PCI, 2MB VRAM, Conner CFA540A IDE, 524MB hard disk, 10ms access time, on board PCI IDE. Windows for WorkGroups 3.11, 20MB SWAP file, 32 bit file and disk access.
Dell P90: 90 MHz Pentium® processor, 16MB DRAM, 256K L2 cache, Number Nine PCI, Conner CFA540A IDE, 524MB hard disk, 10ms access time, on board PCI IDE. Windows for WorkGroups 3.11, 20MB SWAP file, 32 bit file and disk access.
Pentium® is a registered trademark of Intel Corporation. All other brand or product names are trademarks or registered trademarks of their respective holders.

Cyrix 5x86/100 Performance Benchmarks*

*All systems tested with 256K L2 cache.
(1) All systems tested with identical hard drive and graphics card.