Hubbry Logo
R4200R4200Main
Open search
R4200
Community hub
R4200
logo
7 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
R4200
R4200
R4200
View on Wikipedia
from Wikipedia

The R4200 is a microprocessor designed by MIPS Technologies that implemented the MIPS III instruction set architecture and was initially referred to as the VRX during development. MIPS, which had no production capability of its own, licensed the design to NEC, which fabricated and marketed it as the VR4200. The first VR4200, running at 80 MHz, was introduced in 1993, with a faster 100 MHz version following in 1994.

Primarily aimed at low-power Windows NT systems such as personal computers and laptops,[1]: 468  the R4200 was marketed as offering "Pentium processor performance at a tenth of the price." It was initially expected to deliver twice the performance of a 66 MHz Intel 486DX2 processor. SPECint benchmark results showed the microprocessor’s integer performance at approximately 85% of the original Pentium, while its floating-point performance was about half that of the Pentium.

The R4300i is a derivative of the R4200, designed by MIPS for embedded applications. A variant of the R4300i was used in the widely popular Nintendo 64 and SNK's Hyper Neo Geo 64 arcade board.

The R4200 never saw use in personal computers and was eventually repositioned as an embedded microprocessor complementing the R4600.

Description

[edit]

The R4200 is a scalar design with a five-stage classic RISC pipeline. Notably, floating point mantissa calculation reused the 64-bit integer datapath (only the exponent needed a separate 12-bit datapath).[2] This scheme reduced chip size and transistor count, reducing cost and power consumption. Whilst this reduced floating point performance, the R4200's intended applications did not require high floating point performance.

The R4200 has a 16 kB instruction cache and an 8 kB data cache. Both caches are direct-mapped. The instruction cache has a 32-byte line size, whereas the data cache has 16-byte line size. The data cache uses the write-back write protocol.

The R4200 has a 32-entry translation lookaside buffer (TLB) for data, and a 4-entry TLB for instructions. A 33-bit physical address is supported. The system bus is 64 bits wide and operates at half the internal clock frequency.

The R4200 contained 1.3 million transistors and had an area of 81 mm2. NEC fabricated the R4200 in a 600 nm process with three levels of interconnect. It was packaged in a 179-pin ceramic pin grid array that was compatible with the R4x00PC and R4600, or a 208-pin plastic quad flat pack (PQFP). It used a 3.3 V power supply, dissipating 1.8 W typically and a maximum of 2 W at 80 MHz.

In comparison to the Pentium, SPECint ratings had the Pentium at 64.5 and the R4200 at 55. SPECfp ratings had the Pentium at 56 versus the R4200 at 30.[3]

R4300i

[edit]
CPU-NUS in a Nintendo 64. It is a custom version of the NEC R4300 and is based on the R4300i

The R4300i is a derivative of the R4200, designed by MIPS for embedded applications and announced on 17 April 1995.[4] It improves on the R4200 with a faster integer multiplier featuring lower latency and a simplified 32-bit system bus to reduce costs. The chip, measuring 45 mm², was fabricated using a 350 nm process and was packaged in a low-cost 120-pin PQFP by employing multiplexed address and data lines. It operates on a 3.3 V power supply and dissipating 1.5 W at 40 MHz (80 MHz internally),[5] 1.8 W at 100 MHz and 2.2 W at 133 MHz.

The R4300i was licensed to NEC and Toshiba, who marketed it as the VR4300 and TX4300, respectively, with 100 and 133 MHz versions.

NEC produced a version of the VR4300 for the Nintendo 64 called the CPU-NUS, clocked at 93.75 MHz with a performance of 125 million instructions per second.[6] Popular Electronics compared its processing power to that of contemporary Pentium desktop processors.[7] Though constrained by a narrower 32-bit system bus, the VR4300 retained the computational capabilities of the more powerful 64-bit R4300i.[8] However, software rarely utilized 64-bit precision, as Nintendo 64 games primarily relied on faster and more compact 32-bit operations.[9]

NEC also developed two other derivatives for the embedded market, the VR4305 and VR4310, announced on 20 January 1998.[10] The VR4310, available at 100, 133, or 167 MHz, was manufactured in a 250 nm process and packaged in a 120-pin PQFP.

References

[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia

R4200

View on Grokipedia
from Grokipedia
The R4200 is a low-power, low-cost (RISC) designed by that implements the 64-bit and is fully compatible with the earlier family. Targeted primarily at portable computing devices such as laptops and embedded applications like printers and routers, it achieves its efficiency through a compact featuring a unified and floating-point data path, a 5-stage , and on-chip caches totaling 24 KB (16 KB for instructions and 8 KB for data). Introduced in 1993, the R4200 operates at internal clock speeds up to 80 MHz (with a 40 MHz external bus) on a 0.6-micron process with approximately 1.3 million transistors, consuming just 1.5 W at full speed and supporting low-power modes including a power-down state for quick reactivation in intermittent-use scenarios like notebook computing. Performance benchmarks for the R4200, based on simulations at the time of its announcement, include 55 SPECint92 for integer operations and 30 SPECfp92 for floating-point tasks, positioning it to outperform a 66 MHz 486DX2 by approximately 2:1 while competing with emerging processors like the and / PowerPC 601 in low-power contexts. The chip was manufactured initially by Electronics under an exclusive agreement, with sampling beginning in the third quarter of 1993 and volume production ramping up in early 1994, aimed at enabling cost-effective systems priced between $1,000 and $2,000, particularly for running Microsoft Windows NT on mobile platforms. Its design innovations, such as an optional removable (MMU) and system interface for embedded variants, along with support for plastic packaging options like a 208-pin PQFP, underscored MIPS's focus on versatility and energy efficiency in the early microprocessor market.

Overview

Description

The R4200 is a scalar reduced instruction set computing (RISC) microprocessor that implements the MIPS III (ISA), developed by and announced in May 1993. It features a 5-stage . Its primary goals centered on lowering costs and power dissipation relative to the R4000 series, achieved through a simplified that reuses the 64-bit integer datapath for arithmetic operations on the mantissa portion of floating-point numbers, while providing a dedicated multiplier for exponents without a full separate . Initially targeted at low-power workstations and laptops to deliver competitive performance in portable , the R4200 was also positioned for embedded applications due to its efficiency and optional logic core removals for customized designs. Fabricated by on a 600 nm process, the R4200 incorporates 1.3 million transistors across a compact die measuring 81 mm².

Key Specifications

The MIPS R4200 was fabricated using a 600 nm process with three levels of interconnect by , which produced it under the designation VR4200. It supports initial clock speeds of 80 MHz for the core (with the at half speed, or 40 MHz), and later variants reached 100 MHz in 1994. The processor operates on a 3.3 V , with typical power consumption of 1.5 W and a maximum of 2 W at 80 MHz. It includes a 16 KB direct-mapped instruction cache and an 8 KB direct-mapped data cache. Packaging options include a 179-pin ceramic (PGA) compatible with the R4000PC and R4600, or a 208-pin quad flat pack (PQFP) for the low-profile R4200LP variant. The R4200 features a 33-bit space and a 64-bit multiplexed that runs at half the core clock frequency.

Development

Historical Background

The R4200 microprocessor was announced by on in 1993 as a cost-reduced follow-up to the , which had debuted in 1991 and established the foundation for MIPS's 64-bit RISC architecture. Designed to implement the MIPS III instruction set at lower power and cost, the R4200 targeted emerging opportunities in portable computing and entry-level systems, with an 80 MHz internal clock speed (40 MHz external bus) and a focus on enabling efficient 64-bit operation in compact form factors. MIPS Technologies, the semiconductor design arm originally formed as part of MIPS Computer Systems in and restructured to emphasize licensing by the early , positioned the R4200 as a strategic response to intensifying RISC competition. The company aimed to challenge Intel's dominance in the low-end PC segment by providing an affordable alternative for Unix and Windows NT-based systems, where demand was growing for RISC processors that balanced performance with reduced complexity and power consumption. This move reflected broader market pressures in the early , as RISC architectures from competitors like and PowerPC proliferated, prompting MIPS to prioritize scalable, licensable designs over high-end workstation exclusivity. Initially targeted at PCs and laptops to capitalize on the rise of and Microsoft's platform, the R4200's market positioning evolved rapidly after its introduction due to shifting industry dynamics, including the entrenched x86 ecosystem and slower adoption of non-Intel RISC in consumer desktops. By the mid-1990s, as and PC opportunities waned amid fierce competition, MIPS redirected emphasis toward embedded applications, where the R4200's low-power profile and MIPS III compatibility proved advantageous for networking, , and industrial systems. This pivot aligned with MIPS's licensing model, enabling broader proliferation in specialized markets and contributing to the architecture's sustained relevance beyond general-purpose computing.

Design and Production

The R4200 microprocessor was engineered with key optimizations to achieve a compact die size and low power consumption while supporting 64-bit operations, targeting cost-sensitive portable and embedded applications. To minimize silicon area, designers reused the integer datapath for floating-point mantissa calculations, forgoing a dedicated floating-point unit and instead handling such operations through the existing integer hardware. This approach, combined with the omission of specialized hardware for integer and floating-point multiplication, secondary cache support, and multiprocessor capabilities, significantly simplified the design relative to the higher-end R4000, reducing transistor count to 1.3 million on an 81 mm² die. The processor adopted a straightforward five-stage pipeline, abandoning superpipelining techniques to curb complexity and energy use, while fully incorporating the MIPS III instruction set architecture, including 64-bit integer extensions for compatibility with advanced workloads. Production of the R4200 was handled exclusively by , which branded it as the VR4200 and funded the effort independently. Fabrication utilized NEC's 0.6-micron three-layer-metal process, chosen for its balance of yield, cost, and performance in volume . secured sole production rights for the first year, with the chip offered in 179-pin PGA and 208-pin PQFP packages to suit different needs. Addressing challenges in delivering 64-bit MIPS III capabilities at low power for portable systems, the R4200 operated at 3.3 , dissipating just 1.5 W typically through features like cycle-by-cycle shutdown of idle blocks, banked cache organization to reduce active power, and a low-speed quarter mode at 20 MHz. These measures ensured suitability for battery-powered environments without sacrificing core architectural integrity. Release milestones included design in June 1993, first shortly thereafter, sampling in the third quarter of 1993, and full production ramp-up by the first half of 1994, starting with an 80 MHz internal clock (40 MHz ).

Architecture

Processing Core

The R4200 microprocessor features a classic five-stage RISC pipeline tailored for scalar processing, consisting of instruction fetch, decode, execute, memory access, and writeback stages that handle both integer and floating-point operations efficiently. This design draws from the simpler R3000 pipeline, avoiding the eight-stage superpipelining of the R4000 to prioritize low power and cost in portable applications, while introducing single-cycle stalls for branches and loads to manage dependencies without excessive complexity. The processor implements the full MIPS III instruction set architecture, supporting 64-bit integer arithmetic, load/store operations, and floating-point instructions compliant with standards, but excludes advanced MIPS IV extensions such as paired single-precision floating-point operations. This compatibility ensures seamless execution of existing MIPS software, with the instruction set emphasizing reduced instruction set principles for high performance in embedded and portable systems. Central to the execution core are single-issue units: a 64-bit arithmetic logic unit (ALU) that performs additions, subtractions, shifts, operations, and leading zero counts in a single cycle, paired with a that reuses the integer datapath's mantissa path—supplemented only by an exponent shifter—for double-precision computations. This shared minimizes area and power draw, enabling floating-point additions in three cycles and multiplications in 11 or 20 cycles depending on precision, without a dedicated multiplier hardware block; instead, it employs a 3-bit Booth recoding for efficiency. Branch instructions adhere to the MIPS architecture's mechanism, where the subsequent instruction is always executed irrespective of the branch resolution, combined with a one-cycle to resolve the condition and update the . This approach, standard across MIPS implementations, balances simplicity and performance by allowing compiler optimizations to fill s productively, without dynamic prediction hardware to keep the core lightweight.

Memory and Bus System

The R4200 integrates a compact on-chip in configurations that include it, optimized for low power and cost, featuring a 16 KB direct-mapped instruction cache with 32-byte line size and an 8 KB direct-mapped data cache with 16-byte line size. Both caches are physically tagged and virtually indexed to enable parallel address translation and access, with the data cache employing a write-back protocol and write allocation to minimize external bus traffic. Neither cache includes write-back for instructions nor a victim cache mechanism, and they are divided into four banks each for early decoding that reduces dynamic power by deactivating unused banks during accesses. The caches, along with the and system interface, can be omitted in embedded variants to reduce cost and power. Virtual-to-physical address translation is handled by a (MMU) with a 32-entry main (TLB) shared primarily for data operations and a separate 2-entry micro-TLB dedicated to instructions, allowing simultaneous translation of two sequential instruction addresses without stalling the . The TLB entries support both standard 4 KB pages and large 16 MB pages, facilitating efficient mapping in embedded and portable systems. A one-entry write buffer further aids memory operations by absorbing write requests to prevent CPU stalls on stores to main memory or I/O. The processor employs a 33-bit space, enabling up to 8 GB of addressable memory, while supporting 64-bit virtual addressing in compliance with the MIPS III . Externally, it interfaces via a 64-bit multiplexed (SysAD) that runs at half the core clock frequency—for example, 40 MHz when the core operates at 80 MHz—and supports big-endian or little-endian byte ordering, selectable via a configuration pin, for consistency in MIPS-compatible systems. This bus design maintains compatibility with R4000-family chipsets and supports external secondary cache controllers for further hierarchy extension.

Variants

R4300i

The R4300i, introduced on April 17, 1995, by in collaboration with , serves as the primary embedded variant of the R4200, tailored for cost-sensitive and interactive applications. It incorporates a 0.35 μm process technology, achieving a compact die size of 45 mm² while maintaining the core MIPS III . A key adaptation is the reduction to a 32-bit external bus—despite preserving the internal 64-bit —to minimize pin count and enable simpler integration in embedded systems. Optimized for low power, the R4300i operates across clock speeds from 40 MHz to 133 MHz on a 3.3 V supply, with typical consumption of 1.8 W at 100 MHz and up to 2.2 W at higher frequencies. It is housed in a 120-pin plastic (PQFP), facilitating use on compact boards. Compared to the R4200, these modifications prioritize pin efficiency and reduced power draw without sacrificing the unified integer and floating-point execution core. produced a licensed version known as the VR4300, with samples available from May . Notable applications include its integration as the main processor in the gaming console.

NEC Derivatives

developed the VR4305 and VR4310 as specialized variants in the R4300i lineage, targeting embedded applications with enhanced integration. These processors were announced in February 1998, and fabricated using a to enable compact, efficient designs suitable for space-constrained systems. The VR4310, a key model in this series, operated at clock speeds ranging from 100 to 167 MHz while maintaining a 32-bit external bus interface for compatibility with existing embedded infrastructures. Compared to the R4300i, it demonstrated improved power efficiency, consuming approximately 50% of the power of its predecessor at equivalent frequencies due to process optimizations and architectural refinements. Design enhancements emphasized suitability for real-time systems, including advanced handling with support for five general-purpose interrupts, non-maskable interrupts, and precise exception mechanisms via the ERET instruction and status register controls. These features facilitated low-latency responses without altering the core MIPS III . The processors retained the R4300i core as their foundation, adapting it for embedded constraints. Production of the VR4305 and VR4310 was phased out by the early , coinciding with ' transition to the MIPS32 and MIPS64 architectures for next-generation embedded processors.

Applications

Embedded Systems

The R4200 family, particularly its R4300i derivative, found significant adoption in embedded systems due to its low power consumption, cost-effectiveness, and compatibility with the MIPS III instruction set, making it suitable for consumer and industrial hardware. The primary application was in gaming consoles, where the VR4300—a customized version of the R4300i—powered the , released in 1996. This processor operated at 93.75 MHz, providing approximately 125 MIPS for game logic processing, which balanced performance with the console's compact design. Another notable gaming implementation was in the arcade system, launched in 1997, which utilized an VR4300 clocked at 100 MHz to handle 64-bit and audio I/O tasks in a high-end arcade environment. Beyond gaming, the R4300i targeted embedded controllers in networking devices and other industrial applications, such as routers and printers, by the late 1990s, leveraging its integrated and high-bandwidth interface for efficient data handling. These deployments capitalized on the processor's modular design, allowing cache and MMU components to be omitted for space-constrained systems. The family's market impact was profound, enabling affordable 64-bit processing in embedded contexts and contributing to millions of shipments through the alone, which sold 32.93 million units worldwide. This widespread integration demonstrated the R4200 lineage's role in democratizing advanced RISC capabilities for non-workstation uses. Its legacy influenced ongoing MIPS licensing in , paving the way for transitions to higher-performance series like the R5000 in the mid-1990s.

Performance Metrics

The MIPS R4200 achieved SPECint92 performance of 55 and SPECfp92 performance of 30 when operating at an internal clock speed of 80 MHz. In comparison, the processor at 66 MHz delivered 64.5 SPECint92 and 56 SPECfp92, highlighting the R4200's competitive integer performance but lower floating-point throughput relative to contemporary x86 designs. These benchmarks underscored the R4200's efficiency in low-power scenarios, with typical power dissipation of 1.5 W at 80 MHz, enabling its suitability for portable systems. Among variants, the R4300i, a licensed to multiple manufacturers, provided 60 SPECint92 and 45 SPECfp92 at 100 MHz, alongside approximately 125 MIPS. In the console, the VR4300 implementation of the R4300i core ran at 93.75 MHz and sustained around 125 MIPS in peak operation, demonstrating effective real-world throughput for embedded tasks. Later derivatives, such as the VR4310 fabricated on a 0.28-micron process, supported clock speeds up to 167 MHz internally, scaling performance toward 221 MIPS while halving power consumption relative to prior generations at equivalent frequencies. Performance was influenced by the processor's five-stage pipeline, which achieved high efficiency through branch delay slots and minimal stalls due to the integrated 16 KB instruction and 8 KB data caches. The power-per-performance ratio stood at approximately 1.5 W for 55 SPECint92, prioritizing energy efficiency over raw speed. Relative to the higher-end R4400, which reached 117 SPECint92 at 200 MHz, the R4200 offered lower overall throughput but excelled in 64-bit integer tasks compared to 32-bit contemporaries like the 80486, where native 64-bit support provided up to 30% better efficiency in compatible applications.

References

  1. https://ieeexplore.ieee.org/document/282950
  2. https://websrv.cecs.uci.edu/~papers/mpr/MPR/ARTICLES/070701.pdf
  3. https://www.halfhill.com/byte/1993-7_low-power.html
  4. https://ieeexplore.ieee.org/document/282950/
  5. https://www.electronicproducts.com/low-power-mips-processor-has-plastic-package-power-management/
  6. http://kpolsson.com/micropro/proc1993.htm
  7. https://semiwiki.com/x-subscriber/imagination-technologies/1881-a-brief-history-of-the-mips-architecture/
  8. https://cpushack.com/CIC/announce/1995/MIPSR4300i.html
  9. https://datasheets.chipdb.org/MIPS/R4300i_datasheet.pdf
  10. https://foldoc.org/MIPS%2BTechnologies%2C%2BInc.
  11. https://www.cecs.uci.edu/~papers/mpr/MPR/ARTICLES/090601.pdf
  12. https://datasheet.datasheetarchive.com/originals/distributors/Datasheets-313/224339.pdf
  13. https://datasheets.chipdb.org/NEC/Vr-Series/Vr43xx/U10504EJ7V0UMJ1.pdf
  14. https://www.toppy.org.uk/documents/U11455EU4V0SG00.pdf
  15. http://ieeexplore.ieee.org/document/282950/
  16. https://www.copetti.org/writings/consoles/nintendo-64/
  17. http://www.sgidepot.co.uk/sgi2.html
  18. https://www.system16.com/hardware.php?id=870
  19. https://www.nintendo.co.jp/ir/en/finance/hard_soft/index.html
  20. https://www.cpushack.com/MIPSCPU.html
  21. http://www.sgidepot.co.uk/mips-nec.html
  22. https://openlab.citytech.cuny.edu/ggordon-eportfolio/files/2018/09/Nintendo-64-Product-Overview.pdf
  23. https://www.semanticscholar.org/paper/R4200%253A-a-high-performance-MIPS-microprocessor-for-Zivkov-Ferguson/af0a8fe3f0d415d0d04781d67e8a9750c9bfa294
  24. http://archive.irixnet.org/siliconsurf/news/MIPS_sampling.html
Add your contribution
Related Hubs
User Avatar
No comments yet.