Pentium II
The Pentium II is a sixth-generation x86-compatible microprocessor developed by Intel Corporation and introduced on May 7, 1997, succeeding the Pentium Pro and incorporating enhancements from the Pentium MMX for multimedia processing.[1] It employs the P6 microarchitecture, featuring a single-core design with superscalar, out-of-order execution, dynamic branch prediction, and support for MMX instructions to accelerate vector operations.[2] The processor was fabricated initially on a 0.35-micrometer process with 7.5 million transistors, delivering approximately 36% greater performance than its predecessor through a 12-stage pipeline and dual integer execution units.[1][2] Key specifications include clock speeds starting at 233 MHz and scaling up to 450 MHz in later models, with an integrated 32 KB L1 cache (16 KB instruction and 16 KB data, four-way set-associative) and external L2 cache ranging from 256 KB to 2 MB housed in a Single Edge Contact (SEC) cartridge for the Slot 1 interface.[2] The design utilized a 66 MHz front-side bus (upgradable to 100 MHz in some variants) and supported up to 64-bit data transfers, enabling efficient burst modes for memory access.[2] Variants included the Klamath core (0.35 μm) for initial desktop releases and the Deschutes core (0.25 μm) for higher-performance iterations, alongside mobile and server-oriented Pentium II Xeon models introduced in 1998 with larger caches and support for multi-processor configurations.[2] The Pentium II marked a pivotal shift in consumer computing by packaging the CPU, L2 cache, and thermal interface in a removable cartridge, facilitating easier upgrades and cooling while targeting multimedia and general desktop applications.[1] It powered early systems like those with the Intel 440BX chipset, contributing to the rise of 3D graphics and internet-enabled PCs in the late 1990s, before being superseded by the Pentium III in February 1999 with added SSE instructions.[2]Introduction and Development
Overview
The Pentium II is a sixth-generation x86 microprocessor developed by Intel, introduced on May 7, 1997, and officially discontinued on January 1, 2005.[3][4] It served as a key product in Intel's lineup, evolving directly from the Pentium Pro to address growing demands in both consumer and enterprise computing environments.[5] While retaining compatibility with server systems through its shared architecture, the Pentium II was primarily positioned to drive multimedia-enabled consumer desktops, incorporating enhancements that improved performance for everyday applications and emerging digital content.[6][7] At its core, the Pentium II utilized the P6 microarchitecture, an advanced superscalar design originally pioneered in the Pentium Pro, but with the addition of MMX instructions to support multimedia processing.[5] Early desktop variants, known as the Klamath core, featured approximately 7.5 million transistors fabricated on a 0.35-micrometer process.[8] Later mobile versions utilized a 0.25-micrometer process for improved power efficiency. This progression allowed the processor to balance power efficiency and performance, making it suitable for a wide range of PCs during the late 1990s. Upon launch, Intel priced the 233 MHz Pentium II model at $637 in quantities of 1,000 units, reflecting its premium status in the market for high-performance computing components.[9] This pricing strategy helped position the Pentium II as an accessible upgrade for consumers seeking enhanced multimedia capabilities without fully diverging from enterprise-grade reliability.Historical Context
The Pentium II processor originated as a consumer-oriented derivative of Intel's Pentium Pro project, which began development in the early 1990s and culminated in the Pentium Pro's release on November 1, 1995, primarily targeting server and workstation markets with the new P6 microarchitecture. Recognizing the limitations of the Pentium Pro for mainstream personal computing—particularly its suboptimal handling of legacy 16-bit code—Intel initiated the Pentium II as an adaptation to broaden market appeal while retaining the core P6 design for compatibility and performance scaling. This evolution addressed the need for a processor that could serve the growing consumer PC segment without requiring a complete architectural overhaul.[10] Central to the Pentium II's development goals were enhancements to boost 16-bit code execution efficiency, making it more suitable for the prevalent legacy software in consumer environments, alongside the integration of MMX technology to accelerate emerging multimedia workloads such as audio, video, and graphics processing. The transition from the Pentium Pro's Socket 8 to the new Slot 1 form factor was a strategic move to streamline manufacturing by offloading the L2 cache to a separate cartridge, which improved production yields compared to on-package integration and facilitated easier end-user upgrades through a standardized slot-based interface. These changes aimed to reduce costs and enhance scalability for desktop systems.[10] Facing intensifying competition from AMD's K6 processor, which was poised for release in April 1997 and threatened Intel's dominance in the mid-range PC market, Intel accelerated the Pentium II's production timeline to ensure a swift market entry. This rush contributed to early manufacturing hurdles on the 0.35-micron (350 nm) process used for the initial Klamath core, including lower initial yields that temporarily constrained supply. Ultimately, Intel announced and launched the Pentium II on May 7, 1997, positioning it as a high-performance option for business and consumer desktops with initial speeds of 233 MHz and 266 MHz.[11][3]Architecture and Key Features
Microarchitecture
The Pentium II processor is based on Intel's P6 microarchitecture, which employs a triple-issue superscalar design capable of dispatching up to three instructions per clock cycle to enhance instruction-level parallelism. This design decodes x86 instructions into micro-operations (μops) and supports a 12-stage pipeline that allows for deeper pipelining and higher clock frequencies compared to prior architectures, with stages including fetch, decode, dispatch, execute, and retire. Advanced dynamic branch prediction is integrated via a 512-entry branch target buffer (BTB), which stores branch history and target addresses to predict branch outcomes and minimize pipeline stalls, achieving prediction accuracies often exceeding 90% in typical workloads.[12][13] Out-of-order execution is facilitated by a reservation station and reorder buffer, enabling the processor to dynamically schedule μops based on data dependencies rather than program order, thereby hiding latency from long-execution instructions. This mechanism uses 20 physical registers for integer operations to support register renaming and avoid false dependencies.[2][12]) The front-end features three decoders: two simple decoders that each handle one basic x86 instruction per cycle into one μop (totaling two), and one complex decoder for more intricate operations that can produce up to four μops from a single instruction, allowing up to three x86 instructions per cycle, ensuring efficient instruction breakdown before dispatch.[2][12]) The execution core includes two integer pipelines, designated U and V, each capable of performing arithmetic and logical operations, with the U-pipe also handling jumps. Floating-point operations are managed by a dedicated unit comprising one multiplier for multiply and fused multiply-add instructions, alongside a divider for reciprocal and square-root approximations. MMX instructions integrate seamlessly with the floating-point unit for SIMD processing. For power efficiency in desktop applications, the microarchitecture incorporates Stop-Grant mode, which halts core clocks while maintaining bus snooping, and Sleep mode, which further reduces power by stopping nearly all internal clocks except the phase-locked loop.[12]Cache and Memory Subsystem
The Pentium II processor features a two-level on-chip cache hierarchy designed to minimize latency for frequently accessed data and instructions. The Level 1 (L1) cache consists of a separate 16 KB instruction cache and a 16 KB data cache, both organized as 4-way set associative with 32-byte cache lines. The instruction cache operates in a non-blocking manner, allowing continued fetches during misses, while the data cache employs a write-back policy across two 8 KB banks to support efficient write operations and reduce bus traffic.[2] This configuration, running at full core clock speed, provides low-latency access—typically 3-4 cycles for hits—enhancing overall instruction throughput in the P6 microarchitecture. The Level 2 (L2) cache is implemented off-die within the Single Edge Contact Cartridge (SECC) packaging, using pipelined burst synchronous static RAM (BSRAM) for high-speed operation. Standard configurations offer 256 KB or 512 KB of unified L2 cache, organized in a 4-way set associative structure with non-blocking access to handle multiple outstanding requests.[2] Unlike the integrated L1, the L2 runs at half the processor core frequency (e.g., 133 MHz for a 266 MHz core) in synchronous burst mode, enabling burst transfers of up to 32 bytes per cycle over a dedicated path, though this speed differential introduces higher latency of 8-12 cycles for hits compared to L1.[14] Later variants, such as the Dixon core, support full-speed L2 operation to further reduce this penalty.[2] The front-side bus (FSB) serves as the primary interface between the processor, L2 cache, and system memory, utilizing a 64-bit data path with GTL+ signaling for reliable high-speed communication. Operating at a standard frequency of 66 MT/s, the FSB supports pipelined, split-transaction protocols for efficient address and data phases, allowing up to four pending transactions to overlap and improve bandwidth for cache misses.[12] Subsequent revisions enable upgrades to 100 MT/s, doubling potential throughput to approximately 800 MB/s while maintaining compatibility with the 36-bit physical address bus.[12] Memory addressing in the Pentium II leverages Physical Address Extension (PAE) for a 36-bit physical address space, theoretically supporting up to 64 GB of RAM through signals A[35:3]. However, early implementations, such as the Klamath core, constrain cacheable memory to 512 MB due to limitations in the external tag RAM size, beyond which accesses fall back to slower uncached modes.[14] Later models expand this to 4 GB cacheable via enhanced tag addressing (e.g., using ASZ[1:0]# for regions 0-4 GB or 4-64 GB), with optional ECC support on the FSB for error correction in server environments.[12] This subsystem prioritizes MESI cache coherency to maintain consistency across multiprocessor configurations.[12]Instruction Set Extensions
The Pentium II processor incorporated the MMX (MultiMedia eXtensions) instruction set, which added 57 new single instruction, multiple data (SIMD) instructions to the P6 architecture for enhanced multimedia processing.[15][16][17] These instructions operate on 64-bit packed integer data, enabling parallel computations across multiple elements such as 8 bytes, 4 words, or 2 doublewords within a single register, thereby accelerating tasks like audio and video manipulation without requiring dedicated hardware.[15][17] MMX supports a range of integer operations tailored for multimedia applications, including arithmetic, logical, shift, and multiply functions on packed data types. For instance, the PADDSB instruction performs saturated addition on eight packed bytes, preventing overflow by clamping results to the valid range, which is particularly useful for signal processing in audio and video streams.[16][17] Additionally, the EMMS instruction clears the MMX state by setting all floating-point stack tags to empty and resetting the top-of-stack pointer, ensuring proper transition back to scalar floating-point code.[15][16] These extensions provided significant performance improvements, achieving up to 4x speedup in multimedia workloads such as MPEG video decoding compared to traditional scalar implementations on prior processors.[16][17] MMX maintains compatibility with the existing x87 floating-point unit (FPU) by aliasing its eight 64-bit registers (MM0 through MM7) to the FPU's stack registers, but programmers must explicitly use EMMS to avoid contaminating the FPU state with integer data tags, which could otherwise lead to incorrect floating-point results.[15][16]Packaging and Form Factors
Slot 1 Cartridge
The Slot 1 cartridge for the Pentium II processor employed the Single Edge Contact Cartridge (SECC) packaging, also known as Single Edge Processor Package (SEPP) in its uncovered variant, which utilized a 242-pin edge connector with gold-plated fingers on an FR-4 substrate for interfacing with the motherboard's Slot 1 socket.[18] This design featured a two-sided staggered pin configuration at a 1.0 mm pitch, including a key slot for proper alignment and insertion, enabling a socketable connection that protected internal components from handling damage.[18] Key integrated components within the SECC included the CPU die mounted via a flip-chip ball grid array (FC-BGA), off-die L2 cache chips consisting of TagRAM and burst static RAM (BSRAM) dies for up to 512 KB of cache, a voltage regulator module (VRM) for on-cartridge power regulation, and mounting points for a heat sink with an integrated thermal plate in early SECC versions.[18] The cartridge's overall dimensions were approximately 5.5 inches in length and 2.5 inches in height, with the SECC2 variant later omitting the thermal plate to accommodate flip-chip transitions while retaining the edge connector.[18] This form factor provided several advantages over prior Pin Grid Array (PGA) sockets, including simplified assembly processes due to the cartridge's self-contained structure, enhanced support for larger external cache implementations without on-die constraints, and a direct upgrade compatibility with Slot 1 systems from the preceding Pentium Pro processor.[18] Electrically, the design operated with core voltages between 2.0 V and 2.8 V, managed via Voltage Identification (VID) pins connected to the VRM, and early models maintained a thermal design power (TDP) of up to 21 W.[14] Additionally, it incorporated a back-side bus architecture that separated the L2 cache bus—running at half the core frequency—from the system bus, thereby decoupling internal processor operations from external I/O for improved efficiency.[14]Mobile Modules
The Mobile Pentium II processors were packaged in specialized form factors to address the constraints of laptop and notebook systems, prioritizing reduced size, lower power consumption, and improved thermal management compared to desktop variants. The primary packaging for early mobile implementations was the Mobile Module Cartridge (MMC), a miniaturized adaptation of the Slot 1 cartridge designed for integration into compact motherboards. The MMC-1 featured a 280-pin surface-mount connector with a 0.6 mm pitch, enabling a compact footprint of approximately 102 mm x 64 mm while incorporating the processor die, external L2 cache chips, and an integrated voltage regulator module (VRM) to supply core voltages typically ranging from 1.6 V to 1.9 V, depending on the clock speed.[19] This on-module VRM accepted a wide input range of 5 V to 21 V (V_DC) and generated regulated outputs such as 3.3 V for I/O operations and 2.5 V for CPU I/O, facilitating efficient power delivery without requiring extensive external circuitry on the motherboard.[19] Subsequent iterations introduced the MMC-2, which utilized a 400-pin connector to support enhanced interfaces like AGP/PCI integration, maintaining the miniature cartridge design but with improved signal integrity for higher bus speeds. Both MMC variants included power management features tailored for mobile use, such as support for low-power states and a maximum thermal design power (TDP) of around 13.9 W at 300 MHz, though typical consumption was lower at 6–10 W under normal loads to extend battery life.[19] These modules emphasized power efficiency through optimized voltage regulation and reduced leakage, with I/O interfaces operating at 3.3 V to ensure compatibility with laptop chipsets while keeping core power low. Later mobile Pentium II variants based on the Dixon core shifted to a Ball Grid Array (BGA) package, specifically the 615-ball BGA-615 form factor measuring 35 mm x 31 mm, which allowed direct soldering to the motherboard for further size reduction and elimination of the cartridge overhead. This packaging enabled on-die integration of the full-speed L2 cache (256 KB), eliminating the half-speed external cache of earlier modules and contributing to a 15% reduction in power consumption at equivalent frequencies compared to prior designs.[20] For instance, the 366 MHz Dixon BGA model operated at a 1.6 V core voltage with a TDP of 9.5 W and typical usage around 6.6 W, supporting thinner notebook profiles under 0.1 inches high.[20] The BGA also facilitated 3.3 V I/O compatibility with a low-voltage core, enhancing overall system efficiency. Thermal management in these mobile modules was critical due to the confined spaces of portable devices, incorporating a Thermal Transfer Plate (TTP) as an integrated heat spreader on the module's top surface to provide a uniform attach point for cooling solutions like heat pipes or spreader plates. This design maintained case temperatures between 0°C and 100°C while ensuring system ambient operation up to 55°C, with active thermal feedback via SMBus to monitor and mitigate hotspots.[19] Such features addressed heat dissipation challenges in mobile environments, enabling reliable performance without aggressive fans or bulky heatsinks.Processor Variants
Klamath Core
The Klamath core served as the debut implementation of the Pentium II processor targeted at desktop computers, launched by Intel on May 7, 1997, with initial clock speeds of 233 MHz, 266 MHz, and 300 MHz. Built on the P6 microarchitecture, it utilized a 0.35 μm CMOS manufacturing process and integrated 7.5 million transistors across a die size of 195 mm².[21][22][23] One notable challenge with the Klamath core was its elevated thermal profile, reaching a thermal design power (TDP) of approximately 28 W in the 300 MHz model, which demanded substantial heat sinks to maintain operational stability and prevent thermal throttling.[24] Furthermore, the off-die 512 KB L2 cache ran at half the processor's core frequency—such as 150 MHz on the 300 MHz variant—leading to performance bottlenecks in memory-bound workloads like database operations and multimedia rendering.[14][25] Early production of the Klamath core encountered low manufacturing yields, contributing to widespread supply shortages that constrained availability for system builders in the months following its release. The core's model variants consisted of the 233 MHz unit (sSpec SL2QR), 266 MHz unit (sSpec SL2BZ), and 300 MHz unit (sSpec SL2SC), each equipped with 512 KB of L2 cache and housed in the Slot 1 cartridge packaging.[26]Deschutes Core
The Deschutes core was Intel's refined desktop implementation of the Pentium II processor, debuting on January 26, 1998, with initial clock speeds of 333 MHz and extending to 450 MHz by mid-1999. Fabricated on a 0.25 μm CMOS process, it maintained the same 7.5 million transistors as the Klamath core but featured a smaller 106 mm² die size, which contributed to improved manufacturing yields and power efficiency.[27][8] Key enhancements addressed the thermal limitations of the Klamath core, including operation at a reduced core voltage of 2.0 V, which lowered power dissipation and heat output while enabling higher frequencies. Later Deschutes models supported an extended 100 MT/s front-side bus (FSB), up from the 66 MT/s of the initial 333 MHz variant, providing bandwidth improvements for memory and I/O operations. Microcode optimizations also enhanced 16-bit code execution performance, benefiting legacy software compatibility and overall application efficiency.[28][29] Representative models included the 350 MHz variant (S-spec SL2WE), 400 MHz (SL2XF), and 450 MHz (SL2ZJ), all equipped with 512 KB of off-die L2 cache running at half the core clock speed. The Deschutes core was adapted for the Pentium II OverDrive processors, allowing upgrades on older Socket 8 systems.[30]Tonga Core
The Tonga core marked Intel's initial foray into mobile Pentium II processors, released in September 1998 with clock speeds of 233 MHz, 266 MHz, and 300 MHz.[31] These processors were fabricated using a 250 nm manufacturing process, the same as the contemporary desktop Deschutes core, but with specific optimizations for reduced power draw in portable systems.[31] Designed for the MMC-1 packaging format, the Tonga core incorporated 512 KB of off-die L2 cache operating at half the processor's clock speed, alongside a 66 MT/s front-side bus to balance performance and efficiency.[32] Power management was enhanced through multiple low-power states, including Quick Start mode for rapid resumption from idle, enabling a thermal design power of approximately 10 W across the lineup—9.0 W for the 233 MHz model, 10.3 W for the 266 MHz, and 11.6 W for the 300 MHz variant.[32] Early notebook implementations often had battery life limited to 2-3 hours under typical loads.[33] The available models—233 MHz (S-Spec SL2AV), 266 MHz (SL2BB), and 300 MHz (SL2BD)—were primarily aimed at business-oriented laptops, providing a transitional step in mobile computing performance before further refinements in subsequent cores.[31]Dixon Core
The Dixon core represented Intel's advanced mobile implementation of the Pentium II processor, introduced in March 1999 as part of efforts to enhance efficiency in notebook computing. Fabricated on a 0.25 μm process, it incorporated 27.4 million transistors, enabling clock speeds ranging from 266 to 400 MHz. This core shifted from the external cache modules of prior mobile designs, prioritizing integration for reduced power draw and improved thermal management.[34] A primary innovation was the inclusion of 256 KB of on-die L2 cache running at full core speed, eliminating the half-speed limitations of off-die cache in earlier variants like the Tonga core. Paired with Ball Grid Array (BGA) packaging, this allowed for a more compact footprint and support for a 100 MT/s front-side bus, while achieving a thermal design power (TDP) of 6–12 W across models—significantly lower than comparable desktop or prior mobile processors. These features directly tackled power inefficiency issues in mobile environments, yielding superior battery life and sustained performance for applications such as office productivity and light multimedia tasks.[35][20] Key models included the 333 MHz variant (S-spec SL4TC), the 366 MHz (S-spec SL4TD), and the 400 MHz (S-spec SL57A), with core voltages reaching up to 1.4 V to balance speed and efficiency. By integrating MMX instructions for multimedia acceleration, the Dixon core provided a versatile foundation for mobile systems without compromising on the P6 microarchitecture's core capabilities.[36]Pentium II OverDrive
Design and Compatibility
The Pentium II OverDrive processor was released in the second half of August 1998 as an upgrade module specifically designed for Socket 8-based systems, targeting users of Intel's Pentium Pro processors. It utilized the 0.25-micrometer Deschutes core, derived from the standard Pentium II architecture, operating at clock speeds of 300 MHz (with a 60 MHz front-side bus) or 333 MHz (with a 66 MHz front-side bus) and a fixed 5× multiplier to ensure compatibility with legacy motherboards. This core contained approximately 7.5 million transistors, incorporating features such as Dual Independent Bus architecture, Dynamic Execution, and Intel MMX technology for enhanced multimedia processing.[37][38][1] In terms of packaging, the Pentium II OverDrive employed a flip-chip module compatible with the Socket 8 interface, integrating the Deschutes core directly with a 512 KB full-speed L2 cache chip sourced from the Pentium II Xeon design, which ran at the processor's core frequency for improved performance over half-speed alternatives. This configuration included an onboard Voltage Regulator Module (VRM) that converted the Socket 8's standard 3.3 V input to the required 2.0 V core voltage, enabling reliable operation in older systems while providing some tolerance for overclocking scenarios up to 3.3 V. Production was limited, with only the 333 MHz model widely sold under the designation PODP66X333, reflecting Intel's focus on extending the lifespan of high-end workstations and servers rather than mass-market deployment. It was priced at a manufacturer's suggested retail price of $599.[38][39][37] Compatibility was optimized as a drop-in replacement for 150–200 MHz Pentium Pro processors in single- or dual-processor configurations, requiring no modifications to the motherboard beyond ensuring the system supported upgradable Socket 8 slots. It maintained backward compatibility with early Pentium II systems but was constrained by the era's chipsets, such as the Intel 440FX, limiting cacheable RAM to 512 MB despite theoretical support for more in the processor architecture. This design choice prioritized seamless integration into existing enterprise environments, avoiding the need for new hardware investments.[37][38][40]Upgrade Performance
The Pentium II OverDrive processor provided substantial performance improvements when upgrading legacy Socket 8 systems originally equipped with Pentium Pro processors. For instance, installing the 333 MHz variant in a 200 MHz Pentium Pro-based system yielded approximately 47% higher performance in standard business applications under Windows 95, as measured by Intel's internal benchmarks. Similarly, under Windows NT 4.0, the upgrade delivered a 39% performance gain in comparable workloads. These enhancements stemmed primarily from the higher clock speeds—up to 333 MHz—and the integration of a full-speed 512 KB L2 cache, which operated at the processor's core frequency unlike the half-speed cache in many Pentium Pro configurations.[37] In multimedia and data-intensive tasks, the inclusion of Intel MMX technology enabled even more pronounced benefits, with up to an 80% increase in performance for MMX-optimized applications such as video encoding and image processing. This acceleration was particularly valuable in environments lacking native MMX support on the original Pentium Pro, allowing the OverDrive to handle tasks like Adobe Premiere video editing roughly twice as efficiently in some scenarios due to the combined effects of MMX instructions and the unified cache design. Real-world tests confirmed these gains, positioning the upgrade as a cost-effective way to extend the life of older workstations without full system replacement.[37] However, the OverDrive's performance was constrained by the underlying Socket 8 architecture, which lacked support for the 100 MT/s front-side bus speeds available on newer Slot 1 platforms, limiting overall system bandwidth to 66 MT/s maximum. In multi-processor configurations, setups exceeding 333 MHz often encountered stability issues, as the processor was not officially validated for higher frequencies in SMP environments beyond dual-CPU operation. These limitations highlighted the OverDrive's role as a bridge solution rather than a long-term platform.[37] The upgrade proved especially suitable for CAD and engineering workstations, where CPU-bound tasks like 3D modeling in AutoCAD benefited from the integer and MMX boosts without requiring peripheral upgrades. By 2001, however, support for Pentium II OverDrive-equipped systems waned due to motherboard obsolescence and the shift to Pentium III and later architectures, rendering further upgrades impractical as operating systems like Windows XP favored newer hardware.[37]Specifications
Desktop Specifications
The desktop variants of the Pentium II processor utilized the Slot 1 cartridge interface and operated on core voltages between 2.0 V and 2.8 V, depending on the core revision. These processors supported a maximum of 4 GB of system RAM following resolutions to early errata related to physical address extension and chipset compatibility. Starting in late 1998, Intel locked the clock multipliers on Deschutes-based models to restrict unauthorized performance enhancements.[41][40] The Klamath core models ran at clock speeds of 233 MHz, 266 MHz, and 300 MHz with a 66 MT/s front-side bus (FSB), featuring 512 KB of L2 cache running at half the core speed and thermal design power (TDP) ratings of 34.8–43 W. Deschutes core models expanded to speeds from 266 MHz to 450 MHz, supporting either 66 MT/s or 100 MT/s FSB, the same 512 KB half-speed L2 cache, and TDP values ranging from 18.6–27.1 W.[41][28]| Core | Clock Speed (MHz) | FSB (MT/s) | L2 Cache | TDP (W) | Example sSpec Numbers |
|---|---|---|---|---|---|
| Klamath | 233 | 66 | 512 KB (half-speed) | 34.8 | SL25C, SL2ZJ, SL2ZD |
| Klamath | 266 | 66 | 512 KB (half-speed) | 38.2 | SL25D, SL2ZE, SL2ZF |
| Klamath | 300 | 66 | 512 KB (half-speed) | 43 | SL2QC, SL2QG, SL2QH |
| Deschutes | 266 | 66 | 512 KB (half-speed) | 19.5 | SL2RF, SL2RG |
| Deschutes | 300 | 66 | 512 KB (half-speed) | 20.8 | SL2RH, SL2RJ |
| Deschutes | 333 | 66 | 512 KB (half-speed) | 23.7 | SL2SK, SL2SL |
| Deschutes | 350 | 100 | 512 KB (half-speed) | 25.5 | SL357, SL35A |
| Deschutes | 400 | 100 | 512 KB (half-speed) | 26.1 | SL2S7, SL362 |
| Deschutes | 450 | 100 | 512 KB (half-speed) | 27.1 | SL3CA, SL3CB |
Mobile Specifications
The Mobile Pentium II processors were designed for portable computing, featuring the Tonga and Dixon cores optimized for lower power consumption compared to desktop variants. The Tonga core, introduced in 1998, operated at clock speeds of 233 MHz, 266 MHz, and 300 MHz with a 66 MT/s front-side bus (FSB) and 512 KB of off-die L2 cache running at half the processor speed.[34][32] These models had thermal design power (TDP) ratings ranging from 9 W to 11.6 W, enabling battery life estimates of 2–4 hours under typical loads in contemporary laptops.[34][42] Packaging utilized Mini-Cartridge or MMC-1 formats, with core voltages around 1.6 V, and supported up to 1 GB of system RAM without integrated graphics, necessitating discrete GPU solutions for display output.[32] The Dixon core, launched in 1999, advanced mobile efficiency with on-die 256 KB L2 cache operating at full processor speed, available in speeds from 266 MHz to 400 MHz, primarily at 66 MT/s FSB.[34][43] TDP values spanned 10.3 W to 13.1 W, with some low-voltage variants at 1.5 V core supply to further extend battery runtime to 2–4 hours under load, while thermal throttling activated around 90–100 °C via integrated sensors to prevent overheating.[34][20] Packaging shifted to MMC-2, BGA, or micro-PGA for thinner profiles, maintaining 1.5–2.0 V voltage ranges and 1 GB maximum RAM support, with no onboard graphics requiring external accelerators.[32]| Core | Clock Speed (MHz) | FSB (MT/s) | L2 Cache Speed | TDP (W) | Package Type |
|---|---|---|---|---|---|
| Tonga | 233–300 | 66 | Half-speed | 9–11.6 | Mini-Cartridge, MMC-1 |
| Dixon | 266–400 | 66 | Full-speed | 10.3–13.1 | MMC-2, BGA, micro-PGA |