Socket A
View on Wikipedia
| Release date | 2000-06 |
|---|---|
| Type | PGA-ZIF |
| Chip form factors | |
| Contacts | 462 |
| FSB protocol | EV6 |
| FSB frequency | 200 MT/s, 266 MT/s, 333 MT/s, 400 MT/s |
| Voltage range | 1.0–2.05V |
| Processor dimensions | 49.53 mm × 49.53 mm[1] |
| Processors | |
| Predecessor | Slot A |
| Successor | |
| Memory support | DDR |
This article is part of the CPU socket series | |
Socket A (also known as Socket 462) is a zero insertion force pin grid array (PGA) CPU socket used for AMD processors ranging from the Athlon Thunderbird to the Athlon XP/MP 3200+, and AMD budget processors including the Duron and Sempron. Socket A also supports AMD Geode NX embedded processors (derived from the Mobile Athlon XP). It compliments (and later supersedes) the prior Slot A CPU interface used in some Athlon Thunderbird processors. The front-side bus frequencies supported for the AMD Athlon XP and Sempron are 133 MHz, 166 MHz, and 200 MHz. Socket A supports 32-bit CPUs only.
The socket is a zero insertion force pin grid array type with 462 pins, hence the alternative name Socket 462. About nine pins in the socket are blocked to discourage accidental insertion of Socket 370 CPUs on Socket A motherboards.
Socket A was replaced by Socket 754 and Socket 939 during 2003 and 2004 respectively, except for its use with Geode NX processors.
Technical specifications
[edit]- Support of processor clock-speeds between 600 MHz (Duron) to 2333 MHz (Athlon XP 3200+)[2]
- Double data rate 100, 133, 166 and 200 MHz front-side bus on Duron, XP and Sempron processors, based on the Alpha 21264 EV6 bus.

Initially launched with 100 MHz FSB support in the earliest chipsets it evolved stepwise to faster 200 MHz FSB while maintaining pin compatibility throughout its lifetime. However, clock, timing, BIOS and voltage differences restrict compatibility between older chipsets and later processors.[3]
Socket dimensions are 5.59 cm (5.24 cm without lever) × 6.55 cm or 2.2" (2.06" without lever) × 2.58", slightly bigger than Socket 370 sockets.
Heatsink
[edit]Heatsinks were commonly attached directly to the CPU socket, but some motherboards also had 4 holes for fastening bigger heatsinks to the motherboard. Those holes are placed in a rectangle with lateral lengths of 35 mm and 65 mm.
Socket A mechanical load limits
[edit]
AMD recommends that the mass of a Socket A CPU cooler to not exceed 300 grams (10.6 ounces). Heavier coolers may result in damage to the die when the system is not properly handled.
All Socket A processors (Athlon, Sempron, Duron and Geode NX) have the following mechanical maximum load limits[1] which should not be exceeded during heatsink assembly, shipping conditions, or standard use. They came with a warning that load above those limits may crack the processor die and make it unusable. The limits are included in the table below.
| Location | Dynamic | Static |
|---|---|---|
| Die Surface | 445 N (100 lbf) | 133 N (30 lbf) |
| Die Edge | 44 N (10 lbf) | 44 N (10 lbf) |
These load limits are quite small compared to the load limits of Socket 370, Socket 423 and Socket 478 processors. Indeed, they were so small that many users more often than not ended up with cracked processors while trying to remove or attach heatsinks to their fragile processor core. This makes installing non-standard or non-certified heatsink solutions a risky business.[citation needed] OEM aluminium heatsinks typically provided smaller thermal tolerances, so the improper application or absence of a thermal pad or thermal grease or operation in high room temperatures may result in some Socket A CPUs overheating and crashing, permanently damaging them and rendering them unusable.
Chipsets
[edit]AMD offered official chipsets for the Slot A/Socket A CPUs. These are included in the table below.
| Model | Code name | Released | CPU support | FSB/HT (MHz) | Southbridge | Features / Notes |
|---|---|---|---|---|---|---|
| AMD-750 chipset | AMD-751 | August 1999[4] | Athlon, Duron (Slot A, Socket A), Alpha 21264[citation needed] | 50 (100MT/s) | AMD-756, VIA-VT82C686A | AGP 2×, SDRAM Irongate chipset family; early steppings had issues with AGP 2×; drivers often limited support to AGP 1×; later fixed with "super bypass" memory access adjustment.[5] |
| AMD-760 chipset | AMD-761 | Nov 2000 | Athlon, Athlon XP, Duron (Socket A), Alpha 21264[citation needed] | 66 (133MT/s) | AMD-766, VIA-VT82C686B | AGP 4×, DDR SDRAM |
Third-party chipsets include the nForce, nForce2, and a large number of VIA K-series chipsets.
In practice, third-party chipsets were heavily favoured by motherboard manufacturers. Stability problems and compatibility quirks from these chipsets abounded from manufacturers not following chipset designers' guidelines. This caused long-lasting damage to AMD's reputation, despite AMD having nothing to do with the poorly-realised hardware.[citation needed] A similar incident happened with third-party chipsets for Super Socket 7 CPUs, of which AMD tried to remedy it by putting quality assurance measures for the Athlon, which used Slot A/Socket A CPUs. Despite this, however, the aforementioned problems mentioned above still lingered on for quite a while, even for Athlon CPUs.
See also
[edit]References
[edit]- ^ a b "AMD Athlon Processor Model 4 Data Sheet" (PDF). AMD. Archived from the original on April 7, 2005. Retrieved November 8, 2001.
- ^ "CPU Sockets Chart". PC Hardware Links. Retrieved April 4, 2009.
- ^ "CTechnology Evolution". AMD. Retrieved April 4, 2009.
- ^ "AMD-750 Chipset Overview" (PDF). AMD. Retrieved August 1, 2001.
- ^ Orozco, Silvino (December 29, 1999). "AMD's Super Bypass - AMD Improves their 750 Chipset". Tom's Hardware. Retrieved November 11, 2022.
Socket A
View on GrokipediaHistory and Development
Origins and Introduction
Socket A, also known as Socket 462, is a zero insertion force (ZIF) pin grid array (PGA) CPU socket featuring 462 pins, designed specifically for AMD's consumer desktop processors. It was released in June 2000 as the primary interface for AMD's 32-bit x86 processors, succeeding the Slot A design and establishing a more standardized platform for mainstream computing.[6] The socket was introduced alongside the Athlon Thunderbird core, which integrated the processor's L2 cache on-die, enabling a transition from the slot-based cartridge of earlier Athlons to a simpler PGA package for easier installation and handling.[6] This shift marked AMD's strategic move toward PGA for its consumer lineup, supporting the Athlon and Duron processor families while reducing manufacturing costs associated with slot designs and improving overall production scalability.[6][7] Key design objectives included compatibility with emerging memory architectures, such as DDR SDRAM through dedicated chipsets like the AMD 760, which provided up to 100% greater memory bandwidth compared to SDRAM contemporaries. This forward-looking approach positioned Socket A as a versatile foundation for AMD's desktop ecosystem until the eventual transition to sockets like Socket 754 in 2003.[1]Evolution from Slot A
Slot A, introduced by AMD in June 1999, served as the initial interface for the Athlon processor family, employing a 242-lead single-edge connector (SEC) cartridge design that encased the CPU and external L2 cache components.[8] This slot-based architecture, inspired by Intel's Slot 1 but rotated 180 degrees for incompatibility, facilitated higher power delivery and larger cache sizes compared to contemporary pin-grid array sockets, enabling the Athlon to compete effectively in performance against Intel's Pentium III.[9] However, the cartridge assembly proved cumbersome, contributing to elevated manufacturing expenses due to the additional materials and assembly steps required for the plastic housing and integrated components.[10] The transition to Socket A was driven by several practical limitations of Slot A, including its high production costs from the cartridge construction, challenges in motherboard integration due to the larger slot mechanism, and the lack of a zero insertion force (ZIF) lever, which complicated processor installation and upgrades compared to simpler socket-based systems.[11] By integrating the L2 cache directly onto the CPU die, AMD eliminated the need for the external cartridge, reducing costs and simplifying the overall package while maintaining compatibility with the EV6 bus architecture.[12] This shift also addressed scalability issues as clock speeds rose, making the bulky Slot A less viable for mass production and easier overclocking through unlocked multipliers on the new socket.[9] Slot A was phased out by mid-2000, with Socket A (also known as Socket 462) becoming the standard interface starting with the Athlon Thunderbird revision, launched on June 5, 2000.[8] This timeline aligned with AMD's strategic positioning against Intel's Socket 370 platform for Pentium III and Celeron processors, emphasizing cost efficiency and upgrade simplicity to attract original equipment manufacturers (OEMs) and enhance market penetration in the consumer and enthusiast segments.[13] The first motherboards supporting Socket A debuted in mid-2000, such as the ASUS A7V with the VIA KT133 chipset for SDRAM support; the AMD-760 chipset, introducing DDR SDRAM support, arrived in November 2000 and further bolstered AMD's competitive edge in memory bandwidth.[14][15] (wait, non-wiki; use [16])Technical Specifications
Electrical Characteristics
Socket A employs a front-side bus (FSB) based on the double data rate (DDR) EV6 protocol, operating at speeds ranging from 200 to 400 MT/s (mega-transfers per second), which corresponds to clock rates of 100 to 200 MHz with data transferred on both rising and falling edges.[17] This design enables efficient bandwidth of up to 3.2 GB/s on the highest-speed configurations, while CPU clock generation relies on internal multipliers applied to the FSB frequency, allowing processors to run at multiples independent of the bus rate.[17] The EV6 protocol is source-synchronous and packet-based, facilitating point-to-point communication between the CPU and chipset with strobe signals for timing alignment.[17] The socket supports a core voltage (VCC_CORE) range of 1.1 V to 1.85 V, adjustable via the Voltage Identification (VID) pins to match specific processor requirements, with nominal values around 1.65 V for many models.[18] Additional rails include the analog supply (VCCA) at 2.25 V to 2.75 V (nominal 2.5 V) for phase-locked loops (PLLs), drawing up to 50 mA per GHz of operation, and a 2.5 V supply for DDR SDRAM memory interfaces on the motherboard.[18] Input/output (I/O) signaling on the FSB operates at core voltage levels, with reference voltage (VREF) set to 50% of VCC_CORE (±50 mV) to support low-voltage differential signaling.[17] Power delivery through Socket A accommodates current draws up to approximately 46.5 A for high-end processors in active states, enabling clock speeds from 600 MHz in initial Duron models to over 2200 MHz in Athlon XP configurations like the 3200+ model.[17] Thermal design power reaches up to 76.8 W under maximum load, with multiple distributed VCC_CORE pins ensuring stable delivery and reduced inductance.[17] The asynchronous nature of the FSB permits flexible CPU-to-memory clock ratios, such as 6:4 or 10:5, which enhances compatibility across components but necessitates chipset firmware support for stable operation and has historically enabled user overclocking by adjusting multipliers or bus ratios.[18] Signaling utilizes 462 pins, with allocations including approximately 40 for address lines (split into 13-bit input and output channels like SADDIN[14:2]# and SADDOUT[14:2]# for packet-based transfers), 64 bidirectional data lines (SDATA[63:0]#), plus 8 for error-correcting code (ECC), and various control lines for commands, strobes, and resets.[17] The interface employs AGTL+ (Advanced Gunning Transceiver Logic Plus) logic levels, featuring low-voltage swing push-pull drivers with impedance matching via on-die termination (ODT) to minimize noise and reflections on the FSB traces.[17] This signaling reduces power consumption compared to earlier standards and supports the high-speed transfers essential for performance.[17]Physical Dimensions
Socket A employs a pin grid array (PGA) zero insertion force (ZIF) design with 462 pin positions arranged in a staggered grid pattern, of which 453 are active pins, forming a roughly square array. The pins feature a standard pitch of 1.27 mm, enabling compact integration on motherboards while accommodating the high pin density required for Athlon processors.[17] The socket body measures 5.59 cm in length (5.24 cm excluding the lever) by 6.55 cm in width, providing a footprint compatible with standard ATX motherboard layouts. The associated CPU package is an organic PGA (OPGA) with square dimensions of approximately 49.5 mm by 49.5 mm and a mass of 11.0 g, ensuring precise alignment within the socket..html)[17] Insertion and securing of the processor are facilitated by a ZIF cam lever mechanism located on one side of the socket; the lever is raised to approximately 90-100 degrees to open the contacts with zero force, allowing the processor to drop into place, after which lowering the lever clamps the pins. Alignment keys on both the socket and processor package notches prevent misorientation and ensure correct seating.[19] The socket's pins are made of phosphor bronze, a resilient alloy chosen for its spring-like properties and fatigue resistance, with a gold plating over nickel underlayer for superior electrical conductivity, low contact resistance, and corrosion protection. This construction supports 30-50 insertion/removal cycles while maintaining reliable connections.[20] Unlike its predecessor Slot A, which used a slot-based cartridge interface for processor modules, Socket A adopts a land-side pin configuration and distinct grid layout, rendering it physically and electrically incompatible and requiring dedicated motherboards.[21]Compatible Components
Supported Processors
Socket A, also known as Socket 462, supported a range of AMD K7-based processors primarily designed for desktop and entry-level server applications, spanning from 2000 to 2004. These processors utilized a 462-pin PGA package and operated on front-side bus (FSB) frequencies of 100 MHz, 133 MHz, 166 MHz, or 200 MHz, with clock multipliers that allowed for scalable performance within the socket's electrical limits. The Athlon series formed the flagship lineup for Socket A, beginning with the Thunderbird core introduced in 2000. Manufactured on a 0.18 μm process, Thunderbird processors featured clock speeds from 600 MHz to 1400 MHz and integrated 256 KB of L2 cache directly on the die, marking a shift from earlier off-die cache designs for improved efficiency.[22] This was followed by the Palomino core in 2001, also on 0.18 μm, which improved power efficiency through architectural enhancements and the 0.18 μm process, enabling higher clock speeds with reduced heat output compared to Thunderbird, and supported speeds up to 1733 MHz while retaining 256 KB L2 cache. In 2002, the Thoroughbred core transitioned to a 0.13 μm process, enabling higher densities and speeds from 1450 MHz to 2250 MHz with 256 KB L2 cache, contributing to better thermal performance and overclocking potential.[22] The Barton core, launched in 2003 on the same 0.13 μm node, doubled the L2 cache to 512 KB and reached speeds up to 2333 MHz, providing a significant boost in application performance for gaming and productivity tasks. Additionally, the Athlon MP variant, optimized for multi-processor configurations in entry-level servers, used these same cores but included enhancements for symmetric multiprocessing support.[1] For budget-oriented systems, the Duron series offered cost-effective alternatives to the Athlon, sharing the same Socket A interface but with reduced cache. The initial Spitfire core, released in 2000 on 0.18 μm, operated at 600 MHz to 950 MHz with 64 KB L2 cache, targeting value desktops.[23] The Morgan core in 2001, still on 0.18 μm, extended speeds to 950 MHz through 1300 MHz while maintaining the 64 KB L2 cache for basic computing needs.[23] By 2002, the Applebred core adopted the 0.13 μm process from Thoroughbred, reaching up to 1800 MHz but continuing with 64 KB L2 cache to keep pricing low.[24] In 2004, AMD extended Socket A support with the Sempron series, rebranding select Duron and Athlon XP models for continued market presence amid the shift to newer sockets. These processors, based on Thoroughbred or Barton cores at 0.13 μm, featured 256 KB or 512 KB L2 cache and clock speeds up to 2000 MHz, often with a 333 MHz FSB to maximize compatibility with existing motherboards.[25] For embedded and low-power applications, the Geode NX series, introduced in 2003, provided specialized Socket A processors derived from the Thoroughbred architecture. These chips emphasized reduced power consumption for industrial and non-desktop uses, with clock speeds from 667 MHz to 1800 MHz, 256 KB L2 cache, and optimizations for fanless operation in compact systems.[26]Compatible Chipsets
Socket A motherboards primarily utilized chipsets from AMD and third-party manufacturers such as NVIDIA and VIA Technologies, enabling support for Athlon and Duron processor families through various front-side bus speeds.[27] These chipsets evolved from SDRAM-based designs to DDR SDRAM support, incorporating features like accelerated graphics ports and peripheral connectivity to meet the demands of early 2000s computing. AMD's initial chipset for the platform, the AMD-750 released in 1999, provided foundational support through adapters for transitioning from Slot A to Socket A configurations, featuring the AMD-751 system controller and AMD-756 peripheral bus controller with PC-100 SDRAM compatibility up to 768 MB and AGP 2.0 at 1x/2x modes.[28][29] Following this, the AMD-760 chipset launched in 2000 offered native DDR SDRAM support at 200/266 MHz speeds, AGP 2x interface, and USB 1.1 ports via its AMD-761 northbridge and AMD-766 southbridge (or compatible VIA VT82C686B alternatives).[30] NVIDIA entered the market with the nForce chipset in 2001, distinguished by its integrated GeForce2 GPU for graphics processing, 5.1-channel audio capabilities, and dual-channel DDR SDRAM support reaching up to 400 MT/s in later iterations.[31] The successor nForce2, introduced in 2002, provided peak compatibility for Athlon XP processors and included PCI/AGP bus locking features that facilitated stable overclocking by preventing frequency scaling issues.[32][33] VIA Technologies dominated third-party development after the AMD-760, driven by licensing arrangements that allowed broader ecosystem expansion, resulting in over 100 motherboard models across variants.[34] The KT133 chipset from 2000 supported only SDRAM at PC100/133 speeds, while the KT133A update in 2001 added DDR SDRAM compatibility.[27] Subsequent releases included the KT266 in 2001 for enhanced front-side bus handling up to 266 MHz and the KT400 in 2002, which extended late-stage Socket A viability with DDR333 support and AGP 8x, though it transitioned toward Athlon 64 platforms.[35] Overall, more than 20 chipset variants emerged between 2000 and 2004 from these and other vendors like SiS and ALi, reflecting the platform's maturity before Socket A's phase-out.[34]| Manufacturer | Chipset | Release Year | Key Features |
|---|---|---|---|
| AMD | AMD-750 | 1999 | PC-100 SDRAM (up to 768 MB), AGP 2x, adapter-based Socket A support[29] |
| AMD | AMD-760 | 2000 | DDR SDRAM (200/266 MHz), AGP 2x, USB 1.1[30] |
| NVIDIA | nForce | 2001 | Integrated GeForce2 GPU, 5.1 audio, dual-channel DDR (up to 266 MT/s initially)[31] |
| NVIDIA | nForce2 | 2002 | Dual-channel DDR up to 400 MT/s, PCI/AGP locking for overclocking[32] |
| VIA | KT133 | 2000 | SDRAM (PC100/133), basic Socket A integration[27] |
| VIA | KT133A | 2001 | DDR SDRAM support added[27] |
| VIA | KT266 | 2001 | Enhanced FSB (up to 266 MHz), DDR SDRAM |
| VIA | KT400 | 2002 | DDR333, AGP 8x, late Socket A extension[35] |