CPU socket
A CPU socket is the specialized connector on a motherboard that establishes mechanical and electrical connections between a central processing unit (CPU) and the system's other components, enabling the transfer of power, data, and control signals.[1][2] It serves as a removable interface, distinguishing socketed CPUs from soldered ones like those using Ball Grid Array (BGA) packaging, which cannot be easily upgraded.[3] The primary purpose of a CPU socket is to facilitate straightforward installation, removal, and replacement of processors, supporting hardware upgrades without requiring full motherboard replacement and ensuring compatibility between specific CPU models and chipsets.[2][4] This design enhances system flexibility, repairability, and longevity, particularly in desktop and server environments where performance demands evolve rapidly. Sockets typically incorporate mechanisms like levers or clips to secure the CPU and apply even pressure for reliable contact, preventing damage from misalignment or overheating.[5] CPU sockets vary by architecture and manufacturer, with two predominant designs: Pin Grid Array (PGA), where pins protrude from the CPU to insert into the socket, and Land Grid Array (LGA), where flat contacts on the CPU mate with pins in the socket itself.[6][7] Intel has favored LGA since introducing Socket 775 (LGA775) in 2004 for Pentium 4 processors, with subsequent examples including LGA1151 for 6th-9th Gen Core CPUs, LGA1700 for 12th-14th Gen Alder Lake and Raptor Lake series (as of 2023), and LGA1851 for Core Ultra 200 series (Arrow Lake) as of 2024.[8][9][10] In contrast, AMD predominantly uses PGA in sockets like AM4 for Ryzen 1000-5000 series and has transitioned to LGA with AM5 (LGA1718) for Ryzen 7000, 8000, and 9000 series processors, with support promised through at least 2027.[11][12] These socket types dictate CPU-motherboard compatibility, influencing factors such as pin count, voltage regulation, and thermal design power (TDP) handling.[13]Fundamentals
Definition and Purpose
A CPU socket is a mechanical and electrical connector designed to securely house a central processing unit (CPU) on a motherboard, enabling removable installation in contrast to permanently soldered chips. This interface consists of an array of pins or contacts that align with the CPU's package, ensuring precise physical and electrical connectivity between the processor and the system's circuitry.[14] By allowing the CPU to be inserted and removed without specialized equipment, sockets promote modularity in desktop and server computing environments.[4] The primary purposes of a CPU socket include facilitating straightforward replacement and upgrading of processors to improve performance or extend hardware lifespan without replacing the entire motherboard. It serves as the conduit for essential electrical pathways, delivering power, transmitting data, and relaying control signals between the CPU and other motherboard components.[15] Additionally, sockets support effective thermal management by providing a stable mounting point for heat sinks and cooling solutions, which dissipate the heat generated by the CPU during operation.[4] Key benefits of CPU sockets lie in their enhancement of system longevity, particularly through designs that enable backward and forward compatibility across multiple CPU generations within the same socket type. This upgradability contrasts sharply with ball grid array (BGA) packaging commonly used in laptops and mobile devices, where the CPU is soldered directly to the board, making replacements difficult and often requiring full system disposal.[4] Sockets thus prioritize serviceability and cost-efficiency in environments where hardware evolution demands flexibility.[14] CPU sockets emerged in the 1970s alongside the advent of socketed integrated circuits, replacing direct soldering to improve reliability, ease of maintenance, and adaptability in early microprocessor-based systems.[16]Design Principles
CPU sockets are engineered with a robust mechanical structure to facilitate secure and repeatable connections between the processor and motherboard. The core components include a socket body, typically molded from high-strength plastic or composite materials, which encases an array of electrical contacts—either protruding pins in pin grid array (PGA) configurations or flat pads in land grid array (LGA) designs. These contacts are arranged in a precise grid to interface with corresponding features on the processor package. A retention mechanism, often a lever or cam-operated clip, applies uniform pressure to maintain contact integrity once the processor is seated, preventing dislodgement during operation. Alignment keys, such as molded notches or protrusions on the socket body, guide the processor into position and prevent incorrect orientation, ensuring compatibility and reducing the risk of damage during installation.[17][18] Electrically, the socket's contacts form dense arrays optimized for signal integrity, supporting high-speed data transfer rates essential for modern processors. These arrays include dedicated pins or pads for power delivery, such as Vcc rails for core voltage supply and Vtt rails for termination voltage to stabilize signaling on buses like the front-side bus (FSB). Impedance matching is a critical principle, where contact geometry and surrounding ground planes are designed to maintain characteristic impedance (typically 50–100 ohms) along signal paths, minimizing reflections, crosstalk, and electromagnetic interference that could degrade performance. This ensures reliable transmission of clock signals, address lines, and data buses, with multiple ground and power planes interleaved to provide shielding and low-inductance paths.[19][20][21] To handle operational stresses, CPU sockets are built to endure thermal and mechanical loads while integrating with cooling solutions. The design accommodates insertion and removal cycles ranging from 10 to 50, depending on the specification, with materials selected for fatigue resistance to avoid deformation or contact failure over time. Thermal considerations involve compatibility with integrated heat spreaders (IHS) on the processor, where the socket's retention mechanism applies even pressure to promote efficient heat transfer to attached coolers, often via thermal interface materials. Mechanical load limits, such as torque on the retention lever, are strictly defined to prevent warping of the socket body or underlying printed circuit board (PCB).[19][18][22] Keying and anti-misinsertion features further enhance reliability by enforcing proper orientation. Asymmetrical shapes, including cutouts or tabs on the socket perimeter, mate only with matching features on the processor package, such as corner notches marked by indicators like a gold triangle. Zero insertion force (ZIF) mechanisms, common in contemporary designs, employ a lever that lifts contacts during insertion, eliminating the need for force that could bend pins or pads, thereby extending socket lifespan and simplifying user handling. These elements collectively support modularity by allowing processor upgrades without soldering, while safeguarding against errors that could lead to electrical shorts or mechanical damage.[17][23]Historical Evolution
Early Developments (1970s–1990s)
The origins of CPU sockets trace back to the 1970s, coinciding with the advent of commercial microprocessors. Early designs utilized Dual In-line Package (DIP) sockets to accommodate processors like the Intel 4004, introduced in 1971, which featured a 16-pin ceramic DIP configuration often soldered directly onto boards for production systems but inspiring socketed variants for prototyping and maintenance.[24] These sockets provided removable electrical connections via two parallel rows of pins, facilitating easier replacement compared to permanent soldering and supporting basic signal integrity through simple pin-to-hole mating. For industrial and military applications requiring greater durability against temperature extremes and vibration, Ceramic Dual In-line Package (Cer-DIP) sockets emerged as a reliable alternative, offering superior hermetic sealing and thermal conductivity for early microprocessors such as the Intel 8080.[25] This period's sockets prioritized low pin counts—typically 16 to 40 pins—to handle fundamental power, ground, and address lines, reflecting the limited complexity of 4-bit and 8-bit architectures. In the 1980s, CPU sockets evolved to support more advanced 16-bit processors, with the introduction of Pin Grid Array (PGA) designs enabling denser pin arrangements beneath the chip for improved signal routing. The Intel 80386 microprocessor, introduced in 1985, utilized PGA packaging with sockets supporting up to 168 pins to enable 32-bit addressing for up to 4 GB of memory and stable power distribution at clock speeds up to 40 MHz.[26] PGA sockets marked a departure from edge-based DIP layouts, arranging pins in a grid pattern on the underside of the package to accommodate the growing number of I/O signals without increasing the overall footprint. Pin counts remained modest, ranging from around 132 pins for early 386 designs to 168 pins for later 32-bit designs like the 80486, sufficient for basic bus operations and power needs in desktop systems.[27] The 1990s brought expanded functionality to CPU sockets amid the transition to higher-performance 32-bit and early 64-bit processors. Socket 3, introduced by Intel in 1993 for the 80486 and compatible with initial Pentium models, incorporated voltage detection pins for automatic switching between 5 V and 3.3 V supplies, reducing power consumption and heat generation in evolving motherboard designs.[28] This era also saw the widespread adoption of Zero Insertion Force (ZIF) mechanisms in PGA sockets, where a lever-activated clamp gently presses pins into contacts without manual force, minimizing damage during installation—a feature first popularized with 386 processors and standardized for 486 systems.[29] AMD contributed to this landscape with its Am486 series, which replicated Intel's 80486 pinouts and socket compatibility to allow drop-in replacements on existing motherboards, fostering competition and upgrade options.[30] These advancements were driven by practical necessities in microprocessor evolution, particularly the shift away from soldered chips prevalent in 8086-based systems of the late 1970s and early 1980s. Soldered installations exacerbated heat management challenges as clock speeds rose from 5 MHz to 10 MHz, often leading to thermal failures without easy replacement; sockets enabled the attachment of dedicated heatsinks and facilitated field upgrades.[9] This transition was further propelled by Moore's Law-driven reductions in transistor sizes—from micrometer-scale in the 8086 to sub-micron processes by the 1990s—allowing higher densities and speeds that demanded modular, non-permanent mounting for reliability and serviceability.[31]Modern Transitions (2000s–2010s)
In the early 2000s, CPU sockets transitioned to support higher clock speeds and power demands of single- and dual-core processors, while addressing thermal challenges. Intel's Socket 478, launched in 2001 for the Pentium 4 series, utilized a 478-pin micro Pin Grid Array (mPGA) design with zero insertion force (ZIF) mechanism to facilitate easier installation and accommodate the NetBurst architecture's front-side bus (FSB) at speeds up to 800 MHz.[32] Concurrently, AMD's Socket A (Socket 462), introduced in 1999 and spanning into the 2000s, employed a 462-pin PGA configuration for Athlon processors, enabling FSB rates from 100 MHz to 400 MHz and supporting Athlon XP models up to 2.2 GHz. A pivotal shift came in 2004 with Intel's introduction of Socket T (LGA 775), the first major Land Grid Array (LGA) socket, which relocated pins to the motherboard for enhanced durability and improved heat dissipation via direct contact with an integrated heat spreader on the CPU package. By the 2010s, socket designs incorporated integrated components to meet the multi-core era's needs for on-package graphics and memory handling. Intel's Socket 1156 (2009) with 1,156 pins supported the Nehalem and Westmere microarchitectures, including initial integrated graphics in Clarkdale processors, while its successor Socket 1155 (2011) with 1,155 pins supported the Sandy Bridge microarchitecture with built-in Intel HD Graphics 2000/3000 for mainstream desktops, alongside DDR3 memory up to 32 GB. AMD countered with Socket FM1 in 2011 for Llano APUs and Socket FM2 from 2012 to 2013 for Trinity and Richland APUs, both featuring 906-pin PGA layouts to integrate CPU cores with Radeon graphics, supporting DDR3 memory and fused processor-graphics designs for accelerated processing units (APUs). Pin counts surged beyond 900 in high-end sockets like LGA 1366 (2008) and LGA 2011 (2011), enabling direct CPU integration of memory controllers for DDR3 and early DDR4 support, which reduced latency compared to chipset-based designs. These transitions were driven by escalating performance requirements, including thermal design power (TDP) ratings exceeding 100 W in the Core 2 era—for instance, the Core 2 Extreme QX9770 at 136 W—necessitating robust cooling and power delivery. The adoption of serial interfaces like PCI Express (PCIe), standardized in 2003 and integrated into chipsets by the mid-2000s, allowed sockets to route high-bandwidth lanes directly for peripherals such as GPUs, replacing parallel AGP and PCI buses. Additionally, voltage regulation modules (VRMs) became standard on motherboards, providing multi-phase buck converters to step down 12 V supplies to the CPU's low voltages (e.g., 0.8–1.4 V) while handling currents over 100 A for stable operation under load.[33] A landmark event was Intel's Nehalem microarchitecture in 2008, which integrated the memory controller onto the CPU die in Socket B (LGA 1366), eliminating FSB bottlenecks for DDR3 access and enabling three-channel memory configurations up to 24 GB, thus reducing socket dependencies on external northbridge components. This on-die integration marked a broader industry shift toward consolidated die features, influencing subsequent socket evolutions.Contemporary Innovations (2020s)
In the early 2020s, Intel introduced the LGA 1700 socket in 2021 to support its 12th-generation Alder Lake processors, which pioneered a hybrid architecture combining performance-oriented P-cores and efficiency-focused E-cores for improved multitasking and power efficiency.[34] This socket, featuring 1,700 pins and enhanced power delivery capabilities, also accommodated subsequent Raptor Lake processors up to the 14th generation in 2023, enabling thermal design power (TDP) ratings exceeding 200W while maintaining compatibility with DDR4 and DDR5 memory configurations.[34] By 2024, Intel transitioned to the LGA 1851 socket for Arrow Lake processors under the Core Ultra Series 2 lineup, increasing the pin count to 1,851—a 9% rise over LGA 1700—to bolster I/O bandwidth, including expanded PCIe 5.0 lanes and exclusive DDR5 support for higher data throughput in demanding applications.[35][36] AMD countered with the AM5 socket in 2022, designed for its Zen 4-based Ryzen 7000 series processors and extending to Zen 5-based Ryzen 9000 series in 2024, emphasizing long-term platform stability with official support promised through at least 2025 and beyond, potentially reaching 2027 with Zen 6 compatibility.[37][38] This LGA 1718 socket natively supports DDR5 memory and PCIe 5.0, facilitating upgrades across multiple generations without requiring a full platform overhaul, and aligns with AMD's chiplet-based designs that modularize CPU components for scalable performance.[11] The shift from the AM4 socket, which spanned 2016 to 2022, to AM5 marked a deliberate move toward future-proofing amid evolving chiplet architectures that enhance core counts and efficiency for AI and high-performance computing workloads.[39] Key innovations in 2020s CPU sockets revolve around higher pin densities to accommodate AI accelerators and hybrid core configurations, as seen in LGA 1851's expanded contacts that support advanced interconnects for heterogeneous computing.[35] Improved power delivery systems, including more robust voltage regulation and phase arrays, enable sustained operation at 200W+ TDP levels, addressing the thermal and electrical demands of dense core arrangements in both Intel's P/E-core hybrids and AMD's multi-chiplet Zen setups.[34] Sustainability efforts have also emerged, with manufacturers incorporating recyclable materials like post-consumer recycled plastics and aluminum into socket assemblies and surrounding motherboard components to reduce e-waste and carbon footprints in high-volume production.[40] Notable events include Intel's adoption of Foveros 3D stacking technology, which vertically integrates dies to boost density but imposes stricter thermal management on sockets, necessitating advanced cooling interfaces for heat dissipation in stacked hybrid cores.[41] Meanwhile, AMD's AM4-to-AM5 transition in 2022 underscored a strategic pivot to chiplet modularity, allowing sockets to handle disaggregated CPU designs that prioritize higher core counts and AI-optimized interconnects without proportional increases in physical size.[39]Classifications and Types
Contact-Based Categories
CPU sockets are primarily categorized based on their physical contact mechanisms, which determine how electrical connections are established between the processor and the motherboard. The two dominant types are Pin Grid Array (PGA) and Land Grid Array (LGA), each featuring distinct designs for pin placement and contact formation. These categories influence aspects such as installation ease, durability, and electrical performance, with PGA placing pins on the processor and LGA positioning them on the socket itself.[42][43] In PGA sockets, the processor underside contains an array of protruding pins that insert directly into corresponding holes in the socket on the motherboard. This design, commonly used by AMD in sockets like AM4, which uses PGA for Ryzen 1000-5000 series processors, provides robust mechanical alignment and supports high pin counts for modern multi-core processors. The pins establish electrical contact upon insertion, secured by a lever mechanism that applies downward pressure. Advantages include greater durability on the processor side, as damage to the socket does not affect the CPU pins, facilitating easier motherboard socket repairs. However, PGA requires higher insertion force, which can lead to pin bending during handling or upgrades if not done carefully.[44][45][7] Conversely, LGA sockets feature flat contact pads, or lands, on the processor's underside that press against spring-loaded pins within the socket. Intel predominantly employs this approach in designs such as LGA 1700, where the socket's pins provide resilient contact under load from a retention mechanism. This configuration excels in heat dissipation, as the flat CPU surface allows better thermal interface material application, and it simplifies processor upgrades since socket failures do not risk damaging the CPU. LGA also accommodates denser pin arrangements in a compact form factor, beneficial for high-performance applications. Drawbacks include potential wear on the socket pins over multiple installations and higher manufacturing complexity for the socket assembly.[19][34][46] Other contact-based variants include early PGA derivatives like Socket 5 and Socket 7, which used staggered pin arrays for Pentium-era processors and offered backward compatibility through varied pin counts. In embedded systems, Ball Grid Array (BGA) interfaces are occasionally socketed, though typically soldered directly to the board for permanence; these use solder balls for contact, prioritizing space efficiency over upgradability. Comparisons reveal that PGA and LGA exhibit similar electrical performance up to high frequencies (around 8 GHz), with PGA sometimes showing superior signal integrity in certain modeling scenarios due to its rigid pin structure. Trade-offs generally favor PGA for manufacturing simplicity and cost in consumer applications, while LGA supports more frequent socket replacements and enhanced thermal management in upgrade-oriented systems.[47][48]Form Factor Variations
CPU socket form factors vary significantly to balance electrical performance, mechanical stability, thermal management, and compatibility with different system architectures, such as desktop, server, and mobile platforms. These variations primarily involve differences in physical dimensions, pin arrangements, and overall package geometry, which directly influence the socket's ability to handle increasing power delivery and data throughput requirements over time.[19] Pin counts in CPU sockets have evolved from around 168 pins in early designs, such as those for the Intel 486 processors in the late 1980s, to over 4,000 pins in contemporary high-end configurations. For instance, Intel's LGA 4677 socket, used for Xeon Scalable processors like Sapphire Rapids, features 4,677 pins to support extensive I/O bandwidth, including up to 80 PCIe 5.0 lanes, USB ports, and memory channels. AMD's Socket SP5 features 6,096 pins for EPYC 9004 series processors (introduced 2022), while Intel's LGA 7529 uses 7,529 pins for Xeon 6 E-core series (released 2024), supporting extensive PCIe 5.0/CXL interfaces. This increase in pin density enables higher data transfer rates; older sockets like LGA 1156, with 1,156 pins, typically supported only 16 PCIe 2.0 lanes, limiting overall system bandwidth compared to modern variants. The pin count directly impacts the socket's capacity for power distribution and signal integrity, as more pins allow for finer-grained allocation to I/O interfaces like PCIe and USB.[49][50][19][51][52] Form factors differ across use cases to optimize space, cooling, and multi-processor support. Desktop sockets, such as Intel's LGA 1200, employ a compact square layout measuring approximately 37.5 mm by 37.5 mm to fit standard ATX motherboards while accommodating consumer-grade processors. In contrast, server-oriented sockets like AMD's Socket SP3 adopt a larger rectangular form factor (75 mm by 82.5 mm) with 4,094 pins, facilitating multi-socket configurations for enterprise workloads requiring scalable compute resources. Mobile variants often use flip-chip ball grid array (FCBGA) packaging, such as Intel's micro-FCBGA, which is soldered directly to the board rather than socketed, reducing size and height for laptops while maintaining high pin densities up to 2,000 or more for integrated I/O. These form factor choices ensure compatibility with specific chassis designs and thermal envelopes.[19][53][49] Socket layouts are engineered with strategic pin placement to minimize electrical noise and optimize signal propagation. Power and ground pins are typically clustered centrally or distributed evenly around the perimeter, surrounded by signal traces dedicated to data, address, and control lines, which helps reduce voltage droop under high loads. Keying mechanisms, such as notches or tabs at the corners (e.g., one or two missing pins or alignment grooves), prevent incorrect insertion and ensure proper orientation during installation. These layout decisions significantly affect motherboard PCB routing, as higher pin densities demand multi-layer boards with precise trace widths and via placements to avoid crosstalk and maintain signal integrity.[54][19] Standardization efforts, guided by organizations like JEDEC, establish consistent spacing parameters to promote interoperability and manufacturing efficiency. Common pin pitches range from 0.5 mm in dense mobile packages to 1.27 mm in desktop and server sockets, with Intel's LGA designs often using 1.016 mm (0.040 inches) for balanced density and reliability. This evolution in pitch and layout supports expanded I/O capabilities, such as increasing PCIe lanes from 16 in early 2010s sockets to 28 or more in recent server variants, enabling broader connectivity for peripherals and accelerators.[55][19]Sockets by Instruction Set Architecture
x86 Family
The x86 family of CPU sockets, predominantly developed by Intel and AMD, forms the backbone of personal computing, server, and workstation platforms, enabling compatibility between processors and motherboards across multiple generations. These sockets adhere to either Pin Grid Array (PGA) or Land Grid Array (LGA) designs, with Intel favoring LGA for its consumer and server lines since the mid-2000s, while AMD has historically used PGA but introduced LGA in select high-end variants. Key evolutions in x86 sockets reflect advancements in core counts, power delivery, and memory support, balancing backward compatibility with performance gains in microarchitectures like Intel's Core series and AMD's Ryzen lineage. Intel's x86 sockets have undergone frequent iterations to support evolving architectures, starting with early examples like Socket 8 in 1995, which accommodated the Pentium Pro processor with 240 pins in a PGA configuration, marking an early shift toward slot-based designs before full socket adoption. This was followed by LGA 775, introduced in 2004 for the Pentium 4 and later Core 2 processors, featuring 775 contacts and supporting up to 8-core configurations until its phase-out in 2010, with DDR2 and early DDR3 memory compatibility. The LGA 115x series (LGA1150 and LGA1151) represented a period of standardization for consumer platforms, with LGA1150 (2013–2015 for Haswell/Broadwell, 1150 pins) and LGA1151 (2015–2018 for Skylake/Coffee Lake, 1151 pins). Subsequent sockets included LGA1200 (2020–2021 for Comet Lake/Rocket Lake, 1200 pins) and LGA1700 (2021–2024 for Alder Lake/Raptor Lake, 1700 pins), enabling integrated graphics in consumer chips and PCIe 4.0/5.0 support. More recent consumer sockets include LGA 1851, launched in 2024 for Arrow Lake processors, with 1851 contacts to handle increased power demands up to 250W TDP and DDR5 memory. For servers, variants like LGA 3647 debuted in 2017 with Xeon Skylake-SP processors (2017–2023, 3647 contacts), offering dual-socket scalability and up to 28 cores per CPU; this was succeeded by LGA 4677 in 2023 for Sapphire Rapids processors (4677 contacts, up to 60 cores, DDR5 support).[56] AMD's x86 sockets emphasize longevity and upgrade paths, beginning with Socket A (2003–2004 for Athlon XP), evolving to AM2 (2006–2008 for Athlon 64 X2 with DDR2), and AM3 (2009–2011 for Phenom II with DDR3 and HyperTransport 3.0), all using PGA designs with pin counts around 940–962. The FM2 socket, released in 2012 for Trinity APU processors, introduced 906 PGA pins and partial GPU integration but had a short lifespan. A major milestone was AM4 in 2016, a PGA socket with 1331 pins supporting Ryzen 1000 through 5000 series processors from 2016 to 2022, spanning Zen 1 to Zen 3 architectures, DDR4 memory, and up to 16 cores, with an exceptional seven-year compatibility window that allowed multi-generational upgrades without motherboard changes. This was succeeded by AM5 in late 2022, a 1718-pin LGA socket for Zen 4 and Zen 5 processors, introducing exclusive DDR5 support, PCIe 5.0, and integrated graphics in select models, with commitments for support through at least 2027.[57] High-end Threadripper platforms use dedicated sockets, such as sTRX4 (2019–2022 for Zen 2 Threadripper 3000 series, 4094-pin sPGA for up to 64 cores and quad-channel DDR4), sWRX8 (2020–2022 for Threadripper PRO 3000/5000 series, 4094-pin sPGA for up to 64 cores and quad-channel DDR4), and sTR5 (2023+ for Zen 4/5 Threadripper 7000/9000 series, 6096-pin LGA for up to 96 cores and 8-channel DDR5).[58] Shared features across x86 sockets include robust power delivery for high-TDP chips, with consumer lines like Intel's LGA 115x/1200/1700 and AMD's AM4/AM5 often integrating GPUs to reduce system costs, alongside memory transitions from DDR4 (ubiquitous in 2010s sockets) to DDR5 in newer designs for bandwidth improvements up to 100 GB/s. Compatibility windows vary significantly, exemplified by AMD's AM4 supporting seven processor generations, contrasting Intel's shorter cycles of 2–3 years per socket. Pin counts have scaled with complexity, such as AM5's 1718 pins enabling finer signal routing for 5nm processes. These sockets' transitions highlight divergent strategies: Intel's rapid socket changes align with microarchitectural shifts like Meteor Lake's hybrid cores, necessitating new interfaces for efficiency, while AMD's extended support via AM4 and AM5 leverages Zen architectures for cost-effective longevity in consumer and enthusiast markets.| Vendor | Socket | Year Range | Key Processors | Pin Count/Type | Notable Features |
|---|---|---|---|---|---|
| Intel | Socket 8 | 1995 | Pentium Pro | 240 PGA | Early server focus, slot convertible |
| Intel | LGA 775 | 2004–2010 | Core 2 Duo/Quad | 775 LGA | DDR2/3, up to 1333 MHz FSB |
| Intel | LGA 1150 | 2013–2015 | Haswell/Broadwell | 1150 LGA | Integrated GPU, PCIe 3.0 |
| Intel | LGA 1151 | 2015–2018 | Skylake/Coffee Lake | 1151 LGA | Integrated GPU, PCIe 3.0 |
| Intel | LGA 1200 | 2020–2021 | Comet Lake/Rocket Lake | 1200 LGA | Integrated GPU, PCIe 4.0 |
| Intel | LGA 1700 | 2021–2024 | Alder Lake/Raptor Lake | 1700 LGA | Integrated GPU, PCIe 5.0 |
| Intel | LGA 1851 | 2024+ | Arrow Lake | 1851 LGA | DDR5, 250W TDP support |
| Intel | LGA 3647 | 2017–2023 | Skylake-SP Xeons | 3647 LGA | Dual-socket, 28+ cores |
| Intel | LGA 4677 | 2023+ | Sapphire Rapids Xeons | 4677 LGA | Dual-socket, 60+ cores, DDR5 |
| AMD | Socket A/AM2/AM3 | 2003–2011 | Athlon/Phenom II | 940–962 PGA | DDR2/3, HyperTransport |
| AMD | FM2 | 2012–2014 | Trinity APU | 906 PGA | Integrated Radeon GPU |
| AMD | AM4 | 2016–2022 | Ryzen 1000–5000 | 1331 PGA | 7-year support, Zen 1–3 |
| AMD | AM5 | 2022+ | Ryzen 7000/9000 | 1718 LGA | DDR5, PCIe 5.0 |
| AMD | sTRX4 | 2019–2022 | Threadripper 3000 | 4094 sPGA | 64 cores, quad-channel DDR4 |
| AMD | sWRX8 | 2020–2022 | Threadripper PRO 3000/5000 | 4094 sPGA | 64 cores, quad-channel DDR4 |
| AMD | sTR5 | 2023+ | Threadripper 7000/9000 | 6096 LGA | 96 cores, 8-channel DDR5 |