AMD Ryzen Threadripper 9960X

AMD Ryzen Threadripper 9960X

AMD Ryzen Threadripper 9960X: 24-core HEDT processor for workstations and enthusiasts

Ryzen Threadripper 9960X is a top consumer HEDT model in the Threadripper 9000 lineup (Zen 5 architecture) with 24 cores and 48 threads. It targets heavy production workloads and powerful desktop workstations built on the TRX50 platform. Key traits include high clock speeds, a large cache, quad-channel DDR5, and up to 80 lanes of PCIe 5.0 for graphics and storage; an integrated GPU and a dedicated NPU are not present.

Key specifications

• Architecture/codename, process: Zen 5, chiplet platform Shimada Peak; compute dies on 4 nm, I/O die on 6 nm.
• Cores/threads: 24 / 48.
• Frequencies (base; boost): 4.2 GHz; up to 5.4 GHz (actual values depend on thermal budget and motherboard settings).
• L3 cache: 128 MB total; L2 — 24 MB (1 MB per core); combined cache — 152 MB.
• Power envelope: TDP 350 W; supports efficiency profiles (including Eco Mode and PPT/TDC/EDC limits in BIOS/UEFI, depending on the board).
• Integrated graphics: none (discrete GPU required).
• Memory: DDR5, quad-channel; UDIMM support (including ECC-UDIMM on boards that validate it). Recommended effective frequencies depend on the motherboard’s QVL; a typical reference point is up to DDR5-6400.
• I/O: up to 80 PCIe 5.0 lanes from the CPU; platform-level lanes/ports vary by TRX50 motherboard (commonly multiple M.2 PCIe 5.0 x4 slots, PCIe x16/x8 slots, 2.5/10 Gbps Ethernet; USB4 40 Gbps where controller support is provided by the board). Multi-display support is handled by a discrete GPU.
• NPU/Ryzen AI: no; AI acceleration relies on the CPU (AVX-512, VNNI, BF16) and/or a discrete GPU.

What this chip is and where it is used

Ryzen Threadripper 9960X belongs to the HEDT “enthusiast” segment and occupies a position as a high-performance yet non-professional solution: above it sits the Threadripper PRO 9000 WX line with more memory channels and PCIe lanes; below are mainstream Ryzen 9000 processors on the AM5 platform. The model is intended for desktop workstations in ATX/SSI-CEB/EEB form factors on TRX50 boards and is aimed at video editing and encoding, rendering, modeling, compiling large codebases, CAD/EDA, virtualization, and data-heavy workflows.

Architecture and manufacturing process

At its core, the 9960X uses the Zen 5 microarchitecture with a chiplet layout. Compute dies (CCD) are fabricated on a 4 nm node and connected to the I/O die (IOD) via Infinity Fabric. Zen 5 improvements span the front end, branch prediction, vector execution units, and memory subsystem; AVX-512 (implemented through paired 256-bit units) is supported, including VNNI and BF16, which is useful for media and CPU-based AI workloads.
The memory controller is quad-channel DDR5 with EXPO/XMP profile handling at the board level; performance and stable frequencies depend on module quality and motherboard trace design. The large shared L3 cache (128 MB) reduces sensitivity to main-memory latency in many highly threaded tasks, while 1 MB of L2 per core lowers access latency under intensive data processing.
There is no integrated graphics block in this line; hardware video decode/encode is provided by a discrete GPU or performed in software on the CPU using SIMD instructions.

CPU performance

Practical performance of the 9960X is determined by the blend of 24 full Zen 5 cores, high clock speeds, and a large cache. In building large projects (C/C++/Rust/Java), multi-threaded rendering (CPU engines), photo and video conversion, archiving, and analytical pipelines, the model demonstrates a substantial advantage over mainstream desktop CPUs thanks to thread count and I/O width.
Power-limit settings and cooling efficiency are critical: at a 350 W TDP, sustained frequencies under prolonged load depend on the motherboard VRM capabilities, the cooling solution, and the chassis thermal regime. Eco Mode can deliver a balanced “performance/efficiency” profile where modest clock reductions are offset by lower heat and noise. In latency-bound scenarios or where software licensing caps cores, scaling may fall short of linear, but high boost frequencies retain competitive single-thread speed.

Graphics and multimedia (iGPU)

There is no integrated GPU. A discrete graphics card is required for display output and hardware acceleration of codecs. This applies to gaming as well as professional NLE/3D pipelines—rendering, effects, and video encoding in professional applications are typically GPU-accelerated. System memory plays a minimal role in graphics for this model, since graphics workloads rely on the discrete card and its VRAM.

AI/NPU (where applicable)

There is no dedicated NPU in the Ryzen Threadripper 9960X. On-device acceleration of AI models runs on the CPU (AVX-512, VNNI, BF16) and/or a discrete GPU (CUDA/ROCm/DirectML), depending on the framework stack. For inference of large transformers and acceleration of generative graphics, the practical optimization point is the GPU and its VRAM, with the CPU providing high throughput for loading and preprocessing stages.

Platform and I/O

The processor fits the sTR5 socket and operates on TRX50 motherboards. The platform provides up to 80 PCIe 5.0 lanes from the CPU; board vendors distribute them across PCIe x16/x8 slots and M.2 connectors for NVMe drives (PCIe 5.0 x4). Typical configurations include two full-speed x16 slots for graphics/accelerators and 3–5 M.2 sockets supporting PCIe 5.0/4.0, with U.2/OCuLink available on some models.
Networking depends on the board: from 2.5 Gbps to 10 Gbps Ethernet, often with additional Wi-Fi 6E/7. USB ports range from USB 3.2 Gen2x2 to USB4 40 Gbps on boards with the requisite controllers; Thunderbolt support is provided on boards that declare compatibility. RAID across NVMe and SATA is determined by BIOS/UEFI features and chipset drivers.

Power consumption and cooling

The nominal 350 W TDP requires a high-grade power delivery and cooling system. For sustained operation under long multi-threaded loads, 360–420 mm AIO liquid coolers or top-tier tower coolers with official sTR5 support and a wide baseplate for the Threadripper heatspreader are appropriate. The case should ensure a straight-through airflow pattern with adequate intake/exhaust area; dedicated VRM airflow is recommended.
Power parameters (PPT/TDC/EDC) and Precision Boost Overdrive are adjustable in BIOS/UEFI. Eco profiles lower heat and noise at the cost of minor performance loss in long renders/simulations; conversely, aggressive PB/PBO settings raise boost clocks if thermal and electrical budgets allow.

Where the processor is found

Ryzen Threadripper 9960X is used in HEDT workstations from system integrators and in custom builds on TRX50 platforms. These are desktop workstations for video editing, color grading, media encoding, 3D graphics, photo pipelines, compilation and CI farms, simulations and visualization, DCC packages, as well as enthusiast systems with multiple accelerators where PCIe lane count and high storage bandwidth are important.

Comparison and positioning

Within the HEDT Threadripper 9000 “X” series, three key models are common: 9960X (24C/48T), 9970X (32C/64T), and 9980X (64C/128T). All share the same nominal 350 W TDP, similar boost clocks, quad-channel DDR5, and up to 80 PCIe 5.0 lanes.
Compared to the professional Threadripper PRO 9000 WX lineup, the “X” series yields on memory channels (4 vs 8) and total PCIe lanes (80 vs 128 on PRO) but retains high frequencies and uses the more accessible TRX50 platform instead of WRX90.

Who it suits

• Production pipelines: non-linear editing, color work, video encoding, photo catalogs, batch processing.
• 3D rendering and DCC workloads that scale with CPU threads and need fast I/O for caches and assets.
• Software development: compiling large projects, multi-threaded test runs, containerization and CI.
• Virtualization and lab setups: multiple VMs with high CPU and storage demands.
• Enthusiast configurations with multiple accelerators/storage devices that require numerous PCIe lanes.

Pros and cons

Pros

  1. 24 full Zen 5 cores with high boost clocks.

  2. Up to 80 PCIe 5.0 lanes — flexible GPU/SSD/network adapter layouts.

  3. Large cache (128 MB L3) benefiting various compute and media workloads.

  4. Quad-channel DDR5 with ECC-UDIMM support on many boards.

  5. TRX50 platform with a wide choice of boards and interfaces.

Cons

  1. High 350 W TDP — demanding on cooling and motherboard VRM.

  2. No iGPU — a discrete card is mandatory even for basic display output.

  3. No dedicated NPU — on-device AI acceleration is best handled by a powerful GPU.

  4. Performance in some tasks can be bounded by per-core/socket licensing or memory latency.

  5. Platform costs (board, memory, cooling) exceed mainstream AM5 systems.

Configuration recommendations

Memory: install four DDR5 modules to enable all four channels; prioritize stability and QVL compatibility. For data-heavy projects, ECC-UDIMM is sensible (if the board supports it). Choose effective frequencies within board-validated ranges; optimize timings and synchronize with Fabric frequency where applicable.
Storage: for OS and project workspace use a fast NVMe PCIe 5.0/4.0 x4 drive; dedicate a separate SSD for cache and previews; for large media sets with deep read queues, consider scalable arrays or multiple independent NVMe volumes.
Cooling: AIO 360–420 mm or a massive tower cooler with full sTR5 compatibility and reinforced mounting; ensure VRM airflow. Use a quality thermal compound applied evenly to the large heatspreader.
Power and chassis: a PSU with headroom and robust 12VHPWR/PCIe lines, especially for multi-GPU builds; a case with straight airflow and effective dust filtration.
Power profiles: start at stock, then enable Eco Mode or custom PBO limits as needed. For long renders, sustained clocks matter more than short-term peaks.
Software and licensing: account for licensing models (per core/socket); it may be optimal to cap thread counts per job for a better “speed/cost” balance.

Summary

Ryzen Threadripper 9960X is a balanced HEDT processor for workstations, combining 24 Zen 5 cores, high frequencies, a large cache, and rich I/O on TRX50. The absence of an iGPU and NPU is offset by the intended use with discrete accelerators, while 80 PCIe 5.0 lanes simplify building systems with multiple GPUs and fast NVMe arrays. In scenarios that benefit from four memory channels and broad PCIe resources, the 9960X delivers strong practical performance; for cases requiring eight memory channels, up to 128 PCIe lanes, and extended memory features, the Threadripper PRO 9000 WX line is the natural alternative.

Basic

Label Name
AMD
Platform
Desktop
Launch Date
July 2025
Model Name
?
The Intel processor number is just one of several factors - along with processor brand, system configurations, and system-level benchmarks - to be considered when choosing the right processor for your computing needs.
Ryzen Threadripper 9960X
Code Name
Shimada Peak
Foundry
TSMC
Generation
Ryzen Threadripper (Zen 5 (Shimada Peak))

CPU Specifications

Total Cores
?
Cores is a hardware term that describes the number of independent central processing units in a single computing component (die or chip).
24
Total Threads
?
Where applicable, Intel® Hyper-Threading Technology is only available on Performance-cores.
48
Performance-core Base Frequency
4.2 GHz
Performance-core Max Turbo Frequency
?
Maximum P-core turbo frequency derived from Intel® Turbo Boost Technology.
5.4 GHz
L1 Cache
64 KB per core
L2 Cache
1 MB per core
L3 Cache
128 MB shared
Bus Frequency
100 MHz
Multiplier
42.0
Unlocked Multiplier
Yes
CPU Socket
?
The socket is the component that provides the mechanical and electrical connections between the processor and motherboard.
AMD Socket sTR5
Technology
?
Lithography refers to the semiconductor technology used to manufacture an integrated circuit, and is reported in nanometer (nm), indicative of the size of features built on the semiconductor.
4 nm
TDP
350 W
Max. Operating Temperature
?
Junction Temperature is the maximum temperature allowed at the processor die.
95°C
PCIe Version
?
PCI Express is a high-speed serial computer expansion bus standard used for connecting high-speed components, replacing older standards such as AGP, PCI, and PCI-X. It has gone through multiple revisions and improvements since its initial release. PCIe 1.0 was first introduced in 2002, and in order to meet the growing demand for higher bandwidth, subsequent versions have been released over time.
5
Transistor Count
33.26 billions

Memory Specifications

Memory Type
?
Intel® processors come in four different types: Single Channel, Dual Channel, Triple Channel, and Flex Mode. Maximum supported memory speed may be lower when populating multiple DIMMs per channel on products that support multiple memory channels.
DDR5-6400
Max Memory Size
?
Max memory size refers to the maximum memory capacity supported by the processor.
1 TB
Memory Channels
?
The number of memory channels refers to the bandwidth operation for real world application.
4
ECC Memory Support
Yes

GPU Specifications

Integrated Graphics Model
?
An integrated GPU refers to the graphics core that is integrated into the CPU processor. Leveraging the processor's powerful computational capabilities and intelligent power efficiency management, it delivers outstanding graphics performance and a smooth application experience at a lower power consumption.
N/A

Miscellaneous

PCIe Lanes
48

Benchmarks

Geekbench 6
Single Core Score
2970
Geekbench 6
Multi Core Score
26026

Compared to Other CPU

Geekbench 6 Single Core
3978 +33.9%
2605 -12.3%
2431 -18.1%
Geekbench 6 Multi Core
16366 -37.1%
14750 -43.3%
13522 -48%