Quick Verdict
A very high‑core‑count, single‑socket Granite Rapids CPU with strong memory bandwidth, integrated accelerators, and competitive AI performance, best suited for users who can fully utilize 64 cores and justify the 350W TDP and platform cost.
Overview
Launch
2025
Status
LaunchedGeneration
6th Gen Xeon Scalable (Xeon 6 with P-Cores)
Market
1S Server / Workstation
The Intel Xeon 6761P is a 64-core, 128-thread server and workstation processor based on the Granite Rapids-SP architecture, built on Intel’s 3 process node. It targets single-socket platforms requiring high core counts, large memory capacity, and strong AI acceleration, with a 350W TDP and support for DDR5 and MRDIMM memory up to 8000 MT/s.
Intel’s Xeon 6761P sits in the Xeon 6 Performance family with 64 P‑cores, 128 threads, a 2.5 GHz base and 3.9 GHz max turbo, 336 MB of L3 cache, and a 350W TDP.
It supports up to 4 TB of DDR5‑6400 and MRDIMM‑8000 memory across eight channels, and exposes 136 PCIe 5.0 lanes. The chip includes Intel AMX for AI acceleration, along with a full set of server RAS and virtualization features.
It is aimed at dense, single‑socket workloads like virtualization, databases, in‑memory analytics, and AI inference where high core count and memory bandwidth matter more than peak single‑thread speed.
Specifications
Performance
Excels in heavily parallel productivity workloads like large‑scale data processing, in‑memory analytics, and multi‑tenant virtualization where core count and memory bandwidth dominate.
Designed for dense VM environments; 64 cores, 128 threads, and 8‑channel memory make it well suited for consolidation and VDI, though per‑VM licensing and platform cost should be evaluated.
Not targeted at gaming; high core counts do not translate into gaming performance, and platform cost and power are hard to justify for pure gaming use.
Performance per watt is competitive for many server workloads but the 350W TDP and high platform power mean efficiency favors fully loaded systems rather than lightly threaded tasks.
- •Not designed or marketed for gaming
- •Few games scale beyond 16–24 threads
- •Platform cost and power are disproportionate for gaming
- •Intel AMX accelerates matrix operations for inference and low‑precision training
- •DL Boost (AVX‑512 VNNI) improves INT8 inference throughput
- •Best suited for CPU‑based AI or as a host for discrete accelerators, not as a replacement for GPUs in large‑scale training
Architecture
Intel 3
Process Node
Granite Rapids-SP
Codename
64C / 128T
Core Config
336 MB
L3 Cache
350 W
TDP
Architecture Overview
Granite Rapids‑SP is Intel’s 6th‑generation Xeon Scalable server architecture, focusing on high core counts, high memory bandwidth, and integrated accelerators for AI and data processing. It uses Intel’s 3 process node and Redwood Cove P‑cores.
CPU Design
The Xeon 6761P integrates 64 Redwood Cove P‑cores with Hyper‑Threading, yielding 128 threads. Cores are organized with 24 high‑priority and 40 low‑priority cores via Intel Speed Select Technology, allowing operators to adjust frequency and priority per core to match workload requirements.
Memory Subsystem
An 8‑channel DDR5 memory controller supports DDR5‑6400 and MRDIMM‑8000, with a maximum capacity of 4 TB. This provides very high bandwidth for memory‑intensive workloads like in‑memory databases and analytics, and MRDIMM enables higher effective speeds than standard DDR5.
PCIe & I/O
136 PCIe 5.0 lanes are exposed directly from the CPU, supporting high‑speed NICs, GPUs, and NVMe storage. This is particularly valuable in single‑socket systems where PCIe density is critical.
Overclocking
The multiplier is locked; tuning is primarily through Intel Speed Select Technology profiles and platform firmware rather than traditional overclocking.
- Move from Intel 7 to Intel 3 process node
- Higher max turbo (3.9 GHz vs lower frequencies on many Sapphire Rapids SKUs)
- Addition of MRDIMM support for higher memory bandwidth
- Enhanced AI acceleration with Intel AMX and updated DL Boost
Key Highlights
- 64 cores / 128 threads for highly parallel workloads
- 8‑channel DDR5 / MRDIMM up to 8000 MT/s, up to 4 TB capacity
- 136 PCIe 5.0 lanes for dense I/O configurations
- Intel AMX and DL Boost for AI acceleration
- Mature server RAS and virtualization feature set
- Speed Select Technology for fine‑grained per‑core tuning
- High 350W TDP and associated cooling and power requirements
- Single‑socket only; no 2P scalability
- Premium pricing typical of high‑core‑count Xeon SKUs
- No integrated graphics (not expected in this segment)
- Locked multiplier; tuning is enterprise‑oriented, not enthusiast‑oriented
History
The Intel Xeon 6761P launched as part of the Granite Rapids‑SP generation, Intel’s 6th‑generation Xeon Scalable server lineup. It arrived after AMD’s EPYC 9004 “Genoa” and “Genoa‑X” parts had already pushed core counts and memory bandwidth higher in the server segment. Granite Rapids responded with higher per‑socket core counts, MRDIMM support, and integrated accelerators like AMX and IAA.
The 6761P carved out a space as a high‑core‑count, single‑socket part optimized for dense virtualization, in‑memory databases, and AI workloads, complementing dual‑socket designs that could not fully utilize very high core counts per socket. Over time, it became a common option in 1U and 2U servers from major OEMs, particularly where customers wanted a balance between compute density, memory bandwidth, and PCIe connectivity without moving to a 2P platform.
Improvements over Previous Generation
- Move from Intel 7 to Intel 3 process node
- Higher max turbo (3.9 GHz vs lower frequencies on many Sapphire Rapids SKUs)
- Addition of MRDIMM support for higher memory bandwidth
- Enhanced AI acceleration with Intel AMX and updated DL Boost
Alternatives & Competitors
Should You Buy It?
Recommended for the right buyer
Single‑socket servers or workstations that can keep 64 cores busy with parallel, memory‑intensive workloads such as virtualization, databases, analytics, and AI inference, and where high PCIe density and integrated accelerators are valuable.
Avoid if…
- Building a gaming or light‑use desktop
- Needing only 8–16 cores for typical office or light server work
- Prioritizing lowest possible power consumption
- Requiring dual‑socket scalability
Use Cases
Interesting Facts
The Xeon 6761P uses Intel’s Speed Select Technology to designate 24 cores as high‑priority and 40 as low‑priority, allowing different frequency and priority policies per core group.
It is one of the first Xeon 6 Performance parts to support MRDIMM‑8000, giving it a memory bandwidth advantage over DDR5‑only Granite Rapids SKUs.
Granite Rapids‑SP is built on the Intel 3 process node, which Intel positions as a significant density and efficiency improvement over Intel 7 used in Sapphire Rapids.
SPEC CPU2017 results published by multiple OEMs show the 6761P delivering strong integer and floating‑point throughput in 1P configurations, though exact scores vary by platform.
Despite being a server CPU, its 136 PCIe 5.0 lanes make it attractive for workstation‑class configurations with multiple GPUs and high‑speed storage.
The CPU integrates Intel QuickAssist Technology (QAT), DLB, DSA, and IAA accelerators, offloading compression, encryption, and analytics workloads from the cores.
Intel’s Xeon 6 platform introduces CXL 2.0 support alongside PCIe 5.0, though CXL capability depends on the chipset and motherboard implementation.
The 6761P is positioned below the 6767P/6768P in Intel’s stack, trading slight clock speed differences for a lower recommended customer price.
Intel advertises generational performance and performance‑per‑watt gains for Xeon 6 over prior Xeon Scalable generations, but does not publish a single official benchmark score for the 6761P.
OEMs such as Dell, HPE, and Lenovo offer the 6761P in 1U and 2U servers optimized for AI, virtualization, and database workloads.
People Also Ask
Is the Intel Xeon 6761P good for gaming?
It can run games, but it is not designed or priced for gaming. Most games do not benefit from 64 cores, and there are far cheaper CPUs that deliver better gaming performance per dollar and per watt.
What socket does the Intel Xeon 6761P use?
It uses the FCLGA4710 socket, also referred to as LGA 4710, part of Intel’s Xeon 6 platform.
How much memory does the Intel Xeon 6761P support?
It supports up to 4 TB of DDR5‑6400 and MRDIMM‑8000 memory across eight channels, depending on the DIMM types and population rules.
Does the Intel Xeon 6761P have integrated graphics?
No, it does not have integrated graphics. A discrete GPU or BMC graphics is required for display output.
Can the Intel Xeon 6761P be overclocked?
It has a locked multiplier. Performance tuning is done through Intel Speed Select Technology profiles and BIOS settings rather than traditional overclocking.
What is the difference between Intel Xeon 6761P and 6767P?
Both are 64‑core, 128‑thread Granite Rapids‑SP CPUs with 336 MB L3 and 350W TDP. The 6761P has a 2.5 GHz base and 3.9 GHz turbo; the 6767P runs at 2.4 GHz base and 3.9 GHz turbo with a higher recommended price, so the 6761P typically offers better value.
Is the Intel Xeon 6761P a processor or a platform?
It is a processor, but it belongs to the Xeon 6 platform, which includes chipsets, sockets, and features like PCIe 5.0 and CXL 2.0. The full feature set depends on the motherboard and chipset used.
What workloads benefit most from the Intel Xeon 6761P?
Workloads that scale well across many cores and need high memory bandwidth, such as virtualization, in‑memory databases, analytics, and AI inference, benefit most.
Does the Intel Xeon 6761P support ECC memory?
Yes, it supports ECC DDR5 and MRDIMM, which is standard for Xeon processors and important for data integrity in servers and workstations.
How does the Intel Xeon 6761P compare to AMD EPYC 9354?
The Xeon 6761P has twice the cores and threads but fewer memory channels (8 vs 12) and different memory speeds. EPYC 9354 may offer better per‑core licensing and memory bandwidth for some workloads, while the 6761P can provide higher aggregate throughput for highly parallel tasks.
Frequently Asked Questions
What generation is the Intel Xeon 6761P?
It is a 6th‑generation Xeon Scalable processor, marketed as Intel Xeon 6 with P‑Cores (Granite Rapids‑SP).
What process node is the Intel Xeon 6761P built on?
Intel specifies the lithography as Intel 3 for Granite Rapids, which is the node used for the 6761P.
How many PCIe lanes does the Intel Xeon 6761P have?
It provides 136 PCIe 5.0 lanes directly from the CPU.
Does the Intel Xeon 6761P support CXL?
The Granite Rapids architecture supports CXL 2.0, but actual CXL functionality depends on the motherboard and chipset implementation.
What is the maximum turbo frequency of the Intel Xeon 6761P?
The max turbo frequency is 3.9 GHz, with an all‑core turbo of 3.6 GHz.
Can the Intel Xeon 6761P run DDR4 memory?
No, it supports only DDR5 and MRDIMM memory types, not DDR4.
Is the Intel Xeon 6761P unlocked for overclocking?
No, it has a locked multiplier. Tuning is done through Intel Speed Select Technology and platform firmware.
What is the TDP of the Intel Xeon 6761P?
The TDP is 350W, which reflects the high core count and clock speeds under server workloads.
Does the Intel Xeon 6761P include AI acceleration?
Yes, it includes Intel AMX and Intel Deep Learning Boost (AVX‑512 VNNI) for AI inference and low‑precision training workloads.
Is the Intel Xeon 6761P still good in 2026?
For highly parallel server and workstation workloads that can use 64 cores, it remains a strong choice. For lightly threaded or budget‑sensitive builds, newer or lower‑core parts may offer better value.