INTEL

Intel Xeon Gold 6348

Intel processor specifications and benchmark scores

28
Cores
56
Threads
3.5
GHz Boost
235W
TDP
ECC Memory

At a Glance

Intel
Cores / Threads 28C / 56T
Boost Clock 3.5 GHz
Base Clock 2.6 GHz
L3 Cache 42 MB (shared)
TDP 235W
Architecture Ice Lake
Socket Intel Socket 4189
nm
Process 10 nm
Released Apr 2021

Intel Xeon Gold 6348 Specifications

Xeon Gold 6348 Core Configuration

Processing cores and threading

The Intel Xeon Gold 6348 features 28 physical cores and 56 threads, which directly impacts multi-threaded performance in CPU benchmarks. More cores allow the processor to handle parallel workloads efficiently, improving performance in video editing, 3D rendering, and multitasking scenarios. Thread count determines how many simultaneous tasks the CPU can process, with higher thread counts benefiting productivity applications and content creation workflows.

Cores
28
Threads
56
SMP CPUs
2

Gold 6348 Clock Speeds

Base and boost frequencies

Clock speed is a critical factor in Xeon Gold 6348 benchmark performance, measured in GHz. The base clock represents the guaranteed operating frequency, while the boost clock indicates maximum single-core performance under optimal conditions. Higher clock speeds translate to faster single-threaded performance, which is essential for gaming and applications that don't fully utilize multiple cores. The Xeon Gold 6348 by Intel can dynamically adjust its frequency based on workload and thermal headroom.

Base Clock
2.6 GHz
Boost Clock
3.5 GHz
Multiplier
26x

Intel's Xeon Gold 6348 Cache Hierarchy

L1, L2, L3 cache sizes

Cache memory is ultra-fast storage built directly into the Gold 6348 processor die. L1 cache provides the fastest access for frequently used data, while L2 and L3 caches offer progressively larger storage with slightly higher latency. Larger cache sizes significantly improve CPU benchmark scores by reducing memory access times. The Xeon Gold 6348's cache configuration is optimized for both gaming performance and productivity workloads, minimizing data fetch delays during intensive computations.

L1 Cache
64 KB (per core)
L2 Cache
1 MB (per core)
L3 Cache
42 MB (shared)

Ice Lake Architecture & Process

Manufacturing and design details

The Intel Xeon Gold 6348 is built on Intel's 10 nm manufacturing process, which determines power efficiency and thermal characteristics. Smaller process nodes allow for more transistors in the same space, enabling higher performance per watt. The architecture defines how the processor handles instructions and manages data flow, directly impacting benchmark results across different workload types. Modern CPU architectures like the one in Gold 6348 incorporate advanced branch prediction and out-of-order execution for optimal performance.

Architecture
Ice Lake
Codename
Ice Lake-SP
Process Node
10 nm
Foundry
Intel
Generation
Xeon Gold (Ice Lake-SP)

Ice Lake Instruction Set Features

Supported CPU instructions and extensions

The Xeon Gold 6348 by Intel supports various instruction set extensions that enable optimized performance for specific workloads. SIMD instructions like SSE and AVX accelerate multimedia, scientific computing, and AI workloads by processing multiple data points simultaneously. Features like AES-NI provide hardware-accelerated encryption, while AVX-512 (if supported) enables advanced vector processing for data centers and high-performance computing. These instruction sets are critical for software compatibility and performance in modern applications.

MMX
SSE
SSE2
SSE3
SSSE3
SSE4.1
SSE4.2
AVX
AVX2
AVX-512
FMA3
SHA
AES-NI
F16C
BMI1
BMI2
Intel 64
VT-x
VT-d

Gold 6348 Power & Thermal

TDP and power specifications

The Intel Xeon Gold 6348 has a TDP (Thermal Design Power) of 235W, indicating the cooling solution required for sustained operation. TDP affects both system power consumption and the type of cooler needed. Lower TDP processors are ideal for compact builds and laptops, while higher TDP chips typically offer better sustained performance in demanding CPU benchmarks. Understanding power requirements helps ensure your system can deliver consistent performance without thermal throttling.

TDP
235W

Intel Socket 4189 Platform & Socket

Compatibility information

The Xeon Gold 6348 uses the Intel Socket 4189 socket, which determines motherboard compatibility. Choosing the right platform is essential for building a system around this processor. The socket type also influences available features like PCIe lanes, memory support, and upgrade paths. When comparing CPU benchmarks, ensure you're looking at processors compatible with your existing or planned motherboard to make informed purchasing decisions.

Socket
Intel Socket 4189
PCIe
Gen 4, 64 Lanes(CPU only)
Package
FC-LGA4189
DDR5

Intel Socket 4189 Memory Support

RAM compatibility and speeds

Memory support specifications for the Gold 6348 define which RAM types and speeds are compatible. Faster memory can significantly improve CPU benchmark performance, especially in memory-intensive applications and gaming. The memory controller integrated into the Xeon Gold 6348 determines maximum supported speeds and channels. Dual-channel or quad-channel memory configurations can double or quadruple memory bandwidth, providing noticeable performance gains in content creation and scientific workloads.

Memory Type
DDR4
Memory Bus
Eight-channel
Memory Bandwidth
204.8 GB/s
ECC Memory
Supported

Xeon Gold 6348 Product Information

Release and pricing details

The Intel Xeon Gold 6348 is manufactured by Intel and represents their commitment to delivering competitive CPU performance. Understanding the release date and pricing helps contextualize benchmark comparisons with other processors from the same generation. Launch pricing provides a baseline for evaluating value, though street prices often differ. Whether you're building a new system or upgrading, the Xeon Gold 6348 by Intel offers a specific balance of performance, features, and cost within Intel's product lineup.

Manufacturer
Intel
Release Date
Apr 2021
Market
Server/Workstation
Status
Active
Part Number
SRKHPCD8068904572204

Xeon Gold 6348 Benchmark Scores

cinebench_cinebench_r15_multicoreSource

Cinebench R15 multi-core renders a complex 3D scene using all CPU threads simultaneously. This test reveals how Intel Xeon Gold 6348 performs in parallel rendering workloads like video production and 3D animation. The R15 version remains useful for comparing against older hardware benchmarks. Higher scores directly correlate with faster render times in Cinema 4D and similar 3D applications.

cinebench_cinebench_r15_multicore #148 of 1945
4,441
30%
Max: 14,978

cinebench_cinebench_r15_singlecoreSource

Cinebench R15 single-core measures the speed of one CPU thread rendering 3D geometry. This score indicates how Intel Xeon Gold 6348 handles tasks that can't be parallelized across multiple cores. Games and many desktop applications still rely heavily on single-thread performance. A higher single-core score means snappier system responsiveness in everyday use.

cinebench_cinebench_r15_singlecore #143 of 1351
626
30%
Max: 2,114

cinebench_cinebench_r20_multicoreSource

Cinebench R20 multi-core uses a scene requiring 4x more computational power than R15. This test better reflects modern CPU capabilities for professional rendering on Intel Xeon Gold 6348. The more demanding workload provides better differentiation between current-generation processors.

cinebench_cinebench_r20_multicore #148 of 1945
18,507
30%
Max: 62,412
Compare with other CPUs

cinebench_cinebench_r20_singlecoreSource

Cinebench R20 single-core tests one thread against a more demanding scene than R15. This reveals the true single-thread rendering capability of Intel Xeon Gold 6348. The increased complexity provides more accurate performance differentiation between modern CPUs.

cinebench_cinebench_r20_singlecore #143 of 1935
2,612
30%
Max: 8,811

cinebench_cinebench_r23_multicoreSource

Cinebench R23 multi-core is the current standard for CPU rendering benchmarks with a 10-minute minimum runtime. This extended test reveals sustained performance of Intel Xeon Gold 6348 after thermal limits kick in. The longer duration exposes cooling limitations that shorter benchmarks miss.

cinebench_cinebench_r23_multicore #148 of 1945
44,066
30%
Max: 148,601
Compare with other CPUs

cinebench_cinebench_r23_singlecoreSource

Cinebench R23 single-core measures sustained single-thread performance over 10 minutes. This reveals how Intel Xeon Gold 6348 maintains boost clocks under continuous load. The extended runtime shows whether thermal throttling affects single-core performance.

cinebench_cinebench_r23_singlecore #135 of 1932
6,221
30%
Max: 20,979

About Intel Xeon Gold 6348

Intel Xeon Gold 6348 is a 28-core, 56-thread Ice Lake-SP processor for Intel Socket 4189, built for server and workstation workloads. It sits in the 72nd percentile of all CPUs in the database, with an average benchmark score of 12746. This places it in a peculiar spot: its raw multi-threaded throughput is substantial, but its nearest rivals in aggregate scoring are a mix of mobile and entry-level desktop parts, indicating that the average score hides a very specific performance profile.

Who Should Consider It

The benchmark data indicates this chip is exclusively for multi-threaded, server-class workloads. In Cinebench R23 multi-core, it scores 44066, which is an extreme outlier compared to its nearest rivals and represents the core strength of the design. Anyone running heavily threaded rendering, scientific simulation, or virtualized server environments where 28 physical cores can be fully utilized will find this processor's score compelling. The 56 threads and 42 MB of shared L3 cache suggest workloads that scale linearly with core count will benefit most.

For gaming, this is not a sensible choice. The single-core scores are modest — 6221 in Cinebench R23 single-core, 2612 in R20, and 626 in R15. These numbers place it far below typical desktop gaming CPUs. The data does not suggest any gaming-oriented feature, and the high core count does not translate to the single-thread responsiveness that gaming workloads require. A builder assembling a gaming rig should look elsewhere.

For office and general productivity, the processor is overkill. The 204.8 GB/s of eight-channel DDR4 memory bandwidth and 64 PCIe Gen 4 lanes are server features that a standard office workload will not leverage. Daily tasks like web browsing or document editing do not scale with 28 cores, and the modest single-thread performance means it will not feel snappier than a modern consumer chip. This is a tool for a specific job, not a general-purpose desktop part.

Content creation is where this processor finds its niche, but only for the rendering and export phases. In video encoding, 3D rendering, or batch image processing, the multi-core score of 44066 in R23 indicates it will crush these tasks. However, the interactive portions of creation work — timeline scrubbing, applying filters, previewing effects — depend on single-thread performance, where the chip's 6221 R23 score is unremarkable. The data suggests a mixed workflow would benefit from pairing this with a separate high-single-thread machine or accepting slower interactive performance.

Power and Thermals

The TDP is 235 watts. This is a high-power part that demands a serious cooling solution. The data does not specify cooler requirements, but a 235W TDP class processor implies a large tower air cooler or a robust liquid cooling solution is necessary for sustained all-core loads. In a server chassis, this typically means high-static-pressure fans and careful airflow design.

The Ice Lake architecture on a 10 nm process from Intel does not change the thermal reality. With 28 cores running at a 2.60 GHz base clock and boosting to 3.50 GHz, the power draw under full multi-threaded load will approach the TDP limit. The eight-channel memory controller also adds to the thermal load. Any system integrating this processor must plan for the heat output of both the CPU and the surrounding voltage regulation circuitry.

Practical implications: a consumer-grade compact cooler will not suffice. The data implies a workstation or server motherboard with substantial VRM cooling is necessary. The 235W TDP also has implications for the power supply and chassis. This is not a chip for a small form factor build; it requires the airflow and space of a proper workstation tower or rack-mount system.

How It Compares

Intel Core i7-1185G7: The Xeon Gold 6348 has an average score of 12746, which is 0.2% higher than the i7-1185G7's 12715. This negligible delta is misleading. The i7-1185G7 is a low-power mobile chip with 4 cores; the Xeon achieves a nearly identical average score with 28 cores. This comparison highlights how the Xeon's single-thread performance is a severe drag on its average, while its multi-core advantage is massive. In any threaded workload, the Xeon will be several times faster, but in single-thread tasks, the mobile chip is competitive or better.

AMD EPYC 7502: The Xeon leads the EPYC 7502 by 0.3% in average score (12746 vs 12709). The EPYC 7502 is a server part like the Xeon, making this a more relevant comparison. The near-parity in average score suggests that, despite architectural differences, the overall benchmark aggregate lands in the same tier. The Xeon's advantage likely comes from its higher boost clock or cache hierarchy, but the data does not break down per-test deltas. The practical takeaway is that these two server chips are comparable in overall performance, and the choice may come down to platform features rather than raw score.

Intel Core i3-12100: The Xeon is 0.5% ahead of the i3-12100 in average score (12746 vs 12682). This is the most striking comparison. The i3-12100 is a 4-core, 8-thread desktop chip. The fact that a 28-core Xeon only edges it out in aggregate benchmark score is a damning indictment of the Xeon's single-thread performance. The i3-12100's modern architecture gives it strong single-core scores, which prop up its average. In multi-core tests, the Xeon will dominate, but the average score masks this. This comparison underscores that the Xeon is not a general-purpose CPU.

Intel Core i3-1210U: The Xeon trails the i3-1210U by 0.6% in average score (12746 vs 12824). The i3-1210U is a 2-core, 4-thread ultra-low-power mobile chip. The Xeon loses to this part in the aggregate, which is entirely due to the i3-1210U's superior single-thread performance relative to its multi-thread weakness. This is the clearest signal that the Xeon's benchmark profile is heavily skewed. Anyone purchasing this Xeon must ignore the average score and focus on the multi-core results, which are the actual reason to buy it.

FAQ

Q: Is the Intel Xeon Gold 6348 good for gaming?

A: No. The single-core scores are modest (6221 in Cinebench R23, 2612 in R20), which are far below what gaming workloads require. The 28-core design is optimized for multi-threaded server tasks, not the low-latency single-thread performance that games need.

Q: How much L3 cache does it have?

A: The processor has 42 MB of shared L3 cache. Each core also has 64 KB of L1 and 1 MB of L2 cache.

Q: What memory does it support?

A: It supports DDR4 memory with an eight-channel memory bus, providing a memory bandwidth of 204.8 GB/s. ECC memory is supported.

Q: What socket does this processor use?

A: It uses Intel Socket 4189. This is a server platform socket, not compatible with consumer desktop motherboards.

Q: What is the production status?

A: The production status is listed as Active, and the release date is April 5, 2021.

Q: How does it compare to the Intel Core i3-12100?

A: In average benchmark score, the Xeon Gold 6348 is 0.5% ahead of the i3-12100. However, this average is misleading. The Xeon has vastly superior multi-core performance (44066 in R23 multi-core), while the i3-12100 has better single-thread performance relative to its core count.

Benchmark Performance

The benchmark data reveals a processor with extreme multi-core strength and modest single-core capability. In Cinebench R23 multi-core, the Xeon scores 44066. This is the headline number. In Cinebench R20 multi-core, it scores 18507, and in R15 multi-core, 4441. These scores scale consistently across versions, indicating strong all-core scaling.

The average benchmark score of 12746 is dragged down by the single-core results. In Cinebench R23 single-core, it scores 6221; in R20, 2612; and in R15, 626. These single-core numbers are roughly in line with what a modern mid-range desktop chip might achieve, but they are far below what the multi-core score would suggest for a flagship server part.

Comparing to rivals, the Xeon's multi-core dominance is clear. Against the AMD EPYC 7502, which has a similar average score (12709 vs 12746), the Xeon's multi-core performance in real workloads would be competitive, but the aggregate hides the specific distribution. Against the Core i3-12100 and i3-1210U, the multi-core gap is enormous — the Xeon is likely several times faster in threaded tests — yet the average score is nearly identical due to the rivals' superior single-core efficiency.

The percentile rank of 72 indicates that 28% of all CPUs in the database score higher on average. This is a low percentile for a 28-core server chip, again reflecting the single-thread penalty. A more balanced processor with similar multi-core performance would rank higher. The data suggests that the Xeon Gold 6348 is a specialized tool that trades single-thread performance for massive parallel throughput.

Single-Thread vs Multi-Thread Behavior

The split between single-thread and multi-thread scores is the defining characteristic of this processor. The multi-core scores are exceptional: 44066 in R23, 18507 in R20, and 4441 in R15. The single-core scores are 6221, 2612, and 626 respectively. The ratio between multi-core and single-core in R23 is roughly 7.1x, which is lower than the ideal 28x scaling from 28 cores. This indicates that while the chip does scale well, there are diminishing returns due to thermal or power limits.

For real workloads, this means the Xeon Gold 6348 excels in batch processing and parallelizable tasks. Video rendering, 3D animation, code compilation, and database queries that can use all 56 threads will see near-linear speedups. The 42 MB L3 cache helps keep data close to the cores, and the eight-channel memory bandwidth of 204.8 GB/s feeds the cores adequately.

However, for latency-sensitive tasks that rely on a single thread — such as running a database with a single-threaded query path, or any interactive application — the chip will feel sluggish. The 2.60 GHz base clock and 3.50 GHz boost clock are low by modern standards, and the single-core scores confirm this. A workload that cannot use more than one or two threads will be bottlenecked by this processor's modest single-thread throughput.

The practical advice is to match the workload to the architecture. If the software scales with core count, the Xeon Gold 6348 is a strong choice with a high multi-core ceiling. If the software is single-threaded, the chip's performance will be unimpressive, and the high TDP and platform cost are not justified. The benchmark data is unambiguous: this is a multi-threaded workhorse, not a balanced all-rounder.

The AMD Equivalent of Xeon Gold 6348

Looking for a similar processor from AMD? The AMD Ryzen 5 5600G offers comparable performance and features in the AMD lineup.

AMD Ryzen 5 5600G

AMD • 6 Cores

View Specs Compare

Popular Intel Xeon Gold 6348 Comparisons

See how the Xeon Gold 6348 stacks up against similar processors from the same generation and competing brands.

Compare Xeon Gold 6348 with Other CPUs

Select another CPU to compare specifications and benchmarks side-by-side.

Browse CPUs