EPYC 9354 vs Xeon Platinum 8168
Aggregate performance score
EPYC 9354 outperforms Xeon Platinum 8168 by a whopping 105% based on our aggregate benchmark results.
Primary details
Comparing Xeon Platinum 8168 and EPYC 9354 processor market type (desktop or notebook), architecture, sales start time and price.
| Place in the ranking | 259 | 53 |
| Place by popularity | not in top-100 | not in top-100 |
| Cost-effectiveness evaluation | 3.75 | 9.97 |
| Market segment | Server | Server |
| Series | Intel Xeon Platinum | AMD EPYC |
| Power efficiency | 9.62 | 14.44 |
| Architecture codename | Skylake (server) (2017−2019) | Genoa (2022−2023) |
| Release date | 25 April 2017 (7 years ago) | 10 November 2022 (1 year ago) |
| Launch price (MSRP) | $5,890 | $3,420 |
Cost-effectiveness evaluation
Performance per price, higher is better.
EPYC 9354 has 166% better value for money than Xeon Platinum 8168.
Detailed specifications
Xeon Platinum 8168 and EPYC 9354 basic parameters such as number of cores, number of threads, base frequency and turbo boost clock, lithography, cache size and multiplier lock state. These parameters indirectly say of CPU speed, though for more precise assessment you have to consider their test results.
| Physical cores | 24 (Tetracosa-Core) | 32 (Dotriaconta-Core) |
| Threads | 48 | 64 |
| Base clock speed | 2.7 GHz | 3.25 GHz |
| Boost clock speed | 3.7 GHz | 3.25 GHz |
| Multiplier | 27 | 32.5 |
| L1 cache | 1.5 MB | 2 MB |
| L2 cache | 24 MB | 32 MB |
| L3 cache | 33 MB | 256 MB (shared) |
| Chip lithography | 14 nm | 5 nm, 6 nm |
| Die size | no data | 8x 72 mm2 |
| Maximum core temperature | 85 °C | no data |
| Number of transistors | no data | 52,560 million |
| 64 bit support | + | + |
| Windows 11 compatibility | + | no data |
Compatibility
Information on Xeon Platinum 8168 and EPYC 9354 compatibility with other computer components: motherboard (look for socket type), power supply unit (look for power consumption) etc. Useful when planning a future computer configuration or upgrading an existing one. Note that power consumption of some processors can well exceed their nominal TDP, even without overclocking. Some can even double their declared thermals given that the motherboard allows to tune the CPU power parameters.
| Number of CPUs in a configuration | 8 (Multiprocessor) | 2 |
| Socket | FCLGA3647 | SP5 |
| Power consumption (TDP) | 205 Watt | 280 Watt |
Technologies and extensions
Technological solutions and additional instructions supported by Xeon Platinum 8168 and EPYC 9354. You'll probably need this information if you require some particular technology.
| Instruction set extensions | Intel® SSE4.2, Intel® AVX, Intel® AVX2, Intel® AVX-512 | no data |
| AES-NI | + | + |
| AVX | + | + |
| vPro | + | no data |
| Enhanced SpeedStep (EIST) | + | no data |
| Speed Shift | + | no data |
| Turbo Boost Technology | 2.0 | no data |
| Hyper-Threading Technology | + | no data |
| TSX | + | - |
| Turbo Boost Max 3.0 | - | no data |
| Precision Boost 2 | no data | + |
Security technologies
Xeon Platinum 8168 and EPYC 9354 technologies aimed at improving security, for example, by protecting against hacks.
| TXT | + | no data |
| EDB | + | no data |
Virtualization technologies
Virtual machine speed-up technologies supported by Xeon Platinum 8168 and EPYC 9354 are enumerated here.
| AMD-V | - | + |
| VT-d | + | no data |
| VT-x | + | no data |
| EPT | + | no data |
Memory specs
Types, maximum amount and channel quantity of RAM supported by Xeon Platinum 8168 and EPYC 9354. Depending on the motherboard, higher memory frequencies may be supported.
| Supported memory types | DDR4-2666 | DDR5-4800 |
| Maximum memory size | 768 GB | 6 TiB |
| Max memory channels | 6 | no data |
| Maximum memory bandwidth | 128.001 GB/s | 460.8 GB/s |
| ECC memory support | + | - |
Peripherals
Specifications and connection of peripherals supported by Xeon Platinum 8168 and EPYC 9354.
| PCIe version | 3.0 | 5.0 |
| PCI Express lanes | 48 | 128 |
Synthetic benchmark performance
Various benchmark results of the processors in comparison. Overall score is measured in points in 0-100 range, higher is better.
Combined synthetic benchmark score
This is our combined benchmark performance rating. We are regularly improving our combining algorithms, but if you find some perceived inconsistencies, feel free to speak up in comments section, we usually fix problems quickly.
Passmark
Passmark CPU Mark is a widespread benchmark, consisting of 8 different types of workload, including integer and floating point math, extended instructions, compression, encryption and physics calculation. There is also one separate single-threaded scenario measuring single-core performance.
GeekBench 5 Single-Core
GeekBench 5 Single-Core is a cross-platform application developed in the form of CPU tests that independently recreate certain real-world tasks with which to accurately measure performance. This version uses only a single CPU core.
GeekBench 5 Multi-Core
GeekBench 5 Multi-Core is a cross-platform application developed in the form of CPU tests that independently recreate certain real-world tasks with which to accurately measure performance. This version uses all available CPU cores.
Pros & cons summary
| Performance score | 21.63 | 44.33 |
| Recency | 25 April 2017 | 10 November 2022 |
| Physical cores | 24 | 32 |
| Threads | 48 | 64 |
| Chip lithography | 14 nm | 5 nm |
| Power consumption (TDP) | 205 Watt | 280 Watt |
Xeon Platinum 8168 has 36.6% lower power consumption.
EPYC 9354, on the other hand, has a 104.9% higher aggregate performance score, an age advantage of 5 years, 33.3% more physical cores and 33.3% more threads, and a 180% more advanced lithography process.
The EPYC 9354 is our recommended choice as it beats the Xeon Platinum 8168 in performance tests.
Should you still have questions on choice between Xeon Platinum 8168 and EPYC 9354, ask them in Comments section, and we shall answer.
Similar processor comparisons
We picked several similar comparisons of processors in the same market segment and performance relatively close to those reviewed on this page.
