AMD EPYC 9554P vs Intel Xeon D-1746TER Comparison

Head-to-Head Benchmarks

The benchmark data presents a stark, one-sided picture. Across all six shared Cinebench tests, the AMD EPYC 9554P dominates the Intel Xeon D-1746TER with a consistent delta of roughly 572%. This is not a marginal lead; it is a categorical performance gap that defines the entire comparison.

Starting with multi-core workloads, the EPYC 9554P scores 9,020 in Cinebench R15 multi-core versus the Xeon's 1,341, a 572.6% advantage. The pattern repeats in Cinebench R20, where the AMD part hits 37,585 against 5,590, and in Cinebench R23, where it reaches 89,490 compared to 13,311. The deltas remain virtually identical at 572.4% and 572.3%, respectively. This consistency across rendering generations suggests the performance ratio is structural, not workload-specific.

Single-core results tell the same story, which is surprising for a server-focused comparison. The EPYC 9554P scores 1,273 in Cinebench R15 single-core versus 189 for the Xeon, a 573.5% lead. In R20, it is 5,305 against 789, and in R23, 12,633 versus 1,879. The fact that the single-core delta (573.5%) slightly exceeds the multi-core delta (572.6%) in R15 indicates the AMD architecture's per-thread efficiency is not merely a byproduct of core count.

The head-to-head table records six wins for the AMD EPYC 9554P and zero for the Intel Xeon D-1746TER. No benchmark in the shared set favors Intel, even by a small margin. This is notable because the Xeon D series targets dense, power-constrained edge deployments, yet the data shows no workload where its design philosophy yields a competitive score.

Architecture Differences

The underlying architectures explain the benchmark chasm. The AMD EPYC 9554P uses Zen 4 architecture on a 5 nm TSMC process, codenamed Genoa, with 64 cores and 128 threads. The Intel Xeon D-1746TER uses Ice Lake architecture on Intel's 10 nm process, codenamed Ice Lake-D, with 10 cores and 20 threads. That is a 6.4x core count difference, which alone accounts for most of the multi-core gap.

Cache hierarchies diverge sharply. The EPYC 9554P features 64 KB L1 and 1 MB L2 per core, plus a massive 256 MB shared L3 cache. The Xeon D-1746TER has 80 KB L1 and 1.25 MB L2 per core, but only 15 MB shared L3. The AMD part's L3 cache is over 17x larger, which directly impacts data-heavy server workloads and explains why the single-core results also skew heavily toward AMD.

Memory architecture reinforces the divide. The EPYC 9554P supports DDR5 on a twelve-channel bus, delivering 460.8 GB/s of memory bandwidth. The Xeon D-1746TER uses DDR4 on a triple-channel bus, providing 64.0 GB/s. That is a 7.2x bandwidth advantage for AMD. PCIe connectivity also differs: the EPYC offers Gen 5 with 128 lanes, while the Xeon provides Gen 4 with 16 lanes.

Physical specifications follow the same pattern. The EPYC 9554P has a 360 W TDP, uses AMD Socket SP5, and is built from 52,560 million transistors across 8x 72 mm² dies. The Xeon D-1746TER has a 67 W TDP and uses Intel BGA 2227, with no transistor or die size data provided. The EPYC's launch MSRP is $7104, while the Xeon's is $1069.

The Verdict

The data points to two entirely different market positions. The AMD EPYC 9554P is a high-core-count, high-bandwidth server processor designed for compute-dense workloads where every percentage point of throughput matters. The Intel Xeon D-1746TER is a low-power, compact BGA part aimed at edge or embedded scenarios where thermals and space constrain the design.

For pure performance, the verdict is unambiguous. The EPYC 9554P wins every benchmark by over 570%, and its average benchmark score of 21,899 sits at the 75th percentile of all CPUs. The Xeon D-1746TER also posts a 75th percentile ranking with an average score of 21,635, but that ranking is achieved through different benchmark mixes, including PassMark tests not shared between the two.

The data does not support recommending the Xeon for any benchmark where the EPYC is a viable option. However, the Xeon's 67 W TDP versus 360 W suggests a power efficiency story that the current benchmark set does not quantify. If power draw is the primary constraint, the Xeon's lower envelope is its only demonstrable advantage from the fact pack.

Specification Differences

| Specification | AMD EPYC 9554P | Intel Xeon D-1746TER |

|---|---|---|

| Cores | 64 | 10 |

| Threads | 128 | 20 |

| Base Clock | 3.10 GHz | 2000.00 MHz |

| Boost Clock | 3.75 GHz | 3.10 GHz |

| TDP | 360 W | 67 W |

| Socket | AMD Socket SP5 | Intel BGA 2227 |

| Architecture | Zen 4 | Ice Lake |

| Process Node | 5 nm | 10 nm |

| Foundry | TSMC | Intel |

| L1 Cache | 64 KB (per core) | 80 KB (per core) |

| L2 Cache | 1 MB (per core) | 1.25 MB (per core) |

| L3 Cache | 256 MB (shared) | 15 MB (shared) |

| Memory Support | DDR5 | DDR4 |

| Memory Bus | Twelve-channel | Triple-channel |

| Memory Bandwidth | 460.8 GB/s | 64.0 GB/s |

| PCIe | Gen 5, 128 Lanes | Gen 4, 16 Lanes |

| Launch MSRP | $7104 | $1069 |

FAQ

Q: Which CPU has more cores?

A: The AMD EPYC 9554P has 64 cores and 128 threads, while the Intel Xeon D-1746TER has 10 cores and 20 threads.

Q: What is the biggest performance difference in the shared benchmarks?

A: In Cinebench R15 single-core, the EPYC 9554P leads by 573.5%, scoring 1,273 versus 189 for the Xeon D-1746TER.

Q: How much memory bandwidth does each CPU support?

A: The EPYC 9554P supports 460.8 GB/s via twelve-channel DDR5, while the Xeon D-1746TER supports 64.0 GB/s via triple-channel DDR4.

Q: Do both CPUs support ECC memory?

A: Yes, both the AMD EPYC 9554P and the Intel Xeon D-1746TER have ECC memory support enabled.

Q: What is the TDP difference?

A: The EPYC 9554P has a TDP of 360 W, whereas the Xeon D-1746TER has a TDP of 67 W.

Q: Which CPU has a larger L3 cache?

A: The EPYC 9554P has 256 MB shared L3 cache, compared to 15 MB shared L3 on the Xeon D-1746TER.

Where Each One Wins

The AMD EPYC 9554P wins every shared benchmark category. Its multi-core scores in Cinebench R15 (9,020), R20 (37,585), and R23 (89,490) are all over 570% higher than the Xeon's corresponding 1,341, 5,590, and 13,311. This makes it the clear choice for rendering, compilation, and any parallel compute workload where the benchmark data applies.

The EPYC also wins single-core tests decisively, with R15 at 1,273 versus 189, R20 at 5,305 versus 789, and R23 at 12,633 versus 1,879. This suggests that even lightly threaded workloads benefit from the AMD architecture's higher clock speeds and larger cache footprint.

The Intel Xeon D-1746TER does not win any head-to-head benchmark, but it has structural advantages that the shared tests do not capture. Its 67 W TDP, compact BGA 2227 socket, and DDR4 memory support position it for low-power edge servers or network appliances where the EPYC's 360 W TDP and SP5 socket would be impractical. The Xeon's PassMark results, including 191,594 in data compression and 54,482 in integer math, indicate it can handle real-world server tasks, just at a different performance tier.

The data implies a use-case split based on power and form factor rather than benchmark scores. If the workload fits within the Xeon's thermal envelope and the performance level is acceptable, it could be the only viable option. If performance is the sole criterion, the EPYC 9554P wins without qualification.