CPU Comparison

The Intel Xeon 6737P and AMD EPYC 9275F are two high-end server processors aimed at demanding workloads, but they approach performance from very different design philosophies. The Intel part fields 32 cores and 64 threads, while the AMD EPYC counters with 24 cores and 48 threads, yet the benchmark data reveals a complex picture where each chip dominates distinct types of tasks. This analysis draws exclusively on the provided benchmark results, specifications, and nearest-rival comparisons to break down where each processor excels.

FAQ

Q: Which processor has more cores and threads?

A: The Intel Xeon 6737P has 32 cores and 64 threads, compared to the AMD EPYC 9275F’s 24 cores and 48 threads.

Q: How do the two chips compare in single-threaded performance?

A: The AMD EPYC 9275F is significantly ahead in single-threaded tests, scoring 3810 in PassMark single-thread versus the Intel’s 3048, a 20% advantage. In Cinebench R23 single-core, the AMD leads with 10154 versus Intel’s 67688 (note: this is the multi-core score; the R23 single-core value for Intel is not listed, but the R15 and R20 single-core deltas are both -5.9% favoring AMD).

Q: Which processor wins in multi-threaded benchmarks?

A: The AMD EPYC 9275F wins the majority of multi-threaded tests, including Cinebench R23 multi-core (71927 vs 67688) and PassMark multithread (84620 vs 79634), despite having fewer cores. However, the Intel chip wins in floating-point math and extended instructions.

Q: What is the difference in memory bandwidth?

A: The AMD EPYC 9275F supports twelve-channel memory with 576.0 GB/s bandwidth, while the Intel Xeon 6737P uses eight-channel memory with 409.6 GB/s.

Q: Which processor has more PCIe lanes?

A: The AMD EPYC 9275F provides 128 PCIe Gen 5 lanes, whereas the Intel Xeon 6737P offers 88 PCIe Gen 5 lanes (CPU only).

Q: How does the average benchmark score compare?

A: The Intel Xeon 6737P has an average benchmark score of 140694, which is 5.3% higher than the AMD EPYC 9275F’s 133174, according to the nearestRivals data.

Architecture Differences

The two processors are built on fundamentally different silicon. The Intel Xeon 6737P uses the Granite Rapids architecture, fabricated on Intel’s 5 nm process with a die size of 598 mm². It is part of the Xeon 6 (Granite Rapids-SP) generation. In contrast, the AMD EPYC 9275F employs the Zen 5 architecture (codenamed Turin), manufactured by TSMC on a 4 nm node, with a die composed of eight chiplets, each measuring 70.6 mm², totaling 66,520 million transistors.

Cache hierarchies diverge sharply. The Intel part has 112 KB of L1 cache per core, 2 MB of L2 per core, and a shared 144 MB L3 cache. The AMD EPYC 9275F features 80 KB L1 per core, 1 MB L2 per core, and a much larger shared 256 MB L3 cache. This larger L3 pool on the AMD chip likely contributes to its strong performance in data-compression and random-string-sorting workloads.

Memory subsystems also differ. Intel uses an eight-channel DDR5 interface with 409.6 GB/s bandwidth, while AMD employs a twelve-channel DDR5 setup delivering 576.0 GB/s. Both support ECC memory. PCIe connectivity favors AMD with 128 Gen 5 lanes versus Intel’s 88 Gen 5 lanes.

Clock speeds tell a story of trade-offs. The Intel Xeon 6737P has a base clock of 2.90 GHz and a boost of 4.00 GHz, while the AMD EPYC 9275F runs at a higher 4.10 GHz base and 4.80 GHz boost. This clock advantage helps explain AMD’s single-threaded wins. However, the Intel chip consumes less power with a 270 W TDP compared to AMD’s 320 W.

Head-to-Head Benchmarks

The benchmark suite reveals a clear split: AMD wins 12 of 16 head-to-head tests, but Intel’s 4 wins are decisive in specific domains. Starting with Cinebench, the AMD EPYC 9275F wins all three multi-core tests (R15, R20, R23) by a consistent 5.9% margin. For instance, in Cinebench R23 multi-core, AMD scores 71927 versus Intel’s 67688. Single-core Cinebench results also favor AMD, with R15 and R20 both showing a 5.9% delta in AMD’s favor.

PassMark tests paint a more nuanced picture. The AMD chip wins PassMark multithread (84620 vs 79634, a 5.9% edge), data compression (1212560 vs 1157255, a 4.6% edge), find prime numbers (991 vs 697, a massive 29.7% advantage), physics (12089 vs 9362, a 22.6% win), random string sorting (144037 vs 129510, a 10.1% edge), and single-thread (3810 vs 3048, a 20% victory).

Intel’s wins are concentrated in compute-heavy tasks. The Xeon 6737P dominates floating-point math with a score of 258811 versus AMD’s 201888, a 28.2% advantage. It also leads in extended instructions (105453 vs 94889, an 11.1% edge), data encryption (65615 vs 62664, a 4.7% win), and integer math (330756 vs 317777, a 4.1% margin). These results suggest Intel’s architecture is particularly strong in scientific computing and cryptographic workloads.

The overall win count (4 for Intel, 12 for AMD) might suggest a lopsided contest, but the magnitude of Intel’s floating-point victory is notable. A 28.2% lead in floating-point math is a significant differentiator for applications like simulation and rendering.

Specification Differences

The two processors differ across nearly every major specification field. Core and thread counts are the most obvious: Intel offers 32 cores and 64 threads, while AMD provides 24 cores and 48 threads. Clock speeds favor AMD, with a base of 4.10 GHz and boost of 4.80 GHz compared to Intel’s 2.90 GHz base and 4.00 GHz boost.

Power consumption is another key difference. The Intel Xeon 6737P has a TDP of 270 W, while the AMD EPYC 9275F draws 320 W. This means Intel achieves its performance with 50 W less thermal headroom.

Process technology differs by node and foundry. Intel uses its own 5 nm process, while AMD relies on TSMC’s 4 nm node. Die size is dramatically different: Intel’s monolithic die measures 598 mm², while AMD’s multi-chip design uses eight chiplets at 70.6 mm² each. AMD lists 66,520 million transistors, while Intel does not provide a transistor count.

Cache configurations are notably different. Intel has 112 KB L1 per core, 2 MB L2 per core, and 144 MB shared L3. AMD has 80 KB L1 per core, 1 MB L2 per core, and 256 MB shared L3. The larger L3 on AMD is a clear advantage for cache-sensitive workloads.

Memory support shows AMD with a wider interface: twelve-channel DDR5 versus Intel’s eight-channel, resulting in 576.0 GB/s versus 409.6 GB/s bandwidth. PCIe lanes also favor AMD at 128 Gen 5 lanes versus Intel’s 88 Gen 5 lanes. Both support ECC memory, and neither has integrated graphics.

Socket and platform are incompatible: Intel uses Socket 4710, while AMD uses Socket SP5. The Intel part released on 2025-02-23, while the AMD EPYC 9275F launched earlier on 2024-10-09. Both are currently in active production and are locked (no unlocked multiplier).

The Verdict

The data suggests a clear division of labor. The AMD EPYC 9275F is the better choice for general-purpose multi-threaded workloads, single-threaded responsiveness, and memory-bandwidth-intensive tasks. Its wins in Cinebench multi-core, PassMark multithread, data compression, and physics indicate strong all-around performance. The 20% single-thread lead (3810 vs 3048) makes it particularly attractive for applications that rely on per-core performance. The larger 256 MB L3 cache and higher memory bandwidth (576.0 GB/s) likely drive its advantages in data compression and string sorting.

The Intel Xeon 6737P, despite losing most head-to-head tests, is the superior option for floating-point-heavy and encryption-heavy workloads. Its 28.2% lead in floating-point math (258811 vs 201888) and 11.1% edge in extended instructions make it a compelling choice for scientific computing, financial modeling, or any task dominated by FPU operations. The 4.7% win in data encryption also suggests strength in security-focused applications.

For users prioritizing raw multi-threaded throughput with fewer cores, the EPYC 9275F delivers more per core. Its 24 cores outperform Intel’s 32 cores in most multi-threaded benchmarks, which is a testament to Zen 5’s efficiency. The higher TDP (320 W vs 270 W) is a trade-off for that performance.

Intel’s advantage in average benchmark score (140694 vs 133174, a 5.3% delta per the nearestRivals data) is interesting but does not translate into consistent head-to-head wins. This discrepancy suggests Intel’s wins are in specific high-score tests that skew the average.

In summary, pick the AMD EPYC 9275F for balanced server workloads, database operations, and virtualization where single-thread speed and memory bandwidth matter. Pick the Intel Xeon 6737P if your workload is dominated by floating-point math, extended instruction sets, or encryption, where Intel’s 28.2% and 11.1% advantages are decisive. The AMD chip wins more tests, but the Intel chip wins the tests that matter most for certain niches.