AMD EPYC 7663 vs Intel Xeon 6737P Comparison

Head-to-Head Benchmarks

The recorded benchmark data shows a clear split between the AMD EPYC 7663 and the Intel Xeon 6737P, with the AMD part taking 10 of 16 head-to-head wins and the Intel part taking 6. The margin distribution, however, tells a more nuanced story. The AMD EPYC 7663 dominates in several high-magnitude workloads, while the Intel Xeon 6737P counters with a smaller but consistent set of single-threaded and specialized wins.

The largest delta in the entire comparison belongs to the AMD EPYC 7663 in PassMark data encryption, where it scores 111,725 against the Intel part’s 65,615, a 70.3% advantage. This is the single biggest performance gap recorded between the two processors. The AMD chip also shows a 41.5% lead in PassMark integer math, scoring 468,096 versus 330,756. Random string sorting favors AMD by 42.4%, with scores of 184,367 and 129,510 respectively. Data compression is another AMD win, 1,447,020 versus 1,157,255, a 25% margin.

The Intel Xeon 6737P’s largest victory comes in PassMark extended instructions, where it scores 105,453 against the AMD chip’s 73,719, a 30.1% lead. The Intel part also wins PassMark single-thread and single-threaded tests by 14.5%, scoring 3,048 versus 2,606 in both cases. PassMark physics shows a 14.3% Intel advantage, 9,362 versus 8,020. The remaining Intel wins are smaller: floating point math at 2.8% (258,811 versus 251,535) and find prime numbers at 3% (697 versus 676).

In the Cinebench suite, the AMD EPYC 7663 wins all five tests, but by narrow margins. Cinebench R15 multicore shows 7,032 versus 6,822, a 3.1% lead. The single-core R15 result is 992 versus 963, a 3% margin. Cinebench R20 multicore is 29,304 versus 28,428, again 3.1%. R20 single-core is 4,137 versus 4,013, also 3.1%. R23 multicore is 69,773 versus 67,688, a 3.1% lead. PassMark multithread follows the same pattern, 82,087 versus 79,634, a 3.1% margin.

Notably, the AMD EPYC 7663’s average benchmark score is 161,973, while the Intel Xeon 6737P’s is 140,694. Both parts sit at the 98th percentile among all CPUs in the database, but the AMD chip’s average is higher by roughly 15%. The nearest rivals for the AMD part include the AMD EPYC 9375F at 162,497 (0.3% ahead) and the AMD EPYC 9355P at 160,358 (1% behind). The Intel part’s nearest rivals are the Intel Xeon 674X at 143,103 (1.7% ahead), the Intel Xeon 6732P at 143,444 (1.9% ahead), and the AMD Ryzen 9 PRO 9965X3D at 143,735 (2.1% ahead).

FAQ

Q: Which processor has the higher single-threaded score?

A: The Intel Xeon 6737P wins both PassMark single-thread tests with a score of 3,048, versus 2,606 for the AMD EPYC 7663, a 14.5% advantage. The AMD chip still wins the Cinebench R15 and R20 single-core tests, but by only 3% and 3.1% respectively.

Q: What is the biggest performance difference between the two?

A: The largest recorded delta is in PassMark data encryption, where the AMD EPYC 7663 scores 111,725 versus the Intel Xeon 6737P’s 65,615, a 70.3% advantage for AMD.

Q: How do the two processors compare in multi-threaded workloads?

A: The AMD EPYC 7663 wins all multi-threaded Cinebench tests (R15, R20, R23) by 3.1% each, and wins PassMark multithread by 3.1% as well. Its 56 cores and 112 threads outnumber the Intel part’s 32 cores and 64 threads, which is reflected in the integer math and data compression margins.

Q: Does the Intel Xeon 6737P win any benchmark by a wide margin?

A: Yes, the Intel part leads by 30.1% in PassMark extended instructions, scoring 105,453 versus 73,719. It also leads by 14.5% in single-threaded PassMark tests and by 14.3% in PassMark physics.

Q: Which processor has the higher average benchmark score?

A: The AMD EPYC 7663 has an average benchmark score of 161,973, while the Intel Xeon 6737P has 140,694. Both are at the 98th percentile among all CPUs.

Q: Are there any benchmarks where the two are nearly tied?

A: PassMark find prime numbers is the closest test, with the Intel part scoring 697 versus 676, a 3% difference. PassMark floating point math is also close, 258,811 versus 251,535, a 2.8% margin.

Architecture Differences

The AMD EPYC 7663 is built on the Zen 3 architecture, codenamed Milan, part of the EPYC 7003 series. It uses a 7 nm process from TSMC and packs 33,200 million transistors across 8 dies, each 81 mm² in size. The Intel Xeon 6737P uses the Granite Rapids architecture, codenamed Granite Rapids, part of the Xeon 6 generation, on Intel’s 5 nm process with a single 598 mm² die.

Core counts differ substantially. The AMD part has 56 cores and 112 threads; the Intel part has 32 cores and 64 threads. Base clocks are 2.00 GHz for AMD and 2.90 GHz for Intel, with boost clocks of 3.50 GHz and 4.00 GHz respectively. The Intel part carries a higher TDP at 270 W versus 240 W for AMD.

Cache layouts are distinct. The AMD EPYC 7663 uses 64 KB L1 per core, 512 KB L2 per core, and 256 MB of shared L3. The Intel Xeon 6737P uses 112 KB L1 per core, 2 MB L2 per core, and 144 MB of shared L3. Memory support differs by generation: AMD uses DDR4 with eight-channel memory and 204.8 GB/s bandwidth; Intel uses DDR5 with eight-channel memory and 409.6 GB/s bandwidth. Both support ECC memory.

PCIe capabilities also differ. The AMD part provides Gen 4 with 128 lanes (CPU only), while the Intel part provides Gen 5 with 88 lanes (CPU only). The Intel part lists integrated graphics as N/A, while the AMD part has none listed. Sockets are incompatible: AMD Socket SP3 for the EPYC 7663, Intel Socket 4710 for the Xeon 6737P. Both are locked multipliers and active in production.

The Verdict

The data indicates that the AMD EPYC 7663 is the stronger overall performer, with a higher average benchmark score of 161,973 versus 140,694, and 10 head-to-head wins. Its advantages are most pronounced in encryption, integer math, random string sorting, and data compression, where it leads by 70.3%, 41.5%, 42.4%, and 25% respectively. These are large, workload-specific margins that suggest a strong fit for data-intensive server tasks.

The Intel Xeon 6737P, however, wins in single-threaded PassMark tests by 14.5%, in extended instructions by 30.1%, and in physics by 14.3%. Its higher base clock (2.90 GHz versus 2.00 GHz) and boost clock (4.00 GHz versus 3.50 GHz) contribute to these wins. The Intel part also delivers higher memory bandwidth, 409.6 GB/s versus 204.8 GB/s, and uses DDR5 rather than DDR4.

The Cinebench suite shows the AMD chip ahead by a consistent 3.1% across all multicore tests and 3% to 3.1% in single-core tests. This indicates that, despite having 24 fewer cores, the Intel part remains competitive in rendering-style workloads. For users prioritizing maximum core count and encryption/integer throughput, the AMD EPYC 7663 is the data-backed choice. For those needing faster single-threaded execution, extended instruction performance, or higher memory bandwidth, the Intel Xeon 6737P has the recorded advantages.

Specification Differences

The two processors differ in every major specification category. The AMD EPYC 7663 has 56 cores and 112 threads; the Intel Xeon 6737P has 32 cores and 64 threads. Base clocks are 2.00 GHz versus 2.90 GHz, and boost clocks are 3.50 GHz versus 4.00 GHz. TDP is 240 W for AMD and 270 W for Intel. Sockets differ: AMD Socket SP3 versus Intel Socket 4710.

Architecture and process differ: AMD uses Zen 3 (Milan) on a 7 nm TSMC process with 33,200 million transistors across 8x 81 mm² dies; Intel uses Granite Rapids on a 5 nm Intel process with a 598 mm² die. Cache structures differ: AMD has 64 KB L1 and 512 KB L2 per core, with 256 MB shared L3; Intel has 112 KB L1 and 2 MB L2 per core, with 144 MB shared L3. Memory support is DDR4 for AMD and DDR5 for Intel, both eight-channel, but bandwidth is 204.8 GB/s versus 409.6 GB/s. PCIe is Gen 4 with 128 lanes for AMD versus Gen 5 with 88 lanes for Intel. The Intel part lists integrated graphics as N/A; AMD lists none. Release dates are 2021-03-14 for AMD and 2025-02-23 for Intel. The launch MSRP for the AMD part is $6366, and for the Intel part it is $4995.

Where Each One Wins

The AMD EPYC 7663 wins in encryption workloads by a wide margin, with a 70.3% lead in PassMark data encryption. It also wins decisively in integer math (41.5%), random string sorting (42.4%), and data compression (25%). These results align with its higher core count and larger shared L3 cache, and they point toward database, compression, and cryptographic workloads as the AMD part’s strongest territory.

The Intel Xeon 6737P wins in extended instructions by 30.1%, indicating a strong advantage for workloads that utilize advanced instruction sets. It also wins single-threaded PassMark tests by 14.5% and physics by 14.3%, suggesting better per-core execution for lightly threaded or physics-heavy tasks. Floating point math and prime number finding are also Intel wins, but by smaller margins of 2.8% and 3%.

In rendering-style multi-threaded workloads, the AMD EPYC 7663 leads all Cinebench tests by 3.1%, but the Intel part remains close despite fewer cores. The Intel chip’s higher clocks and larger per-core L2 cache (2 MB versus 512 KB) help narrow the gap. For memory-bandwidth-sensitive applications, the Intel part’s 409.6 GB/s and DDR5 support give it a structural advantage, though the recorded benchmarks do not directly measure memory bandwidth as a win condition.

The win count is 10 for AMD and 6 for Intel, but the average benchmark scores differ by about 15% in favor of AMD. The Intel part holds the 98th percentile ranking equally, and its nearest rivals are all within 2.5% of its average score, indicating a tightly competitive field. The AMD part’s nearest rivals include two other AMD EPYC parts within 1% of its average, suggesting that the 7663 sits in a dense performance cluster at the top of the database.