AMD EPYC 9335 vs Intel Xeon 6740P Comparison

The AMD EPYC 9335 and Intel Xeon 6740P are both recent, high-core-count server processors, but the benchmark data reveals a decisive performance gap. Across every measured head-to-head workload, the Intel Xeon 6740P posts a higher score. The AMD EPYC 9335 does not win a single comparison in the dataset, with the Intel part leading by margins ranging from a modest 8.2% in single-threaded tests to a massive 80.4% in the physics workload. This analysis breaks down the specific numerical differences, architectural reasons, and practical implications for each processor.

Head-to-Head Benchmarks

The most lopsided result is in the `passmark_physics` test, where the Intel Xeon 6740P scores 9,705 against the AMD EPYC 9335's 1,905. This translates to an 80.4% deficit for the AMD part, a staggering gap that suggests a fundamental difference in how the two processors handle the physics simulation workload. The Intel part is nearly five times faster in this specific test, which heavily favors its higher core count and potentially different memory subsystem behavior.

Another significant win for Intel comes in `passmark_find_prime_numbers`, with the Xeon 6740P scoring 822 versus 340 for the EPYC 9335. The 58.6% delta here is substantial, indicating that integer-heavy, latency-sensitive workloads with high thread counts benefit greatly from the Intel's 48 cores and 96 threads. The AMD part's 32 cores and 64 threads are clearly outmatched in this compute-bound scenario.

In the broader multithreaded test (`passmark_multithread`), the Intel Xeon 6740P scores 88,061, which is 25.3% higher than the EPYC 9335's 65,811. This aligns with the core count advantage, but the margin is smaller than the raw core ratio might suggest. The AMD's higher base clock of 3.00 GHz versus 2.10 GHz helps it narrow the gap, but the Intel's two extra cores per chiplet and higher total thread count still win out.

Memory and I/O intensive tasks also favor Intel, but by smaller margins. In `passmark_data_compression`, the Intel scores 1,537,129 versus 1,203,096, a 21.7% lead. The `passmark_random_string_sorting` test shows a 33.4% advantage for Intel (175,015 vs 116,608), which indicates better handling of pointer-chasing and memory bandwidth-bound operations. The `passmark_data_encryption` test gives Intel a 23% lead (82,028 vs 63,159), and `passmark_floating_point_math` shows a 23% lead as well (296,102 vs 228,123).

The closest competition is in the single-threaded tests. The Intel Xeon 6740P leads with a score of 2,975 versus 2,732 for the AMD EPYC 9335, a delta of only 8.2%. This shows that while Intel has the edge in raw per-core speed, the AMD's Zen 5 architecture with its higher 4.40 GHz boost clock keeps it competitive in lightly-threaded applications. The `passmark_extended_instructions` test also shows a relatively narrow 9.6% gap (116,992 vs 105,706), and `passmark_integer_math` is close at 10.9% (388,500 vs 346,291). Overall, the data shows a consistent, but not uniform, Intel victory across all 11 benchmark comparisons.

Architecture Differences

The two processors are built on fundamentally different designs. The AMD EPYC 9335 uses the Zen 5 architecture, codenamed "Turin," and is part of the EPYC 9005 series. It is fabricated on a 4 nm process at TSMC and uses a chiplet design with a die size listed as 4x 70.6 mm², containing 33,260 million transistors. In contrast, the Intel Xeon 6740P is based on the Granite Rapids architecture, part of the Xeon 6 family, and uses a 5 nm process at Intel with a monolithic die measuring 2x 598 mm². The difference in process node and die size suggests different manufacturing approaches, with AMD opting for smaller chiplets and Intel using larger, more complex dies.

Core counts differ significantly. The AMD part has 32 cores and 64 threads, while the Intel part has 48 cores and 96 threads. This 50% advantage in core count for Intel is the primary driver of its multithreaded performance lead. However, the AMD parts clock higher, with a base of 3.00 GHz and boost of 4.40 GHz, versus Intel's 2.10 GHz base and 3.80 GHz boost. This makes the AMD part more responsive in single-threaded scenarios, as evidenced by the closer single-thread benchmark scores.

Cache hierarchies are also distinct. The AMD EPYC 9335 provides 80 KB of L1 and 1 MB of L2 per core, with a shared 128 MB L3 cache. The Intel Xeon 6740P offers larger per-core caches at 112 KB L1 and 2 MB L2, and a much larger shared 288 MB L3 cache. This larger L3 cache on the Intel part likely contributes to its strong performance in data-heavy workloads like compression and sorting.

Memory support differs in configuration. Both support DDR5, but the AMD uses a twelve-channel memory bus with a peak bandwidth of 576.0 GB/s, while the Intel uses an eight-channel bus with 409.6 GB/s. Despite having a lower theoretical memory bandwidth, the Intel part still wins in memory-intensive tests, suggesting that its larger caches and memory controller efficiency compensate for the narrower bus. PCIe connectivity also varies, with the AMD providing 128 Gen 5 lanes (CPU only) versus Intel's 88 Gen 5 lanes. Both processors support ECC memory and lack integrated graphics, targeting server and workstation deployments.

FAQ

Q: Which processor has more cores and threads?

A: The Intel Xeon 6740P has 48 cores and 96 threads, while the AMD EPYC 9335 has 32 cores and 64 threads. This gives the Intel part a 50% advantage in core count.

Q: How do they compare in single-threaded performance?

A: The Intel Xeon 6740P leads in the `passmark_single_thread` test with a score of 2,975, which is 8.2% higher than the AMD EPYC 9335's score of 2,732. The gap is the narrowest among all tested workloads.

Q: What is the largest performance gap between them?

A: The largest gap is in the `passmark_physics` test, where the Intel Xeon 6740P scores 9,705 compared to 1,905 for the AMD EPYC 9335, a difference of 80.4%.

Q: Does the AMD EPYC 9335 win any benchmark in the head-to-head comparison?

A: No. In the provided head-to-head benchmark data, the Intel Xeon 6740P wins all 11 tests. The AMD EPYC 9335 has zero wins.

Q: Which processor has a higher base clock speed?

A: The AMD EPYC 9335 has a higher base clock at 3.00 GHz, while the Intel Xeon 6740P has a base clock of 2.10 GHz. The AMD also has a higher boost clock at 4.40 GHz versus 3.80 GHz.

Q: What are the memory bus configurations?

A: The AMD EPYC 9335 uses a twelve-channel memory bus, while the Intel Xeon 6740P uses an eight-channel bus. This gives the AMD part a higher theoretical memory bandwidth of 576.0 GB/s versus 409.6 GB/s.

Specification Differences

The two processors differ across most major specifications. The AMD EPYC 9335 uses 32 cores and 64 threads, while the Intel Xeon 6740P has 48 cores and 96 threads. The AMD's base clock is 3.00 GHz and boost is 4.40 GHz, while the Intel's base is 2.10 GHz and boost is 3.80 GHz. The thermal design power (TDP) also differs, with the AMD rated at 210 watts and the Intel at 270 watts.

The process node and foundry are different: the AMD is fabricated on a 4 nm TSMC process, while the Intel uses a 5 nm Intel process. The AMD's die size is listed as 4x 70.6 mm², and it has 33,260 million transistors, while the Intel's die size is 2x 598 mm² with no transistor count provided. The cache hierarchy is also distinct, with the AMD offering 80 KB L1, 1 MB L2, and 128 MB L3 per core, while the Intel offers 112 KB L1, 2 MB L2, and 288 MB L3.

Memory configuration differs: the AMD uses a twelve-channel bus with 576.0 GB/s bandwidth, while the Intel uses an eight-channel bus with 409.6 GB/s. PCIe lane count is also different, with the AMD providing 128 Gen 5 lanes (CPU only) and the Intel providing 88 Gen 5 lanes. The sockets are incompatible, as the AMD uses Socket SP5 and the Intel uses Socket 4710. The release dates are also different, with the AMD launching on 2024-10-09 and the Intel on 2025-02-23.

The Verdict

Based strictly on the benchmark data, the Intel Xeon 6740P is the superior processor for raw performance. It wins every single head-to-head test, from memory-heavy operations like data compression to compute-intensive tasks like finding prime numbers and physics simulations. The Intel part's 48 cores and 96 threads provide a substantial advantage in multithreaded workloads, and its higher per-core L2 and L3 cache sizes help it maintain the lead even in less parallel tasks. The data shows the Intel Xeon 6740P is faster in every measured category.

The AMD EPYC 9335's strengths are relative, not absolute. It has a higher clock speed (3.00 GHz base, 4.40 GHz boost) and a wider memory bus (twelve-channel, 576.0 GB/s), which narrows the gap in some tests but never closes it. The single-threaded benchmark shows the AMD part is closest to Intel, with an 8.2% deficit. For workloads that are extremely latency-sensitive and not fully parallel, the AMD could be considered competitive, but the benchmark data does not show a single scenario where it takes the lead.

For buyers who prioritize maximum throughput in multi-threaded, data-center scale workloads, the data points unambiguously to the Intel Xeon 6740P. The 80.4% lead in physics and 58.6% lead in prime number finding are decisive. The AMD EPYC 9335 is a capable processor, but the data indicates it is outclassed in this direct comparison. The Intel part also has a larger L3 cache (288 MB vs 128 MB), which likely explains its superior performance in memory-bound tasks. The verdict is clear: the Intel Xeon 6740P is the better performer in this matchup.

Where Each One Wins

The Intel Xeon 6740P wins in every benchmark category in the head-to-head data, so its "winning" use cases are all of them. It is particularly dominant in physics simulations (80.4% lead), prime number computation (58.6% lead), and random string sorting (33.4% lead). These are workloads that heavily utilize many cores and benefit from large amounts of cache memory. The Intel part's 288 MB L3 cache and 48 cores make it the clear choice for high-performance computing, database processing, and any application that can scale to 96 threads.

The AMD EPYC 9335 does not have a single benchmark win, but it has areas where it is less disadvantaged. Its strongest relative performance is in single-threaded tests, where it trails by only 8.2%, and in extended instructions, where the gap is 9.6%. This suggests that for legacy applications that are not well-parallelized, or for workloads that depend heavily on single-core speed, the AMD part is a viable alternative, albeit still slower. Its higher boost clock of 4.40 GHz helps in these scenarios.

The AMD's wider twelve-channel memory bus and higher theoretical bandwidth (576.0 GB/s) could be an advantage in memory-bandwidth-bound workloads, but the data from the memory-heavy tests (compression, encryption, sorting) does not support this. The Intel part wins those tests by double-digit margins. Therefore, the AMD EPYC 9335 is best suited for environments where its higher clock speeds and lower TDP (210 vs 270 watts) are more critical than raw multi-threaded performance, though the benchmark results show it cannot match the Intel Xeon 6740P in any measured performance metric.