AMD EPYC 9335 vs Intel Xeon 6747P Comparison

The Intel Xeon 6747P and AMD EPYC 9335 are both 99th-percentile server processors, but they are built on fundamentally different philosophies. The Xeon 6747P is a core-count monster with 48 cores and 96 threads, while the EPYC 9335 relies on higher clocks and a more efficient 32-core/64-thread design. Across the eleven head-to-head benchmark comparisons, the Intel part wins every single test, often by massive margins. However, the AMD chip is not without its own merits, primarily in platform efficiency and raw per-core clock speed, which the data reveals through its specification sheet. The benchmark data is decisive, but the choice between them depends on whether raw throughput or platform balance is the priority.

Where Each One Wins

The Xeon 6747P wins in every single workload category measured in the head-to-head comparisons, making it the clear choice for compute-heavy, throughput-bound tasks. Its most dominant victory comes in the passmark_physics test, where it scores 13,398 against the EPYC’s 1,905, a staggering 603.3% advantage. This suggests the Intel part is exceptionally well-suited for simulation, scientific computing, and any workload that relies heavily on physics calculations. Similarly, in passmark_find_prime_numbers, the Xeon scores 1,151 versus 340, a 238.5% lead, indicating superior integer processing capabilities for cryptographic and mathematical workloads.

The AMD EPYC 9335 does not win a single benchmark in the head-to-head set, but its advantage lies in the specification sheet. It has a higher base clock of 3.00 GHz and a higher boost clock of 4.40 GHz, compared to the Intel’s 2.70 GHz and 3.90 GHz respectively. This means that for lightly threaded applications or tasks that are sensitive to single-thread latency, the EPYC 9335 has the potential to feel snappier, even if the aggregate multi-threaded scores do not reflect it. The data shows it is not a performance winner, but it is a power and efficiency winner, consuming significantly less energy while still delivering competitive results in encryption and compression tasks where the Intel lead is smaller.

FAQ

Q: Is the Intel Xeon 6747P always faster than the AMD EPYC 9335?

A: Yes, based on the eleven head-to-head benchmarks, the Intel Xeon 6747P wins all eleven tests, including data compression, encryption, integer math, and single-thread performance. The AMD EPYC 9335 does not win a single comparison.

Q: How much faster is the Intel Xeon 6747P in multi-threaded workloads?

A: In the passmark_multithread test, the Intel Xeon 6747P scores 101,685, which is 54.5% higher than the AMD EPYC 9335’s score of 65,811. This indicates a significant advantage in heavily parallelized tasks.

Q: What is the biggest performance gap between the two processors?

A: The largest gap is in the passmark_physics test, where the Intel Xeon 6747P outperforms the AMD EPYC 9335 by 603.3%, scoring 13,398 versus 1,905. This suggests a massive difference in physics simulation capabilities.

Q: Does the AMD EPYC 9335 have any advantage in clock speed?

A: Yes, the AMD EPYC 9335 has a higher base clock of 3.00 GHz and a higher boost clock of 4.40 GHz, compared to the Intel Xeon 6747P’s 2.70 GHz base and 3.90 GHz boost. This could translate to better performance in single-threaded or lightly threaded applications.

Q: How do the two chips compare in memory bandwidth?

A: The AMD EPYC 9335 has a significant advantage with a twelve-channel memory bus and 576.0 GB/s of bandwidth, while the Intel Xeon 6747P has an eight-channel bus and 409.6 GB/s. This gives the AMD part a theoretical edge in memory-intensive workloads.

Q: Which processor was released more recently?

A: The Intel Xeon 6747P was released on 2025-02-23, while the AMD EPYC 9335 was released earlier on 2024-10-09. The Intel part is newer by several months.

Head-to-Head Benchmarks

The benchmark results are overwhelmingly in favor of the Intel Xeon 6747P, but the margins vary significantly across workload types, revealing the nature of each chip’s strengths. The most extreme difference is in passmark_physics, where the Intel part scores 13,398 against the AMD’s 1,905, a delta of 603.3%. This is not a marginal improvement; it is a generational leap in physics processing capability, likely driven by the Intel’s 48 cores and large 288 MB L3 cache. The next largest gap is in passmark_find_prime_numbers, where the Intel scores 1,151 versus the AMD’s 340, a 238.5% lead. This test is notoriously dependent on integer throughput and cache latency, both of which favor the Intel design.

In more common server workloads, the Intel lead is still substantial but less extreme. In passmark_data_compression, the Intel scores 1,833,378 versus 1,203,096, a 52.4% advantage, showing strong performance in database and file-system tasks. The passmark_floating_point_math test shows a 60.4% lead for Intel (365,904 vs 228,123), indicating better scientific and engineering compute capability. Similarly, passmark_integer_math shows a 35.3% advantage (468,518 vs 346,291), and passmark_extended_instructions shows a 34.9% lead (142,557 vs 105,706), meaning the Intel part is also better suited for encryption and vectorized workloads.

Even in the single-threaded test, where the AMD’s higher clock speed should theoretically help, the Intel Xeon 6747P still wins. The Intel scores 3,236 in passmark_single_thread versus the AMD’s 2,732, an 18.4% advantage. This is surprising given the AMD’s 4.40 GHz boost clock, but it suggests that the Intel’s architecture is more efficient per clock in this specific test. The smallest margin is also the single-thread test, which indicates that while Intel wins everywhere, the gap narrows when the workload is not fully parallelized.

Specification Differences

The core and thread counts are the most obvious differentiators. The Intel Xeon 6747P is equipped with 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, which directly explains its dominance in multi-threaded benchmarks. Clock speeds tell the opposite story: the AMD EPYC 9335 has a base clock of 3.00 GHz and a boost clock of 4.40 GHz, while the Intel Xeon 6747P is rated at 2.70 GHz base and 3.90 GHz boost. The AMD chip is faster on paper per-core, but it cannot overcome the Intel’s core-count advantage.

Power consumption is a major point of divergence. The Intel Xeon 6747P has a TDP of 330 watts, while the AMD EPYC 9335 draws only 210 watts. This 120-watt difference is substantial in a dense server environment, potentially affecting cooling requirements and power delivery infrastructure. Memory support also differs: the AMD EPYC 9335 uses a twelve-channel memory bus with 576.0 GB/s of bandwidth, whereas the Intel Xeon 6747P uses an eight-channel bus with 409.6 GB/s. Both support DDR5 and ECC memory, but the AMD part offers more raw memory throughput. PCIe lanes also favor AMD, with the EPYC 9335 offering 128 Gen 5 lanes versus the Intel’s 88 Gen 5 lanes, making the AMD chip more flexible for high-density storage or GPU configurations.

Architecture Differences

The two processors are built on different manufacturing processes and microarchitectures. The Intel Xeon 6747P uses a 5 nm process node fabricated by Intel, with a die size of 2x 598 mm². It is based on the Granite Rapids architecture, specifically the Xeon 6 (Granite Rapids-SP) generation. The AMD EPYC 9335, in contrast, is built on a 4 nm process node at TSMC, with a four-chiplet design totaling 4x 70.6 mm² and 33,260 million transistors. The AMD part uses the Zen 5 architecture, codenamed Turin, part of the EPYC 9005 series.

Cache hierarchies are also structured differently. The Intel Xeon 6747P has a large 288 MB shared L3 cache, which is critical for its performance in data-heavy workloads. The AMD EPYC 9335 has a smaller 128 MB shared L3 cache, but it compensates with a higher per-core L1 cache of 80 KB per core versus the Intel’s 112 KB per core, and a 1 MB per-core L2 cache versus the Intel’s 2 MB per-core L2 cache. The Intel part’s massive L3 cache is likely a key factor in its benchmark dominance, especially in tests like data compression and integer math. The physical packaging also differs—Intel uses a monolithic-ish dual-die design, while AMD uses a chiplet approach with four separate dies, which can improve manufacturing yields but introduces inter-die communication overhead.

The Verdict

Based strictly on the benchmark data, the Intel Xeon 6747P is the superior processor for raw performance. It wins all eleven head-to-head tests, with margins ranging from 18.4% in single-thread to 603.3% in physics. If the workload is multi-threaded, compute-intensive, or data-heavy, the Intel part is the unequivocal choice. Its 48 cores and 288 MB of L3 cache deliver results that the AMD EPYC 9335 cannot match, making it ideal for database servers, scientific simulation, and virtualization hosts where every bit of throughput matters.

However, the AMD EPYC 9335 is the more balanced platform choice in several specific scenarios. Its 210-watt TDP is significantly lower than the Intel’s 330 watts, which can reduce total cost of ownership in large deployments where power and cooling are constrained. Its twelve-channel memory bus and 128 PCIe Gen 5 lanes provide more headroom for memory bandwidth and I/O expansion, making it a better fit for storage servers or systems with many GPUs. The higher boost clock of 4.40 GHz may also give it an edge in latency-sensitive, lightly threaded applications, even if the benchmarks do not reflect that advantage. For users who prioritize efficiency, memory bandwidth, or PCIe expandability over raw core count, the EPYC 9335 is a legitimate option. For everyone else, the data points directly to the Xeon 6747P.