AMD EPYC 9475F vs Intel Xeon 6780E Comparison

FAQ

Q: Which processor has the higher average benchmark score?

A: The AMD EPYC 9475F records an average benchmark score of 350,933, while the Intel Xeon 6780E scores 280,438. The EPYC sits at the 100th percentile of all CPUs in the database, whereas the Xeon is at the 99th percentile.

Q: How do the two chips compare in multithreaded workload performance?

A: The AMD EPYC 9475F leads in the PassMark multithread test with a score of 122,476 versus 86,734 for the Intel Xeon 6780E, a margin of 41.2%. However, the Xeon wins in data compression, scoring 2,557,582 against 2,156,305, a 15.7% advantage for Intel.

Q: What are the core and thread counts of each processor?

A: The AMD EPYC 9475F has 48 cores and 96 threads, while the Intel Xeon 6780E has 144 cores and 144 threads. The Xeon has three times the core count, but the EPYC has simultaneous multithreading that the Xeon lacks.

Q: Which chip offers higher memory bandwidth?

A: The AMD EPYC 9475F provides 576.0 GB/s of memory bandwidth through its twelve-channel DDR5 memory bus. The Intel Xeon 6780E offers 409.6 GB/s via an eight-channel DDR5 configuration.

Q: How does single-thread performance compare?

A: The AMD EPYC 9475F is substantially ahead in the PassMark single-thread test, scoring 3,779 versus 1,923 for the Intel Xeon 6780E, a 96.5% advantage. This reflects the EPYC's much higher boost clock of 4.80 GHz compared to 3.00 GHz on the Xeon.

Q: What are the launch MSRP values for these processors?

A: The AMD EPYC 9475F has a launch MSRP of $7592. The Intel Xeon 6780E has a launch MSRP of $11350.

Architecture Differences

The AMD EPYC 9475F is built on the Zen 5 architecture with the codename Turin, part of the EPYC 9005 series. It uses a 4 nm process node from TSMC, with 66,520 million transistors distributed across a die size of 8x 70.6 mm². The Intel Xeon 6780E uses the Sierra Forest architecture, part of the Xeon 6 (Sierra Forest-SP) generation. It is fabricated on Intel's 5 nm process with a monolithic die of 578 mm².

The cache hierarchies differ substantially. The EPYC 9475F provides 80 KB of L1 cache per core, 1 MB of L2 cache per core, and a large 256 MB shared L3 cache. The Xeon 6780E has 96 KB of L1 per core, 4 MB of L2 per module, and 108 MB of shared L3. While the Xeon has more L1 per core, the EPYC's 256 MB L3 is more than double the Xeon's shared L3 capacity.

Memory architecture also diverges. The EPYC 9475F supports twelve-channel DDR5 memory with 576.0 GB/s bandwidth. The Xeon 6780E uses eight-channel DDR5 with 409.6 GB/s. Both support ECC memory. PCIe connectivity favors AMD: the EPYC provides Gen 5 with 128 lanes (CPU only), while the Xeon provides Gen 5 with 88 lanes (CPU only).

The EPYC 9475F operates at a base clock of 3.65 GHz and boosts to 4.80 GHz, with a TDP of 400 W. The Xeon 6780E runs at 2.20 GHz base and 3.00 GHz boost, with a TDP of 330 W. The AMD part has 48 cores and 96 threads; the Intel part has 144 cores and 144 threads. Neither processor has integrated graphics, and both are locked multipliers.

The socket platforms are incompatible: AMD uses Socket SP5, Intel uses Socket 4710. The EPYC 9475F was released on 2024-10-09, while the Xeon 6780E was released on 2024-06-02. The EPYC's part number is 100-000001143, and the Xeon's is SRPG3.

Head-to-Head Benchmarks

The benchmark results show a split profile between the two processors, with the AMD EPYC 9475F winning 6 of 11 head-to-head tests and the Intel Xeon 6780E winning 5.

The EPYC 9475F's largest win comes in the PassMark find prime numbers test, where it scores 1,507 against 708, a 112.9% advantage. This workload is highly sensitive to per-core performance and clock speed, and the EPYC's 4.80 GHz boost clock dominates the Xeon's 3.00 GHz. A similar pattern appears in single-thread performance, where the EPYC scores 3,779 versus 1,923, a 96.5% lead.

Extended instruction performance also favors AMD heavily. The EPYC 9475F scores 173,169 in PassMark extended instructions, while the Xeon 6780E scores 114,008, a 51.9% advantage. This suggests the Zen 5 cores handle advanced instruction sets more efficiently than Sierra Forest's efficiency-oriented cores.

The multithread test shows the EPYC ahead by 41.2%, scoring 122,476 versus 86,734. This is notable because the Xeon has 144 cores versus 48 on the EPYC, yet the EPYC's higher clock speeds and simultaneous multithreading carry the workload. The physics test follows a similar pattern: the EPYC scores 16,443 against 10,951, a 50.2% lead.

Intel's wins are concentrated in throughput-heavy integer operations. The Xeon 6780E leads in data compression with 2,557,582 versus 2,156,305, a 15.7% margin. The data encryption test shows an even larger Intel advantage: 193,004 versus 116,648, a 39.6% lead. Random string sorting also goes to Intel, with 326,954 versus 253,936, a 22.3% margin.

The floating-point math test is close, with Intel ahead at 433,862 versus 406,524, a 6.3% advantage. Integer math also favors Intel, 641,817 versus 605,696, a 5.6% lead. These results suggest the Xeon's raw core count provides an edge in highly parallel integer workloads that can scale across 144 threads.

The overall picture from the head-to-head data is that the EPYC 9475F wins in latency-sensitive and single-thread-dominated tasks, while the Xeon 6780E wins in bulk data processing that can saturate its many cores. The average benchmark scores reflect this: the EPYC's 350,933 average is 25.1% higher than the Xeon's 280,438.

Looking at the nearest rivals in the database, the EPYC 9475F sits close to other high-end EPYC parts. It trails the AMD EPYC 9754 by 3.7%, the Intel Xeon 6960P by 3.9%, the AMD EPYC 9655 by 6%, and the AMD EPYC 9535 by 7.5%. The Xeon 6780E, meanwhile, is nearly matched with the AMD Ryzen Threadripper 9970X, which is only 0.2% behind, while the EPYC 9565 is 1.8% behind, the Xeon 696X is 2% behind, and the EPYC 9555P is 2.3% behind.

Specification Differences

The two processors differ across nearly every major specification. The EPYC 9475F has 48 cores and 96 threads; the Xeon 6780E has 144 cores and 144 threads. The EPYC's base clock is 3.65 GHz versus 2.20 GHz, and its boost clock is 4.80 GHz versus 3.00 GHz. TDP is 400 W for the EPYC and 330 W for the Xeon.

The cache layouts are different in structure as well as size. The EPYC uses 80 KB L1 per core and 1 MB L2 per core, with 256 MB shared L3. The Xeon uses 96 KB L1 per core and 4 MB L2 per module, with 108 MB shared L3. The EPYC has a 4 nm process from TSMC with 66,520 million transistors on an 8x 70.6 mm² die; the Xeon uses a 5 nm Intel process on a 578 mm² die.

Memory channels differ: twelve for the EPYC versus eight for the Xeon. Memory bandwidth follows, with 576.0 GB/s for AMD and 409.6 GB/s for Intel. PCIe lanes also differ: 128 Gen 5 lanes for the EPYC versus 88 Gen 5 lanes for the Xeon.

The sockets are entirely different: AMD Socket SP5 versus Intel Socket 4710. Release dates differ by about four months, with the Xeon launching 2024-06-02 and the EPYC launching 2024-10-09. The launch MSRP values are $7592 for the EPYC and $11350 for the Xeon. Both are active production parts, both lack integrated graphics, and both have locked multipliers.

The Verdict

The data paints a clear picture of two different design philosophies. The AMD EPYC 9475F is a high-clock, high-frequency processor built for workloads that reward fast single-core execution, deep cache, and memory bandwidth. The Intel Xeon 6780E is a massive-core processor built for throughput-oriented tasks that can scale across 144 threads.

For single-thread and latency-sensitive workloads, the EPYC 9475F is the stronger choice. Its 96.5% lead in single-thread performance and 112.9% lead in prime number finding indicate that applications with serial dependencies or branch-heavy code will run dramatically faster on the AMD part. The 51.9% advantage in extended instructions reinforces this, as does the 50.2% lead in physics simulation.

For bulk data processing, the Xeon 6780E holds advantages in specific workloads. Data encryption shows a 39.6% lead for Intel, and data compression shows a 15.7% lead. Random string sorting is also 22.3% faster on the Xeon. These are workloads that can use the Xeon's 144 cores effectively and are less sensitive to per-core clock speed.

The multithread result is the most revealing data point. Despite having only 48 cores against 144, the EPYC 9475F leads the multithread test by 41.2%. This suggests that the EPYC's higher clocks and simultaneous multithreading more than compensate for the core count disadvantage in this particular benchmark. The Xeon's 144 threads cannot overcome the EPYC's per-thread efficiency.

The average benchmark score favors the EPYC 9475F by 25.1%, and the EPYC sits at the 100th percentile of all CPUs in the database. The Xeon 6780E sits at the 99th percentile. The EPYC's nearest rivals in the database are all within 7.5%, while the Xeon's nearest rivals are within 2.3%, indicating the Xeon is more tightly grouped with other high-core-count parts.

Buyers should choose based on workload characteristics. The EPYC 9475F is the pick for applications that emphasize single-thread speed, low latency, high memory bandwidth, and advanced instruction throughput. The Xeon 6780E is the pick for workloads that are purely throughput-oriented in integer-heavy, highly parallel data operations such as encryption, compression, and sorting. The EPYC's overall average benchmark score and percentile ranking make it the stronger general-purpose processor in the database's measurements.