AMD EPYC 9015 vs AMD Ryzen AI 9 HX PRO 470 Comparison
AMD EPYC 9015
Ryzen AI 9 HX PRO 470
PERFORMANCE BENCHMARKS
Analysis: AMD EPYC 9015 vs AMD Ryzen AI 9 HX PRO 470
The AMD EPYC 9015 and AMD Ryzen AI 9 HX PRO 470 represent two radically different interpretations of the Zen 5 architecture, and the benchmark data shows a clear split: the mobile Ryzen part wins the majority of workloads, while the server EPYC dominates in specific, narrow tasks. Despite the EPYC 9015’s higher average benchmark score of 57,555 versus the Ryzen’s 56,306, the head-to-head results tell a more nuanced story, with the Ryzen AI 9 HX PRO 470 taking 9 of 11 tests, often by substantial margins.
FAQ
Q: Which processor has the higher single-threaded performance?
A: The AMD Ryzen AI 9 HX PRO 470 is decisively ahead, scoring 4,050 in PassMark single-thread tests compared to the EPYC 9015’s 3,265. This represents a 19.4% advantage for the mobile chip, driven by its higher boost clock of 5.20 GHz versus the EPYC’s 4.10 GHz.
Q: How do the two chips compare in multi-threaded workloads?
A: The Ryzen AI 9 HX PRO 470 wins the PassMark multi-thread test with 36,402 points against the EPYC 9015’s 30,689, a 15.7% gap. This is notable because the EPYC has fewer cores (8) and threads (16) than the Ryzen’s 12 cores and 24 threads, though the EPYC does win the prime number finding test by a massive 123.5%.
Q: What is the biggest performance disparity between the two?
A: The largest difference is in the PassMark find prime numbers test, where the EPYC 9015 scores 304 versus the Ryzen’s 136, giving the server chip a 123.5% advantage. This is the only test where the EPYC wins by a triple-digit percentage.
Q: Which processor has faster memory bandwidth?
A: The EPYC 9015 is in a different class entirely, featuring a twelve-channel memory bus with 576.0 GB/s bandwidth. The Ryzen AI 9 HX PRO 470 uses a dual-channel bus with 89.6 GB/s, making the EPYC’s memory subsystem approximately 6.4 times wider on paper.
Q: Are both processors built on the same manufacturing process?
A: Yes, both the EPYC 9015 and the Ryzen AI 9 HX PRO 470 are fabricated on a 4 nm process at TSMC. However, the EPYC 9015 uses a chiplet design with a die size of 2x 70.6 mm², while the Ryzen uses a monolithic 233 mm² die.
Q: Do both processors support ECC memory?
A: Yes, both the EPYC 9015 and the Ryzen AI 9 HX PRO 470 support ECC memory. This is expected for the server-class EPYC, but it is a notable feature for the mobile Ryzen part, which also supports both DDR5 and LPDDR5X memory types.
Architecture Differences
The fundamental architectural split between these two processors is defined by their target markets. The EPYC 9015 is a server/workstation part within the EPYC 9005 series, codenamed Turin, while the Ryzen AI 9 HX PRO 470 is a mobile processor from the Ryzen AI PRO 400 series, codenamed Gorgon Point. Both use the Zen 5 architecture and are built on TSMC’s 4 nm process, but their implementations diverge sharply.
The EPYC 9015 is a chiplet-based design, with a die size of 2x 70.6 mm² and a transistor count of 16,630 million. It features 8 cores and 16 threads, with a base clock of 3.60 GHz and a boost clock of 4.10 GHz. The Ryzen AI 9 HX PRO 470, in contrast, uses a monolithic 233 mm² die and packs 12 cores and 24 threads, with a much lower base clock of 2.00 GHz but a significantly higher boost clock of 5.20 GHz. The Ryzen’s generation also notes a hybrid design of Zen 5 / Zen 5c cores, whereas the EPYC is solely Zen 5.
Cache configurations highlight another major divergence. Both share the same per-core L1 (80 KB) and L2 (1 MB) caches, but the EPYC 9015 offers a massive 64 MB of shared L3 cache, while the Ryzen AI 9 HX PRO 470 provides only 16 MB of L3. This 4x difference in L3 capacity is a key server advantage for handling large datasets. The EPYC also has a far more extensive I/O subsystem, with 128 PCIe Gen 5 lanes versus the Ryzen’s 16 PCIe Gen 4 lanes. Memory support follows the same pattern: the EPYC uses DDR5 over a twelve-channel bus with 576.0 GB/s bandwidth, while the Ryzen uses dual-channel DDR5 or LPDDR5X with 89.6 GB/s. The Ryzen includes integrated Radeon 890M graphics, while the EPYC has no integrated GPU, and the EPYC’s TDP of 125 watts dwarfs the Ryzen’s 28 watts.
Head-to-Head Benchmarks
The head-to-head PassMark data reveals a pattern of consistent, often large, wins for the Ryzen AI 9 HX PRO 470 in most compute tasks, but with two notable exceptions where the EPYC 9015 asserts its server-class strengths.
The Ryzen AI 9 HX PRO 470’s largest victory comes in integer math, scoring 127,573 against the EPYC’s 92,524, a 27.5% lead. Data compression follows closely, with the Ryzen at 461,195 versus 353,014 for the EPYC, a 23.5% difference. Floating-point math shows a 23% advantage for the Ryzen (80,023 vs. 61,615), and data encryption shows a 22.4% edge (22,929 vs. 17,790). The Ryzen also wins the extended instructions test by 14.7% (32,803 vs. 27,970) and random string sorting by 18.5% (48,824 vs. 39,772). In the multi-thread test, the Ryzen’s 36,402 score beats the EPYC’s 30,689 by 15.7%, and the single-thread test shows a 19.4% gap in favor of the Ryzen (4,050 vs. 3,265).
The EPYC 9015’s wins are narrow in count but dramatic in magnitude. In the find prime numbers test, the EPYC scores 304 versus the Ryzen’s 136, an extraordinary 123.5% advantage that reflects the server chip’s specialized integer throughput in this specific workload. The EPYC also wins the physics test, scoring 2,893 versus 1,915, a 51.1% lead. These two wins are significant but isolated; they do not compensate for the Ryzen’s broad dominance across the other nine tests.
Specification Differences
The specification sheets for these two processors read as a study in contrasting priorities. The EPYC 9015 has 8 cores and 16 threads, while the Ryzen AI 9 HX PRO 470 has 12 cores and 24 threads. Base clocks differ substantially: the EPYC runs at 3.60 GHz, while the Ryzen idles at 2.00 GHz. Boost clocks invert the relationship, with the EPYC reaching 4.10 GHz and the Ryzen surging to 5.20 GHz. Thermal design power is heavily skewed, with the EPYC rated at 125 watts and the Ryzen at 28 watts.
The cache hierarchy differs only in L3 capacity: the EPYC offers 64 MB shared, while the Ryzen provides 16 MB. Memory support is a differentiator, with the EPYC limited to DDR5 over twelve channels (576.0 GB/s bandwidth) and the Ryzen supporting both DDR5 and LPDDR5X over two channels (89.6 GB/s). PCIe connectivity is also divergent: the EPYC has 128 Gen 5 lanes, while the Ryzen has 16 Gen 4 lanes. The EPYC has no integrated graphics, but the Ryzen includes Radeon 890M. The EPYC uses the AMD Socket SP5, while the Ryzen uses AMD Socket FP8. The EPYC launched on 2024-10-09 with a launch MSRP of $527, while the Ryzen launched on 2026-01-04 with no listed MSRP.
The Verdict
The data directs a clear verdict: the AMD Ryzen AI 9 HX PRO 470 is the superior general-purpose processor, winning 9 of 11 head-to-head benchmarks, including all major math, compression, and encryption tests. Its 19.4% single-thread lead and 15.7% multi-thread lead make it the better choice for most workloads, from everyday productivity to demanding content creation. The Ryzen’s higher core count (12 vs. 8) and higher boost clock (5.20 GHz vs. 4.10 GHz) translate directly into the benchmark results, and its 92nd-percentile ranking among all CPUs matches the EPYC’s percentile despite a lower average score.
However, the AMD EPYC 9015 is not without a clear purpose. Its 123.5% victory in prime number finding and 51.1% win in physics indicate specialized strengths, likely derived from its massive 64 MB L3 cache and higher base clock. For workloads that stress these specific patterns, the EPYC is significantly faster. Additionally, the EPYC’s twelve-channel memory bus with 576.0 GB/s bandwidth and 128 PCIe Gen 5 lanes are class-defining features for server environments, even if they do not appear in the PassMark suite. The EPYC’s launch MSRP of $527 also positions it as a more expensive part than the unlisted mobile chip.
Choose the Ryzen AI 9 HX PRO 470 for most computing tasks. Choose the EPYC 9015 for its server-class memory bandwidth, PCIe lanes, and specific compute patterns where it excels.
Where Each One Wins
The AMD Ryzen AI 9 HX PRO 470 is the clear winner in data-centric workloads. It leads by 27.5% in integer math, 23% in floating-point math, and 23.5% in data compression, making it the better choice for code compilation, spreadsheet calculations, and file archiving. Its 22.4% lead in data encryption and 14.7% lead in extended instructions further solidify its position for security software and modern instruction set usage. For general multitasking, the Ryzen’s 15.7% multi-thread advantage and 18.5% random string sorting win make it preferable for running multiple applications or sorting large datasets. The 19.4% single-thread advantage also makes it the winner for legacy single-threaded applications and responsive everyday use.
The AMD EPYC 9015 wins in two specific, compute-intensive areas. The find prime numbers test, where it leads by 123.5%, suggests a strong advantage in prime-number generation and related number theory applications, potentially benefiting cryptography or scientific computing. The physics test, with a 51.1% lead, indicates superior performance in physics simulation workloads that rely on the EPYC’s higher base clock and larger cache. Beyond the PassMark suite, the EPYC’s 576.0 GB/s memory bandwidth and 128 PCIe Gen 5 lanes make it the only choice for memory-bandwidth-bound server tasks or systems requiring massive I/O expansion, even though these are not directly measured in the benchmark data.