AMD EPYC 9015 vs AMD Ryzen 5 PRO 9645 Comparison

AMD
AMD

AMD EPYC 9015

CORE STATE Turin
CORE SPECS 8 Cores / 16 Threads
CLOCK SPEED 3.6 Base / 4.1 GHz Turbo
CACHE 64 MB (shared)
MAX TDP 125W
ARCHITECTURE Zen 5
nm
PROCESS 4 nm
LAUNCH DATE 2024
VS
AMD
AMD

Ryzen 5 PRO 9645

CORE STATE Granite Ridge
CORE SPECS 6 Cores / 12 Threads
CLOCK SPEED 3.9 Base / 5.4 GHz Turbo
CACHE 32 MB (shared)
MAX TDP 65W
ARCHITECTURE Granite Ridge
nm
PROCESS 4 nm
LAUNCH DATE 2025

PERFORMANCE BENCHMARKS

passmark_data_compression
353,014
363,736
passmark_data_encryption
17,790
17,599
passmark_extended_instructions
27,970
29,705
passmark_find_prime_numbers
304
235
passmark_floating_point_math
61,615
62,720
passmark_integer_math
92,524
93,019
passmark_multithread
30,689
31,434
passmark_physics
2,893
1,945
passmark_random_string_sorting
39,772
38,410
passmark_single_thread
3,265
4,636
passmark_singlethread
3,265
4,636

Analysis: AMD EPYC 9015 vs AMD Ryzen 5 PRO 9645

The AMD Ryzen 5 PRO 9645 and the AMD EPYC 9015 are both active 4 nm parts from AMD’s 2024-2025 lineup, yet they target different corners of the server and workstation market. The Ryzen 5 PRO 9645 is a 6-core Granite Ridge part on Socket AM5, while the EPYC 9015 is an 8-core Turin part on Socket SP5. Benchmark data shows a clear split: the Ryzen 5 PRO dominates single-threaded and several math workloads, while the EPYC takes decisive wins in physics simulation and prime-number finding, with a 7-4 win tally favoring the Ryzen part. Both CPUs sit at the 92nd percentile among all CPUs, with average benchmark scores of 58,916 for the Ryzen 5 PRO and 57,555 for the EPYC 9015.

Head-to-Head Benchmarks

The most dramatic difference appears in single-thread performance. The Ryzen 5 PRO 9645 scores 4,636 in the PassMark single-thread test, a full 42% ahead of the EPYC 9015’s 3,265. This is the largest delta in the entire comparison, and it reflects the Ryzen part’s 5.40 GHz boost clock versus the EPYC’s 4.10 GHz. For workloads that rely on per-core responsiveness—database queries, latency-sensitive transactions, or lightly threaded application logic—this advantage is substantial. The Ryzen part also leads in extended instructions (29,705 vs 27,970, a 6.2% edge) and data compression (363,736 vs 353,014, a 3% edge), suggesting that its higher clock speed translates into real throughput gains even on AVX-512-style workloads.

The EPYC 9015, however, strikes back in two specific areas. Its physics score of 2,893 is 32.8% higher than the Ryzen’s 1,945, and its prime-number finding score of 304 is 22.7% better than 235. These are heavily parallel, integer-heavy workloads that benefit from the EPYC’s extra two cores and four threads (8 cores/16 threads versus 6 cores/12 threads). The EPYC also wins random string sorting, scoring 39,772 versus 38,410, a 3.4% margin, and data encryption, scoring 17,790 versus 17,599, a 1.1% edge. These results indicate that while the Ryzen’s clock speed dominates single-threaded tasks, the EPYC’s additional cores provide a meaningful advantage in workloads that scale linearly with core count.

The middle ground is closer than the extremes suggest. In floating-point math, the Ryzen 5 PRO wins 62,720 to 61,615, a modest 1.8% margin. In integer math, it wins 93,019 to 92,524, a razor-thin 0.5% edge. The multithread benchmark also goes to the Ryzen, 31,434 versus 30,689, a 2.4% advantage, which is surprising given the EPYC’s extra cores. This suggests that the Ryzen’s higher boost clock compensates for its lower core count in mixed-thread workloads, and that the EPYC’s 3.60 GHz base clock holds it back when all cores are active. The average benchmark scores reinforce this picture: the Ryzen 5 PRO averages 58,916 across all tests, while the EPYC averages 57,555, a 2.4% overall difference. The Ryzen’s nearest rivals include the AMD Ryzen AI 9 HX 470 at 58,826 (0.2% behind) and the Intel Xeon w5-2545 at 58,504 (0.7% behind), while the EPYC sits close to the AMD Ryzen 9 9900X at 57,498 (0.1% behind) and the Intel Core i9-14900 at 58,115 (1% ahead).

FAQ

Q: Which CPU has the higher single-thread performance?

A: The AMD Ryzen 5 PRO 9645 is decisively ahead, scoring 4,636 in the PassMark single-thread test versus 3,265 for the AMD EPYC 9015, a 42% advantage. This is the largest performance gap in the head-to-head comparison.

Q: Does the EPYC 9015 win any benchmarks?

A: Yes, the EPYC 9015 wins four of the eleven head-to-head tests: physics (2,893 vs 1,945, a 32.8% lead), find prime numbers (304 vs 235, a 22.7% lead), random string sorting (39,772 vs 38,410, a 3.4% lead), and data encryption (17,790 vs 17,599, a 1.1% lead).

Q: How do the average benchmark scores compare?

A: The Ryzen 5 PRO 9645 has an average benchmark score of 58,916, while the EPYC 9015 averages 57,555. This gives the Ryzen part a 2.4% overall advantage, despite the EPYC having more cores and threads.

Q: What is the core and thread configuration for each processor?

A: The Ryzen 5 PRO 9645 has 6 cores and 12 threads, while the EPYC 9015 has 8 cores and 16 threads. The EPYC’s two additional cores contribute to its wins in parallel workloads like physics and prime-number finding.

Q: Which CPU has a higher boost clock?

A: The Ryzen 5 PRO 9645 boosts to 5.40 GHz, compared to 4.10 GHz for the EPYC 9015. This clock advantage is the primary driver of the Ryzen’s 42% lead in single-thread performance.

Q: Are both CPUs in the same performance percentile?

A: Yes, both the Ryzen 5 PRO 9645 and the EPYC 9015 are at the 92nd percentile among all CPUs, indicating they are both high-end parts, but with different strengths in different workload types.

Architecture Differences

The two processors share a common foundation: both are built on TSMC’s 4 nm process node, and both use the Zen 5 architecture, with the Ryzen 5 PRO using the Granite Ridge codename and the EPYC using Turin. However, the physical implementation diverges significantly. The Ryzen 5 PRO 9645 is a single-die design with a die size of 70.6 mm² and 8,315 million transistors. The EPYC 9015 uses a dual-die configuration, with a total die size of 2x 70.6 mm² and 16,630 million transistors, exactly double the transistor count. This dual-die setup is typical for server parts, enabling higher core counts and memory bandwidth.

Cache hierarchies are also different. Both parts have 80 KB of L1 cache per core and 1 MB of L2 per core, but the L3 cache differs: the Ryzen 5 PRO has 32 MB shared L3, while the EPYC 9015 has 64 MB shared L3. The EPYC’s larger L3 cache, combined with two extra cores, helps it in workloads that benefit from larger data residency and more parallel execution units. Memory support also diverges. The Ryzen 5 PRO uses dual-channel DDR5 with 89.6 GB/s of bandwidth, while the EPYC 9015 uses twelve-channel DDR5 with 576.0 GB/s of bandwidth—a 6.4x difference. Both support ECC memory, but the EPYC’s massive bandwidth advantage makes it suited for memory-intensive server workloads like virtualization or large in-memory databases.

PCIe connectivity is another major differentiator. The Ryzen 5 PRO offers Gen 5 with 24 CPU-only lanes, while the EPYC 9015 offers Gen 5 with 128 CPU-only lanes. This fivefold difference in lane count, combined with the EPYC’s twelve-channel memory, positions it as a platform for dense storage arrays, high-speed networking, and multi-GPU configurations. The Ryzen 5 PRO, by contrast, is more constrained, with a smaller memory bus and fewer PCIe lanes, but it compensates with a much higher boost clock and integrated Radeon Graphics. The EPYC has no integrated graphics. Both parts have locked multipliers and are marked as server/workstation segments, with the EPYC launching earlier in October 2024 and the Ryzen in September 2025.

The Verdict

The data points to two distinct use cases. The AMD Ryzen 5 PRO 9645 is the choice for single-threaded and lightly threaded performance. Its 42% lead in single-thread tests, along with wins in data compression, extended instructions, floating-point math, and multithread, make it ideal for workloads where clock speed matters more than core count—for example, front-end application servers, developer workstations, or real-time analytics that depend on low latency per query. The fact that it wins the multithread benchmark despite having two fewer cores suggests that its 5.40 GHz boost clock is highly effective in mixed-thread scenarios, and its average score of 58,916 places it ahead of rivals like the AMD Ryzen 7 9850X3D (58,386, 0.9% behind) and the Intel Xeon Platinum 8260M (58,323, 1% behind).

The AMD EPYC 9015 is the better pick for parallel, core-scalable workloads. Its 32.8% lead in physics and 22.7% lead in prime-number finding indicate that it excels in simulations, scientific computing, and encryption tasks that can use all 16 threads. The twelve-channel memory bus and 64 MB L3 cache provide a platform-level advantage that the Ryzen cannot match, and the 128 PCIe Gen 5 lanes offer far more expansion headroom. Its average score of 57,555 is close to the AMD Ryzen 9 9900X (57,498, 0.1% behind) and ahead of the Intel Core i9-14900 (58,115, 1% ahead of the EPYC), showing that it competes well with desktop and server parts alike. The EPYC’s launch MSRP is $527, while the Ryzen has no listed launch MSRP.

For a benchmark database perspective, the Ryzen 5 PRO 9645 is the higher-scoring part overall, with 7 wins to the EPYC’s 4. But the EPYC’s wins are in categories that matter for server-scale parallelism, and its memory and PCIe capabilities are not reflected in PassMark scores. Users with single-threaded or latency-sensitive workloads should choose the Ryzen 5 PRO 9645. Users running heavily threaded, memory-bandwidth-bound applications should choose the EPYC 9015. The 2.4% average score difference is small enough that platform considerations—socket type, memory channels, PCIe lanes—should be the deciding factor.

Specification Differences

| Specification | AMD Ryzen 5 PRO 9645 | AMD EPYC 9015 |

|---|---|---|

| Cores | 6 | 8 |

| Threads | 12 | 16 |

| Base Clock | 3.90 GHz | 3.60 GHz |

| Boost Clock | 5.40 GHz | 4.10 GHz |

| TDP | 65 W | 125 W |

| Socket | AMD Socket AM5 | AMD Socket SP5 |

| Codename | Granite Ridge | Turin |

| Die Size | 70.6 mm² | 2x 70.6 mm² |

| Transistors | 8,315 million | 16,630 million |

| L3 Cache | 32 MB (shared) | 64 MB (shared) |

| Memory Bus | Dual-channel | Twelve-channel |

| Memory Bandwidth | 89.6 GB/s | 576.0 GB/s |

| PCIe Lanes | Gen 5, 24 Lanes (CPU only) | Gen 5, 128 Lanes (CPU only) |

| Integrated Graphics | Radeon Graphics | N/A |

| Release Date | 2025-09-15 | 2024-10-09 |

| Launch MSRP | None listed | $527 |

| Part Number | 100-000001409 | 100-000001553 |

DETAILED SPECIFICATIONS

SPECIFICATION
EPYC 9015
5 PRO 9645
Core Specs
Cores
8
6 -25.0%
Threads
16
12 -25.0%
Base Clock (GHz)
3.6
3.9 +8.3%
Boost Clock (GHz)
4.1
5.4 +31.7%
Frequency (GHz)
3.6
3.9 +8.3%
Turbo Clock (GHz)
4.1
5.4 +31.7%
Multiplier
36
39 +8.3%
SMP CPUs
2
1 -50.0%
Cache
L1 Cache
80 KB (per core)
80 KB (per core)
L2 Cache
1 MB (per core)
1 MB (per core)
L3 Cache
64 MB (shared)
32 MB (shared)
Power
TDP (W)
125
65 -48.0%
PPT
88 W
Configurable TDP
120-155 W
Architecture
Architecture
Zen 5
Codename
Turin
Granite Ridge
Generation
EPYC (Zen 5 (Turin))
Ryzen 7 (Zen 5 (Granite Ridge))
Process Size
4 nm
4 nm
Transistors
16,630 million
8,315 million
Die Size
2x 70.6 mm²
70.6 mm²
Foundry
TSMC
TSMC
Memory
Memory Support
DDR5
DDR5
Memory Bus
Twelve-channel
Dual-channel
Memory Bandwidth
576.0 GB/s
89.6 GB/s
ECC Memory
Yes
Yes
Platform
Socket
AMD Socket SP5
AMD Socket AM5
Chipsets
X870E, X870, B850, B840, X670E, X670, B650E, B650, A620
PCIe
Gen 5, 128 Lanes(CPU only)
Gen 5, 24 Lanes(CPU only)
AMD Multi-Die
IO Process Size
6 nm
6 nm
Interconnect
CXL
Gen 2.0
Graphics
Integrated Graphics
Radeon Graphics
Other
Market
Server/Workstation
Server/Workstation
Production Status
Active
Active
Launch Price
$527
Part Number
100-000001553
100-000001409
Package
FC-LGA6096
FC-LGA1718
Tj Max
95°C
Bundled Cooler
Wraith Stealth
View EPYC 9015 Details View Ryzen 5 PRO 9645 Details