AMD EPYC 9565 vs AMD EPYC 9754 Comparison

AMD
AMD

AMD EPYC 9565

CORE STATE Turin
CORE SPECS 72 Cores / 144 Threads
CLOCK SPEED 3.15 Base / 4.3 GHz Turbo
CACHE 384 MB (shared)
MAX TDP 400W
ARCHITECTURE Zen 5
nm
PROCESS 4 nm
LAUNCH DATE 2024
VS
AMD
AMD

EPYC 9754

CORE STATE Bergamo
CORE SPECS 128 Cores / 256 Threads
CLOCK SPEED 2.25 Base / 3.1 GHz Turbo
CACHE 256 MB (shared)
MAX TDP 360W
ARCHITECTURE Zen 4
nm
PROCESS 5 nm
LAUNCH DATE 2023

PERFORMANCE BENCHMARKS

cinebench_cinebench_r15_multicore
11,585
8,460
cinebench_cinebench_r15_singlecore
1,635
1,194
cinebench_cinebench_r20_multicore
48,273
35,254
cinebench_cinebench_r20_singlecore
6,814
4,977
cinebench_cinebench_r23_multicore
114,937
83,939
cinebench_cinebench_r23_singlecore
16,226
11,850
passmark_data_compression
2,579,631
3,558,043
passmark_data_encryption
141,936
231,891
passmark_extended_instructions
209,595
224,322
passmark_find_prime_numbers
2,422
604
passmark_floating_point_math
549,422
588,187
passmark_integer_math
717,948
1,026,896
passmark_multithread
135,221
98,752
passmark_physics
18,036
8,793
passmark_random_string_sorting
291,941
306,481
passmark_single_thread
3,696
2,328
passmark_singlethread
3,696
2,328

Analysis: AMD EPYC 9565 vs AMD EPYC 9754

The AMD EPYC 9754 and AMD EPYC 9565 represent two distinct approaches to high-core-count server processing. The 9754 is a 128-core Bergamo part built on Zen 4c, while the 9565 is a 72-core Turin part built on Zen 5. The data shows a clear split: the 9565 dominates compute-heavy and single-threaded workloads, while the 9754 leads in specific throughput and data-processing tasks. This analysis breaks down where each processor wins, the architectural reasons behind those results, and the head-to-head benchmark numbers.

Where Each One Wins

The benchmark results indicate a fundamental division of labor. The AMD EPYC 9565 wins the majority of the head-to-head tests, taking 11 of 17 comparisons. Its victories are concentrated in rendering, physics, and single-thread performance. The Cinebench suite, which measures both multi-core and single-core rendering capability, goes entirely to the 9565. It also wins the PassMark multithread and physics tests. This makes it the stronger choice for general-purpose compute, virtualized environments where per-thread speed matters, and any workload that scales with clock frequency and instruction-level efficiency.

The AMD EPYC 9754, despite losing the overall count, wins 6 tests, and these wins are strategically important. It takes the PassMark data compression, data encryption, extended instructions, floating point math, integer math, and random string sorting tests. These are not marginal wins either; the margins are often substantial. For example, the 9754 is 63.4% ahead in data encryption and 43% ahead in integer math. This indicates a processor optimized for specific data-centric tasks, likely benefiting from its massive core count and larger aggregate cache in certain parallel throughput scenarios.

The use-case split is therefore clear: the 9565 is for latency-sensitive, high-frequency computing, while the 9754 is for high-throughput data processing where raw parallel execution of independent operations is paramount. The 9565’s wins in Cinebench and physics suggest it handles complex, interdependent calculations faster, whereas the 9754’s wins in compression and encryption suggest it can grind through large data streams more effectively.

Architecture Differences

The architectural gap between these two processors is significant and explains the benchmark divergence. The 9754 is based on the Zen 4 and Zen 4c architecture, codenamed Bergamo, and is manufactured on a 5 nm process by TSMC. It packs 128 cores and 256 threads, with a base clock of 2.25 GHz and a boost clock of 3.10 GHz. Its cache layout is 64 KB of L1 per core, 1 MB of L2 per core, and a shared 256 MB L3. The 9565, in contrast, uses the newer Zen 5 architecture, codenamed Turin, on a 4 nm process. It has 72 cores and 144 threads, but runs at much higher clocks: 3.15 GHz base and 4.30 GHz boost. Its cache is 80 KB of L1 per core, 1 MB of L2 per core, and a larger 384 MB shared L3.

The transistor counts and die sizes reflect these differences. The 9754 has 71,000 million transistors across eight 73 mm² dies. The 9565 has 99,780 million transistors across twelve 70.6 mm² dies. Despite having fewer cores, the 9565 has more transistors, indicating a more complex and capable core design per unit. The process node advantage, 4 nm vs 5 nm, and the newer Zen 5 architecture contribute to its higher clock speeds and superior single-thread performance.

Both processors share the same socket (AMD Socket SP5), support DDR5 memory with a twelve-channel bus, and offer PCIe Gen 5 with 128 lanes. However, the memory bandwidth differs: the 9754 is rated at 460.8 GB/s, while the 9565 reaches 576.0 GB/s. The 9565 also has a higher TDP at 400 watts compared to the 9754’s 360 watts, which aligns with its higher clock speeds. The 9565 lacks integrated graphics, while the 9754 has none listed, so both require a discrete GPU for display output.

Head-to-Head Benchmarks

The Cinebench results are uniformly in favor of the 9565, with a consistent delta of -27% for the 9754 across all tests. In Cinebench R15 multi-core, the 9565 scores 11,585 against the 9754’s 8,460. The single-core R15 test shows 1,635 versus 1,194. Cinebench R20 multi-core shows 48,273 versus 35,254, and single-core shows 6,814 versus 4,977. In Cinebench R23, the multi-core gap is 114,937 versus 83,939, and single-core is 16,226 versus 11,850. This consistent 27% deficit across all Cinebench variants indicates a fundamental per-core performance advantage for the 9565 that its lower core count cannot overcome.

The PassMark tests tell a different story. The 9754 wins data compression with a score of 3,558,043 against 2,579,631, a 37.9% lead. Data encryption is a dominant win for the 9754 at 231,891 versus 141,936, a 63.4% advantage. Extended instructions go to the 9754 by a narrower 7% margin (224,322 vs 209,595). Floating point math is another 9754 win, 588,187 versus 549,422, a 7.1% lead. Integer math is a major 9754 victory: 1,026,896 versus 717,948, a 43% margin. Random string sorting also favors the 9754, 306,481 versus 291,941, a 5% edge.

The 9565 counters in other PassMark tests. Find prime numbers is a massive win for the 9565, scoring 2,422 versus 604, a 75.1% lead for the 9565 (the delta is listed as -75.1% for the 9754). Physics is another 9565 win, 18,036 versus 8,793, a 51.2% advantage. Multithread score goes to the 9565, 135,221 versus 98,752, a 27% lead. Single-thread tests show 3,696 versus 2,328, a 37% advantage for the 9565.

FAQ

Q: Which processor has more cores?

A: The AMD EPYC 9754 has 128 cores and 256 threads, while the AMD EPYC 9565 has 72 cores and 144 threads.

Q: Why does the 9565 win Cinebench multi-core tests despite having fewer cores?

A: The 9565’s Zen 5 architecture and higher clocks (3.15 GHz base, 4.30 GHz boost) give it a significant per-core performance advantage. The data shows a consistent 27% lead for the 9565 across all Cinebench tests, indicating its cores are substantially faster individually.

Q: In which workloads does the 9754 outperform the 9565?

A: The 9754 wins in PassMark data compression, data encryption, integer math, floating point math, extended instructions, and random string sorting. Its largest wins are 63.4% in encryption and 43% in integer math.

Q: What is the memory bandwidth difference?

A: The 9565 offers higher memory bandwidth at 576.0 GB/s, while the 9754 is rated at 460.8 GB/s. Both support DDR5 with a twelve-channel memory bus.

Q: Do both processors use the same socket?

A: Yes, both the 9754 and the 9565 use the AMD Socket SP5 and support PCIe Gen 5 with 128 lanes.

Q: Which processor has a larger L3 cache?

A: The 9565 has a larger L3 cache at 384 MB shared, compared to the 9754’s 256 MB shared.

The Verdict

The data points to a clear recommendation based on workload type. For most general-purpose server tasks, rendering, virtualization, and single-threaded applications, the AMD EPYC 9565 is the superior choice. It wins the majority of benchmarks, including all Cinebench tests, PassMark multithread, physics, and single-thread. Its 27% lead in Cinebench multi-core and 37% lead in single-thread performance make it the more responsive and capable processor for diverse compute loads.

The AMD EPYC 9754 is the specialist. It is the better option for data-heavy workloads that can leverage its 128 cores. The benchmark data shows it excels in data compression, encryption, integer math, and floating point math. If the primary workload involves encrypting large datasets, compressing files, or processing massive streams of independent data, the 9754’s 63.4% lead in encryption and 43% lead in integer math are decisive. Its average benchmark score is higher than the 9565’s, at 364,371 versus 285,471, but this average is skewed by its dominant wins in specific tests.

The 9565 is the more balanced and modern processor, with a 99th percentile ranking versus the 9754’s 100th. The 9754’s 100th percentile reflects its specific strengths. The choice is not about which is better overall, but which fits the server’s primary function. For a general-purpose compute node, pick the 9565. For a dedicated data-processing or encryption appliance, the 9754 is the data-backed winner.

Specification Differences

  • Cores: The 9754 has 128 cores; the 9565 has 72 cores.
  • Threads: The 9754 has 256 threads; the 9565 has 144 threads.
  • Base Clock: The 9754 runs at 2.25 GHz; the 9565 runs at 3.15 GHz.
  • Boost Clock: The 9754 boosts to 3.10 GHz; the 9565 boosts to 4.30 GHz.
  • TDP: The 9754 is rated at 360 watts; the 9565 is rated at 400 watts.
  • Architecture: The 9754 uses Zen 4; the 9565 uses Zen 5.
  • Codename: The 9754 is Bergamo; the 9565 is Turin.
  • Process Node: The 9754 is on 5 nm; the 9565 is on 4 nm.
  • Transistors: The 9754 has 71,000 million; the 9565 has 99,780 million.
  • Die Size: The 9754 uses 8x 73 mm² dies; the 9565 uses 12x 70.6 mm² dies.
  • L1 Cache: The 9754 has 64 KB per core; the 9565 has 80 KB per core.
  • L3 Cache: The 9754 has 256 MB shared; the 9565 has 384 MB shared.
  • Memory Bandwidth: The 9754 is rated at 460.8 GB/s; the 9565 is rated at 576.0 GB/s.
  • Integrated Graphics: The 9754 has none listed; the 9565 is listed as N/A.
  • Release Date: The 9754 released on 2023-06-12; the 9565 released on 2024-10-09.
  • Launch MSRP: The 9754 launched at $11900; the 9565 launched at $10486.

DETAILED SPECIFICATIONS

SPECIFICATION
EPYC 9565
EPYC 9754
Core Specs
Cores
72
128 +77.8%
Threads
144
256 +77.8%
Base Clock (GHz)
3.15
2.25 -28.6%
Boost Clock (GHz)
4.3
3.1 -27.9%
Frequency (GHz)
3.15
2.25 -28.6%
Turbo Clock (GHz)
4.3
3.1 -27.9%
Multiplier
31.5
22.5 -28.6%
SMP CPUs
2
2 0.0%
Cache
L1 Cache
80 KB (per core)
64 KB (per core)
L2 Cache
1 MB (per core)
1 MB (per core)
L3 Cache
384 MB (shared)
256 MB (shared)
Power
TDP (W)
400
360 -10.0%
Configurable TDP
320-400 W
320-400 W
Architecture
Architecture
Zen 5
Zen 4
Codename
Turin
Bergamo
Generation
EPYC (Zen 5 (Turin))
EPYC (Zen 4c (Bergamo))
Process Size
4 nm
5 nm
Transistors
99,780 million
71,000 million
Die Size
12x 70.6 mm²
8x 73 mm²
Foundry
TSMC
TSMC
Memory
Memory Support
DDR5
DDR5
Memory Bus
Twelve-channel
Twelve-channel
Memory Bandwidth
576.0 GB/s
460.8 GB/s
ECC Memory
Yes
Yes
Platform
Socket
AMD Socket SP5
AMD Socket SP5
PCIe
Gen 5, 128 Lanes(CPU only)
Gen 5, 128 Lanes(CPU only)
AMD Multi-Die
IO Process Size
6 nm
6 nm
Interconnect
CXL
Gen 2.0
Other
Market
Server/Workstation
Server/Workstation
Production Status
Active
Active
Launch Price
$10486
$11900
Part Number
100-000001447
100-000001234
Package
FC-LGA6096
FC-LGA6096
Bundled Cooler
None
View EPYC 9565 Details View EPYC 9754 Details