AMD EPYC 9335 vs Intel Xeon w9-3575X Comparison

AMD
AMD

AMD EPYC 9335

CORE STATE Turin
CORE SPECS 32 Cores / 64 Threads
CLOCK SPEED 3 Base / 4.4 GHz Turbo
CACHE 128 MB (shared)
MAX TDP 210W
ARCHITECTURE Zen 5
nm
PROCESS 4 nm
LAUNCH DATE 2024
VS
Intel
INTEL

Xeon w9-3575X

CORE STATE Sapphire Rapids
CORE SPECS 44 Cores / 88 Threads
CLOCK SPEED 2.2 Base / 4.8 GHz Turbo
CACHE 97.5 MB
MAX TDP 340W
ARCHITECTURE Sapphire Rapids
nm
PROCESS 10 nm
LAUNCH DATE 2024

PERFORMANCE BENCHMARKS

passmark_data_compression
1,203,096
1,219,584
passmark_data_encryption
63,159
62,258
passmark_extended_instructions
105,706
109,843
passmark_find_prime_numbers
340
687
passmark_floating_point_math
228,123
273,398
passmark_integer_math
346,291
298,224
passmark_multithread
65,811
83,338
passmark_physics
1,905
6,836
passmark_random_string_sorting
116,608
134,723
passmark_single_thread
2,732
3,672
passmark_singlethread
2,732
3,672
cinebench_cinebench_r15_multicore
N/A
7,140
cinebench_cinebench_r15_singlecore
N/A
1,008
cinebench_cinebench_r20_multicore
N/A
29,751
cinebench_cinebench_r20_singlecore
N/A
4,200
cinebench_cinebench_r23_multicore
N/A
70,837

Analysis: AMD EPYC 9335 vs Intel Xeon w9-3575X

Head-to-Head Benchmarks

The benchmark comparison between the AMD EPYC 9335 and the Intel Xeon w9-3575X reveals a clear split in workload dominance. The Intel processor wins 9 of the 11 recorded tests, while the AMD chip takes 2. The largest single gap appears in the passmark_physics test, where the Intel Xeon w9-3575X scores 6836 against the EPYC 9335's 1905, a delta of -72.1% from the AMD part's perspective. That is a massive margin, suggesting the Intel chip handles physics simulations with far greater efficiency in this measurement.

The passmark_find_prime_numbers test shows a similarly lopsided result. Intel scores 687, AMD scores 340, a -50.5% difference. Prime number finding often stresses integer throughput and cache behavior, and the Intel part dominates here. The passmark_single_thread test also favors Intel decisively: 3672 versus 2732, a -25.6% gap. This aligns with the boost clock advantage, as the Intel chip reaches 4.80 GHz compared to the AMD's 4.40 GHz.

However, the AMD EPYC 9335 has its own victories. In passmark_integer_math, AMD scores 346291 against Intel's 298224, a 16.1% lead. This is a substantial win for the AMD part, indicating that its Zen 5 architecture handles integer operations more efficiently despite fewer cores. The passmark_data_encryption test also goes to AMD, but narrowly: 63159 versus 62258, a 1.4% margin. Encryption workloads often benefit from dedicated instructions and memory bandwidth, and AMD's twelve-channel memory bus likely contributes here.

The remaining tests show Intel ahead with varying margins. Passmark_multithread gives Intel 83338 versus AMD's 65811, a -21% delta. This is expected given the core count difference: Intel has 44 cores and 88 threads, AMD has 32 cores and 64 threads. Passmark_floating_point_math shows Intel at 273398 versus AMD's 228123, a -16.6% gap. Passmark_random_string_sorting gives Intel 134723 against 116608, a -13.4% difference. Passmark_extended_instructions shows Intel at 109843 versus 105706, a -3.8% margin. Passmark_data_compression is nearly tied: Intel at 1219584, AMD at 1203096, only a -1.4% difference.

The average benchmark scores in the database tell a different story when placed against their respective rivals. The AMD EPYC 9335 has an average benchmark score of 194228, placing it in the 99th percentile of all CPUs. Its nearest rival, the Intel Xeon 6741P, scores 194901, a delta of -0.3%, meaning the AMD part trails by a hair. The Intel Xeon 678X scores 193477, a 0.4% lead for AMD. The Intel Xeon 6740E scores 187718, a 3.5% lead for AMD. The AMD EPYC 8534P scores 185092, a 4.9% lead for AMD. So the EPYC 9335 sits in a tight cluster at the top.

The Intel Xeon w9-3575X has an average benchmark score of 144323, placing it in the 98th percentile. Its nearest rival, the AMD EPYC 7643P, scores 144824, a -0.3% delta, meaning Intel trails slightly. The AMD Ryzen 9 PRO 9965X3D scores 143735, a 0.4% lead for Intel. The Intel Xeon 6732P scores 143444, a 0.6% lead. The Intel Xeon 674X scores 143103, a 0.9% lead. The Intel part's average score is notably lower than the AMD part's, yet it wins most head-to-head tests. This discrepancy suggests the average score calculation weights certain benchmarks differently, or that the Intel part's wins are concentrated in tests that do not dominate the average.

Where Each One Wins

The Intel Xeon w9-3575X is the clear choice for single-threaded and physics-heavy workloads. The passmark_single_thread score of 3672 versus 2732 indicates a 25.6% advantage, which translates to better responsiveness in applications that rely on one or few threads. The passmark_physics score of 6836 against 1905 is a 72.1% lead, making the Intel part dramatically better for simulation and physics calculations. The passmark_find_prime_numbers result, 687 versus 340, reinforces this pattern: the Intel chip excels at workloads that stress per-core integer throughput and cache latency.

The AMD EPYC 9335 wins in two specific areas. First, passmark_integer_math: a 16.1% lead (346291 versus 298224) shows that for raw integer arithmetic across many threads, the AMD architecture is more efficient. This matters for database operations, financial calculations, and other integer-heavy server tasks. Second, passmark_data_encryption: a 1.4% lead (63159 versus 62258) indicates a slight edge in cryptographic workloads, likely due to the twelve-channel memory bus providing ample bandwidth for encryption algorithms that shuffle data rapidly.

For multithreaded general-purpose work, the Intel part wins with a 21% margin in passmark_multithread (83338 versus 65811). This is a direct consequence of having 12 more cores and 24 more threads. However, the AMD part's higher average benchmark score (194228 versus 144323) suggests that the database's overall ranking favors the AMD chip, possibly because the Intel part's wins in specific tests are offset by its lower scores in other unlisted metrics.

For memory-bandwidth-sensitive tasks, the AMD EPYC 9335 has a theoretical edge with 576.0 GB/s of bandwidth against Intel's 307.2 GB/s. Yet the benchmark data does not show a consistent AMD win in memory-heavy tests like data compression, where Intel actually wins by 1.4%. This implies that raw bandwidth alone does not guarantee performance; the Intel chip's cache hierarchy and memory controller efficiency compensate.

Architecture Differences

The two processors come from fundamentally different design philosophies. The AMD EPYC 9335 uses the Zen 5 architecture, codenamed Turin, built on a 4 nm process at TSMC. It has 32 cores and 64 threads, with a base clock of 3.00 GHz and a boost clock of 4.40 GHz. The transistor count is 33,260 million, spread across four chiplets, each 70.6 mm² in size. The L1 cache is 80 KB per core, L2 is 1 MB per core, and L3 is 128 MB shared. It supports DDR5 memory over a twelve-channel bus, with 128 PCIe Gen 5 lanes. The socket is AMD Socket SP5, and the TDP is 210 watts.

The Intel Xeon w9-3575X uses the Sapphire Rapids architecture, built on a 10 nm process at Intel. It has 44 cores and 88 threads, with a base clock of 2.20 GHz and a boost clock of 4.80 GHz. The die size is 4x 477 mm², which is significantly larger than the AMD chiplets. The L1 cache is 80 KB per core, L2 is 2 MB per core, and L3 is 97.5 MB. It supports DDR5 over an eight-channel bus, with 112 PCIe Gen 5 lanes. The socket is Intel Socket 4677, and the TDP is 340 watts. The Intel part has an unlocked multiplier, while the AMD part does not.

These differences explain the benchmark outcomes. The Intel chip's higher boost clock (4.80 GHz versus 4.40 GHz) drives its single-thread win. Its larger L2 cache per core (2 MB versus 1 MB) likely contributes to the physics and prime number wins, as those workloads benefit from fast access to working sets. The AMD chip's larger L3 cache (128 MB versus 97.5 MB) and higher memory bandwidth (576.0 GB/s versus 307.2 GB/s) should favor large data workloads, yet the benchmarks show Intel winning most of those too, except integer math and encryption.

The process node difference is stark: 4 nm versus 10 nm. The AMD chip achieves higher efficiency per watt, with 210 watts versus 340 watts, yet delivers competitive performance in several tests. The Intel chip uses more power but also more cores and higher clocks. The release dates differ: the Intel part launched on 2024-08-23, the AMD part on 2024-10-09.

FAQ

Q: Which processor has more cores?

A: The Intel Xeon w9-3575X has 44 cores and 88 threads, while the AMD EPYC 9335 has 32 cores and 64 threads.

Q: What is the single-thread performance difference?

A: The Intel part scores 3672 in passmark_single_thread, while the AMD part scores 2732, giving Intel a 25.6% advantage.

Q: Does the AMD chip win any benchmarks?

A: Yes, the AMD EPYC 9335 wins passmark_integer_math by 16.1% (346291 versus 298224) and passmark_data_encryption by 1.4% (63159 versus 62258).

Q: Which processor has higher memory bandwidth?

A: The AMD EPYC 9335 supports twelve-channel DDR5 with 576.0 GB/s, while the Intel Xeon w9-3575X supports eight-channel DDR5 with 307.2 GB/s.

Q: How do their average benchmark scores compare?

A: The AMD EPYC 9335 has an average benchmark score of 194228, placing in the 99th percentile. The Intel Xeon w9-3575X has an average of 144323, placing in the 98th percentile.

Q: What are the launch MSRPs?

A: The AMD EPYC 9335 has a launch MSRP of $3178, and the Intel Xeon w9-3575X has a launch MSRP of $3789.

The Verdict

The data points to a clear role separation. The Intel Xeon w9-3575X is the better choice for anyone whose primary workloads involve physics simulations, prime number calculations, single-threaded applications, or general multithreaded tasks. Its wins in passmark_multithread (21% lead), passmark_physics (72.1% lead), and passmark_single_thread (25.6% lead) make it the superior processor for workstations where per-core speed and raw thread count matter most.

The AMD EPYC 9335 is the better choice for integer-heavy server workloads and encryption tasks. Its 16.1% lead in passmark_integer_math and 1.4% lead in passmark_data_encryption indicate that the Zen 5 architecture extracts more work per thread in these specific areas. Additionally, its higher memory bandwidth (576.0 GB/s versus 307.2 GB/s) and larger L3 cache (128 MB versus 97.5 MB) provide headroom for data-intensive applications, even if the recorded benchmarks do not always reflect that advantage.

The percentile rankings complicate the verdict. The AMD part sits in the 99th percentile with an average score of 194228, while the Intel part sits in the 98th percentile with 144323. The AMD part's nearest rivals cluster within 4.9%, showing it competes at the top tier. The Intel part's nearest rivals are also within 0.9%, but its average score is substantially lower. This suggests that the database's overall ranking favors the AMD chip, possibly because the Intel part's specific benchmark wins are offset by weaker performance in other unlisted tests.

For a server or workstation buyer, the decision hinges on workload mix. If the tasks are dominated by physics, prime number finding, or single-threaded code, the Intel Xeon w9-3575X is the data-backed pick. If the tasks are integer math and encryption, the AMD EPYC 9335 wins. For mixed workloads, the AMD part's higher average score and lower TDP (210 watts versus 340 watts) make it the more balanced option, despite the Intel chip's core count advantage.

Specification Differences

| Specification | AMD EPYC 9335 | Intel Xeon w9-3575X |

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

| Cores | 32 | 44 |

| Threads | 64 | 88 |

| Base Clock | 3.00 GHz | 2.20 GHz |

| Boost Clock | 4.40 GHz | 4.80 GHz |

| TDP | 210 W | 340 W |

| Socket | AMD Socket SP5 | Intel Socket 4677 |

| Architecture | Zen 5 | Sapphire Rapids |

| Codename | Turin | Sapphire Rapids |

| Process Node | 4 nm | 10 nm |

| Foundry | TSMC | Intel |

| Die Size | 4x 70.6 mm² | 4x 477 mm² |

| L2 Cache | 1 MB (per core) | 2 MB (per core) |

| L3 Cache | 128 MB (shared) | 97.5 MB |

| Memory Bus | Twelve-channel | Eight-channel |

| Memory Bandwidth | 576.0 GB/s | 307.2 GB/s |

| PCIe | Gen 5, 128 Lanes | Gen 5, 112 Lanes |

| Multiplier Unlocked | No | Yes |

| Launch MSRP | $3178 | $3789 |

| Release Date | 2024-10-09 | 2024-08-23 |

| Part Number | 100-000001149 | SRN72 |

DETAILED SPECIFICATIONS

SPECIFICATION
EPYC 9335
w9-3575X
Core Specs
Cores
32
44 +37.5%
Threads
64
88 +37.5%
Base Clock (GHz)
3
2.2 -26.7%
Boost Clock (GHz)
4.4
4.8 +9.1%
Frequency (GHz)
3
2.2 -26.7%
Turbo Clock (GHz)
4.4
4.8 +9.1%
Multiplier
30
22 -26.7%
SMP CPUs
2
1 -50.0%
Cache
L1 Cache
80 KB (per core)
80 KB (per core)
L2 Cache
1 MB (per core)
2 MB (per core)
L3 Cache
128 MB (shared)
97.5 MB
Power
TDP (W)
210
340 +61.9%
Configurable TDP
200-240 W
—
Architecture
Architecture
Zen 5
—
Codename
Turin
Sapphire Rapids
Generation
EPYC (Zen 5 (Turin))
Xeon W (Sapphire Rapids)
Process Size
4 nm
10 nm
Transistors
33,260 million
—
Die Size
4x 70.6 mm²
4x 477 mm²
Foundry
TSMC
Intel
Memory
Memory Support
DDR5
DDR5
Memory Bus
Twelve-channel
Eight-channel
Memory Bandwidth
576.0 GB/s
307.2 GB/s
ECC Memory
Yes
Yes
Platform
Socket
AMD Socket SP5
Intel Socket 4677
PCIe
Gen 5, 128 Lanes(CPU only)
Gen 5, 112 Lanes(CPU only)
DMI
—
4.0 x8
AMD Multi-Die
IO Process Size
6 nm
—
Interconnect
CXL
Gen 2.0
—
Other
Market
Server/Workstation
Server/Workstation
Production Status
Active
Active
Launch Price
$3178
$3789
Part Number
100-000001149
SRN72
Package
FC-LGA6096
FC-LGA16A
View EPYC 9335 Details View Xeon w9-3575X Details