AMD Ryzen Threadripper PRO 9975WX vs Intel Xeon 678X Comparison

The Intel Xeon 678X and AMD Ryzen Threadripper PRO 9975WX are both flagship workstation processors, but benchmark results show they serve distinctly different masters. The AMD part wins 11 of 16 head-to-head tests, dominating rendering workloads and general multi-threaded tasks, while the Intel chip wins 5 tests, primarily in specialized compute fields like prime number finding and floating-point math. The data indicates AMD's 32-core Zen 5 design delivers superior performance per core and in mixed workloads, while Intel's 48-core Granite Rapids architecture excels in specific high-throughput calculations.

Where Each One Wins

The AMD Ryzen Threadripper PRO 9975WX is the clear winner for mainstream workstation applications. It takes decisive victories in all three Cinebench versions, with multicore scores of 9430 (R15), 39292 (R20), and 93553 (R23) compared to Intel's 8444, 35185, and 83775 respectively. The AMD chip also leads in integer math, scoring 461724 versus Intel's 405075, a 12.3% advantage. For random string sorting, AMD wins 188014 to 164102, a 12.7% gap. The PassMark multithread test also favors AMD at 104902 versus 98559. This pattern shows AMD's strength in general-purpose computation, encryption (85035 vs 83598), and single-threaded tasks where it posts 4408 versus 3758.

The Intel Xeon 678X wins in more specialized domains. Its most dramatic victory is in prime number finding, scoring 1010 versus AMD's 620, a massive 62.9% lead. Floating-point math also goes Intel's way at 362070 versus 304833, an 18.8% advantage. Data compression shows Intel ahead at 1690896 versus 1644573, a modest 2.8% win. Extended instructions favor Intel at 141431 versus 137733, a 2.7% edge. The physics test also goes to Intel at 7809 versus 7288, a 7.1% margin. These results suggest Intel's architecture handles algorithmic and scientific workloads more efficiently.

Architecture Differences

The fundamental design philosophies diverge sharply. Intel's Xeon 678X uses 48 cores and 96 threads built on Granite Rapids architecture with a 5 nm process node from Intel's own foundry. AMD's Threadripper PRO 9975WX uses 32 cores and 64 threads on Zen 5 architecture (codenamed Shimada Peak) with a 4 nm process node from TSMC. Intel's die size is 2x 598 mm², while AMD uses 4x 70.6 mm² dies totaling 33,260 million transistors. Cache configurations also differ significantly: Intel provides 192 MB shared L3 cache with 2 MB L2 per core and 112 KB L1 per core, whereas AMD offers 128 MB L3 with 1 MB L2 and 64 KB L1 per core.

Both processors support DDR5 memory across eight channels with identical 409.6 GB/s bandwidth, and both feature 128 PCIe Gen 5 lanes. Neither has integrated graphics. Both support ECC memory and have unlocked multipliers. The Intel part has a 300 W TDP and launches at a higher base clock of 2.40 GHz but boosts to 4.90 GHz. AMD operates at a 350 W TDP with a 4.00 GHz base clock and 5.40 GHz boost. The Xeon uses Intel Socket 4710 while AMD uses Socket sTR5. The AMD part's higher clock speeds, combined with TSMC's 4 nm process, explain its single-thread dominance.

Head-to-Head Benchmarks

The Cinebench results are remarkably consistent. AMD leads by 10.5% in R15 multicore (9430 vs 8444), 10.4% in R15 single-core (1331 vs 1192), and the same margins persist in R20 and R23 versions. This uniformity suggests a fundamental per-core performance advantage rather than a workload-specific quirk. In single-thread PassMark, AMD's 4408 score beats Intel's 3758 by 14.7%, the largest single-thread margin in the comparison.

Intel's biggest wins are equally telling. The 62.9% lead in prime number finding (1010 vs 620) indicates a specialized integer algorithm where Intel's wider cores or cache hierarchy excels. Floating-point math shows Intel ahead by 18.8% (362070 vs 304833), which could benefit scientific simulation and financial modeling. The physics test, which often correlates with gaming physics engines, goes to Intel by 7.1% (7809 vs 7288). However, Intel's wins in compression (2.8%) and extended instructions (2.7%) are narrow, suggesting AMD is competitive in those areas.

The data encryption test shows AMD winning by only 1.7% (85035 vs 83598), a near tie. AMD's larger wins in integer math (12.3%) and random string sorting (12.7%) demonstrate stronger general-purpose throughput. The PassMark multithread score gives AMD a 6% edge (104902 vs 98559), which is less than the Cinebench margins, indicating the gap varies by workload type.

FAQ

Q: Which processor has more cores and threads?

A: The Intel Xeon 678X has 48 cores and 96 threads, while the AMD Ryzen Threadripper PRO 9975WX has 32 cores and 64 threads.

Q: Which processor is faster in single-core performance?

A: The AMD part wins all single-core tests, including a 14.7% lead in PassMark single-thread (4408 vs 3758) and a 10.4% lead in Cinebench R15 single-core (1331 vs 1192).

Q: Where does the Intel Xeon 678X have its biggest advantage?

A: Intel's largest win is in PassMark find prime numbers, scoring 1010 versus AMD's 620, a 62.9% advantage. It also leads floating-point math by 18.8%.

Q: Do both processors support the same memory configuration?

A: Yes, both support DDR5 memory with eight-channel buses and 409.6 GB/s bandwidth, and both support ECC memory.

Q: What are the TDP differences between the two?

A: The Intel Xeon 678X has a 300 W TDP, while the AMD Ryzen Threadripper PRO 9975WX has a 350 W TDP.

Q: Which processor has more L3 cache?

A: The Intel Xeon 678X has 192 MB shared L3 cache, while the AMD Ryzen Threadripper PRO 9975WX has 128 MB L3 cache.

The Verdict

The benchmark data supports a clear but nuanced conclusion. For most workstation users running rendering, compilation, or data processing, the AMD Ryzen Threadripper PRO 9975WX is the better choice. It wins all Cinebench tests by roughly 10.5% despite having 16 fewer cores, demonstrating superior core efficiency. Its 4.00 GHz base and 5.40 GHz boost clocks, combined with the 4 nm TSMC process, deliver higher single-thread performance that benefits virtually all interactive workloads. The 12.3% lead in integer math and 12.7% lead in random string sorting make it strong for database operations and code compilation.

The Intel Xeon 678X wins where raw algorithmic throughput matters most. Its 62.9% dominance in prime number finding and 18.8% lead in floating-point math suggest scientific computing, cryptography, and certain simulation workloads will run substantially faster. The 7.1% physics win and 2.8% compression advantage add to its specialized appeal. With 48 cores versus 32, Intel's part also offers more parallel headroom for heavily threaded custom workloads, even if AMD's efficiency closes the gap in standard benchmarks.

Users should choose based on workload profile. If the primary tasks involve Cinebench-style rendering, integer-heavy data processing, or general productivity, the AMD part delivers superior results across the board. If the workload is dominated by floating-point mathematics, prime number calculations, or physics simulations, the Intel part provides meaningful advantages that outweigh its losses elsewhere. The Xeon 678X sits in the 99th percentile of all CPUs with an average benchmark score of 193477, while the Threadripper sits in the 98th percentile with 182700, confirming both are elite performers. The data shows AMD wins more tests, but Intel wins the tests that matter most for specific scientific and mathematical applications.

Specification Differences

| Specification | Intel Xeon 678X | AMD Ryzen Threadripper PRO 9975WX |

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

| Cores | 48 | 32 |

| Threads | 96 | 64 |

| Base Clock | 2.40 GHz | 4.00 GHz |

| Boost Clock | 4.90 GHz | 5.40 GHz |

| TDP | 300 W | 350 W |

| Socket | Intel Socket 4710 | AMD Socket sTR5 |

| Architecture | Granite Rapids | Zen 5 (Shimada Peak) |

| Process Node | 5 nm (Intel) | 4 nm (TSMC) |

| Die Size | 2x 598 mm² | 4x 70.6 mm² |

| Transistors | Not specified | 33,260 million |

| L1 Cache | 112 KB per core | 64 KB per core |

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

| L3 Cache | 192 MB shared | 128 MB |

| Release Date | 2026-02-01 | 2025-07-22 |

| Launch MSRP | $3749 | $4099 |