AMD Ryzen Threadripper PRO 9975WX vs Intel Xeon 6740E Comparison

The Intel Xeon 6740E and AMD Ryzen Threadripper PRO 9975WX occupy the same high-end server/workstation segment, but they approach it from opposite design philosophies. The Xeon packs 96 efficiency-focused cores on Intel's Sierra Forest architecture, while the Threadripper uses 32 high-frequency Zen 5 cores with simultaneous multithreading. Benchmark results show a near tie in average score (187,718 vs 186,447), but the distribution of wins is heavily skewed: the AMD part takes 13 of 17 tests, while the Intel part wins four specialized workloads.

FAQ

Q: Which processor has more cores and threads?

A: The Intel Xeon 6740E has 96 cores and 96 threads (no hyperthreading), while the AMD Ryzen Threadripper PRO 9975WX has 32 cores and 64 threads (with SMT). The Xeon has triple the core count and 50% more threads.

Q: Which CPU is faster in single-threaded workloads?

A: The Threadripper PRO 9975WX dominates single-thread tests. In Passmark single-thread it scores 4,422 vs 1,997 for the Xeon, a 54.8% advantage. Cinebench R23 single-core also favors AMD (13,207 vs 9,140).

Q: How do they compare in multi-threaded rendering?

A: AMD wins every Cinebench multi-core test by 30.8%: R15 (9,430 vs 6,525), R20 (39,292 vs 27,191), and R23 (93,553 vs 64,741). Passmark multithread also goes to AMD (110,740 vs 76,167).

Q: Does the Intel part win any benchmarks?

A: Yes, it wins four Passmark tests: data compression (1,786,845 vs 1,669,837, +7%), data encryption (137,106 vs 94,634, +44.9%), floating point math (309,333 vs 292,100, +5.9%), and random string sorting (220,684 vs 205,730, +7.3%).

Q: What are the memory and PCIe differences?

A: Both support DDR5 with eight-channel memory and identical 409.6 GB/s bandwidth. The Xeon offers 88 PCIe Gen 5 lanes, while the Threadripper provides 128 lanes.

Q: Which has a higher average benchmark score?

A: The Xeon edges out the Threadripper by 0.7% (187,718 vs 186,447), but both sit in the 99th percentile of all CPUs.

Architecture Differences

The two chips embody radically different design strategies. The Intel Xeon 6740E uses the Sierra Forest architecture (Sierra Forest-SP) built on Intel's 5 nm process, with a die size of 578 mm². It features 96 cores with no hyperthreading, giving 96 threads. The cache hierarchy is unusual: 96 KB L1 per core, 4 MB L2 per module (not per core), and 96 MB shared L3. This is a dense, efficiency-oriented layout aimed at scale-out workloads.

The AMD Ryzen Threadripper PRO 9975WX is based on Zen 5 (codenamed Shimada Peak) on TSMC's 4 nm process. It uses four chiplets, each 70.6 mm², totaling 33,260 million transistors. It has 32 cores with SMT, yielding 64 threads. L1 cache is 64 KB per core, L2 is 1 MB per core, and L3 is 128 MB shared. The AMD part also has an unlocked multiplier, while the Xeon is locked. The Threadripper's higher base and boost clocks (4.00 GHz and 5.40 GHz vs 2.40 GHz and 3.20 GHz) reflect its focus on single-thread and lightly threaded performance.

The Xeon's 96 threads are all physical, but the Threadripper's 64 threads benefit from higher per-core throughput. The process node and foundry also differ: Intel 5 nm vs TSMC 4 nm. The Xeon's larger die (578 mm²) contrasts with the Threadripper's four smaller chiplets, which may affect manufacturing yields and thermal density.

Head-to-Head Benchmarks

The head-to-head results reveal a clear pattern. In every Cinebench test (R15, R20, R23, both single and multi-core), the AMD Ryzen Threadripper PRO 9975WX wins by a uniform 30.8% margin. For example, Cinebench R23 multi-core scores 93,553 for AMD vs 64,741 for Intel; single-core is 13,207 vs 9,140. This consistency suggests a fundamental per-thread performance advantage, not a workload-specific quirk.

Passmark results are more mixed. AMD wins extended instructions (126,987 vs 78,968, +37.8% for AMD), find prime numbers (663 vs 526, +20.7%), integer math (485,286 vs 457,614, +5.7%), multithread (110,740 vs 76,167, +31.2%), physics (12,426 vs 7,608, +38.8%), and single-thread (4,422 vs 1,997, +54.8%). The single-thread gap is the largest of any test, underscoring the clock-speed and IPC advantage of Zen 5.

Intel strikes back in four areas. Data encryption shows a massive 44.9% lead (137,106 vs 94,634), likely due to dedicated cryptographic instructions. Data compression is 7% higher (1,786,845 vs 1,669,837). Floating point math is 5.9% ahead (309,333 vs 292,100), and random string sorting wins by 7.3% (220,684 vs 205,730). These wins are concentrated in memory-heavy, latency-tolerant workloads, where the Xeon's 96 cores and large L3 cache (96 MB) can compensate for lower per-core speed.

The overall win count is 13 for AMD and 4 for Intel. Yet the average benchmark score is nearly identical (187,718 vs 186,447), because the Intel wins include some very high absolute scores (e.g., compression at 1.78 million) that pull up its average.

Specification Differences

| Field | Intel Xeon 6740E | AMD Ryzen Threadripper PRO 9975WX |

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

| Cores | 96 | 32 |

| Threads | 96 | 64 |

| Base clock | 2.40 GHz | 4.00 GHz |

| Boost clock | 3.20 GHz | 5.40 GHz |

| TDP | 250 W | 350 W |

| Socket | Intel Socket 4710 | AMD Socket sTR5 |

| Architecture | Sierra Forest | Zen 5 |

| Codename | Sierra Forest | Shimada Peak |

| Process node | 5 nm | 4 nm |

| Foundry | Intel | TSMC |

| Die size | 578 mm² | 4x 70.6 mm² |

| L1 cache | 96 KB (per core) | 64 KB (per core) |

| L2 cache | 4 MB (per module) | 1 MB (per core) |

| L3 cache | 96 MB (shared) | 128 MB (shared) |

| PCIe lanes | Gen 5, 88 | Gen 5, 128 |

| Multiplier unlocked | No | Yes |

| Launch MSRP | $5265 | $4099 |

| Release date | Earlier | Later |

Both CPUs support DDR5 with eight-channel memory and 409.6 GB/s bandwidth, and both have ECC memory support. Neither has integrated graphics. The Xeon's 96 MB L3 is smaller than the Threadripper's 128 MB, but the Xeon has more total cache when counting L2 (though L2 is per module, not per core). The Threadripper's transistor count is listed at 33,260 million, while the Xeon's is not provided.

Where Each One Wins

The AMD Ryzen Threadripper PRO 9975WX is the clear choice for workloads that benefit from high single-thread throughput and moderate thread counts. Its 54.8% lead in Passmark single-thread and 30.8% lead in all Cinebench tests make it superior for 3D rendering, video encoding, physics simulation, and any application where a few fast cores matter more than many slow ones. The 38.8% advantage in Passmark physics and 31.2% in multithread reinforce this. The unlocked multiplier also allows tuning for even higher clocks, though the base 4.00 GHz is already well above the Xeon's boost.

The Intel Xeon 6740E wins in four specific areas: data compression, data encryption, floating-point math, and random string sorting. The 44.9% encryption lead is particularly notable—if a server handles heavy TLS or disk encryption, the Xeon's dedicated instructions provide a tangible edge. Its 96 cores also give it raw throughput for highly parallel integer or compression tasks, as seen in the 7% compression win despite lower per-core speed. The Xeon's lower TDP (250 W vs 350 W) may also be advantageous in dense racks, though this is not a benchmark metric.

For overall average performance, the two are within 0.7% of each other, so the decision hinges on workload mix. If the majority of tasks are rendering, simulation, or single-threaded interactive work, the Threadripper PRO 9975WX is the better performer. If the workload is dominated by encryption, compression, or floating-point math, the Xeon 6740E holds a meaningful advantage. The Xeon also offers more physical cores (96 vs 32), which could be relevant for certain virtualized or containerized environments that scale linearly with core count, but the benchmark data shows that AMD's higher per-core efficiency often overcomes the core deficit.