AMD Ryzen Threadripper PRO 9945WX vs Intel Xeon 638 Comparison

Head-to-Head Benchmarks

The recorded benchmark data presents a clear split between the Intel Xeon 638 and the AMD Ryzen Threadripper PRO 9945WX. The AMD processor wins 13 of the 17 head-to-head comparisons, while the Intel part takes 4 wins. The margin of victory tells a more nuanced story than the raw win count.

In the Cinebench suite, the AMD Ryzen Threadripper PRO 9945WX is consistently ahead. Across all six Cinebench R15, R20, and R23 tests, the AMD part posts a 2.3% advantage. The multicore scores stand at 4871 versus 4757 in R15, 20296 versus 19824 in R20, and 48325 versus 47202 in R23. Single-core results follow the same pattern, with the AMD chip scoring 687 versus 671 in R15, 2865 versus 2798 in R20, and 6822 versus 6663 in R23. These are modest but uniform gains, indicating a small but consistent architectural edge.

The Intel Xeon 638 counterattacks in several Passmark workloads. The largest delta is in floating point math, where the Intel part scores 144757 against 111566, a 29.8% advantage. This is a decisive win for the Xeon. The Intel chip also leads in find prime numbers, scoring 381 versus 335, a 13.7% margin. Data compression favors Intel at 725818 versus 671963, an 8% lead. Extended instructions also go to Intel, with 56498 versus 54406, a 3.8% edge.

The AMD processor wins the remaining Passmark tests, but with varying margins. The most dramatic single-thread result favors AMD: 4573 versus 3670, a 19.7% lead. Physics performance also strongly favors AMD, with 6118 versus 4704, a 23.1% difference. Random string sorting goes to AMD at 84498 versus 74318, a 12% margin. The multithread test shows AMD ahead at 56854 versus 55651, a 2.1% win. Data encryption is close, with AMD at 36540 versus 36030, a 1.4% margin. Integer math is nearly identical, with AMD at 185421 versus 184884, just a 0.3% difference.

The overall picture from the head-to-head data is that the AMD processor dominates in single-threaded and physics-oriented workloads, while the Intel processor excels in floating point and compression tasks. The average benchmark scores confirm this: the Intel Xeon 638 posts an average of 80723, while the AMD part averages 76513. The Intel chip sits at the 95th percentile versus all CPUs, and the AMD chip also sits at the 95th percentile.

Architecture Differences

The two processors are built on fundamentally different architectures. The Intel Xeon 638 uses the Granite Rapids architecture, manufactured on a 5 nm process at Intel's own foundry. The die size is 598 mm². The AMD Ryzen Threadripper PRO 9945WX uses the Zen 5 architecture, codenamed Shimada Peak, manufactured on a 4 nm process at TSMC. The AMD part uses two chiplets, each 70.6 mm², for a total die area of 141.2 mm², and contains 16,630 million transistors.

Core and thread counts differ. The Intel part has 16 cores and 32 threads, while the AMD part has 12 cores and 24 threads. Despite having fewer cores, the AMD processor achieves higher multicore scores in Cinebench, which indicates a higher per-core throughput. The base clock for Intel is 3.20 GHz with a boost of 4.80 GHz. The AMD part runs at a 4.70 GHz base and 5.40 GHz boost. The thermal design power reflects the difference: the Intel chip is rated at 180 W, while the AMD chip is rated at 350 W.

Cache hierarchies are distinct. The Intel Xeon 638 has 112 KB of L1 cache per core, 2 MB of L2 cache per core, and 72 MB of shared L3 cache. The AMD chip has 64 KB of L1 per core, 1 MB of L2 per core, and 64 MB of L3 cache. The Intel part has a larger total cache footprint in L2 and L3, which may contribute to its strong compression and floating point results.

Memory support is a major differentiator. Both support DDR5 and ECC memory, but the Intel part uses a quad-channel memory bus with 204.8 GB/s of bandwidth. The AMD part uses an eight-channel memory bus with 409.6 GB/s of bandwidth, exactly double the Intel figure. PCIe lanes also differ: Intel provides Gen 5 with 80 lanes, while AMD provides Gen 5 with 128 lanes.

The sockets are incompatible. Intel uses Socket 4710, while AMD uses Socket sTR5. Both processors have an unlocked multiplier, and neither has integrated graphics. The Intel part is from the Xeon 600 series (Granite Rapids-WS generation), while the AMD part is from the Ryzen Threadripper 9000 series. The Intel part has a launch MSRP of $899; the AMD part has no recorded launch MSRP.

Where Each One Wins

The benchmark data supports a clear use-case split. The AMD Ryzen Threadripper PRO 9945WX is the stronger choice for single-threaded workloads and physics simulations. Its 19.7% lead in single-thread Passmark and 23.1% lead in physics are the largest margins in the entire comparison. The Cinebench single-core tests also favor AMD, with a consistent 2.3% edge across all three versions. This makes the AMD part well suited for applications that depend on per-core performance, such as interactive rendering, real-time simulation, and lightly threaded legacy code.

The AMD processor also wins in random string sorting, a test that reflects memory access patterns and sorting algorithms. A 12% margin here, combined with the eight-channel memory bus, suggests the AMD part handles data-intensive sorting tasks efficiently. The multithread Passmark score also favors AMD, but the margin is small at 2.1%, indicating that the AMD part does not dominate in all parallel workloads despite its clock advantage.

The Intel Xeon 638 is the better performer in floating point math, where it holds a 29.8% lead. This is a substantial margin and points to strong vector and scientific computing capabilities. The Intel part also wins in find prime numbers by 13.7%, a test that is often sensitive to integer throughput and branch prediction. Data compression favors Intel by 8%, which may be relevant for storage and database workloads. Extended instructions go to Intel by 3.8%, suggesting an advantage in SIMD-heavy code.

For mixed workloads, the Intel part is the safer bet when the application mix includes compression, encryption-adjacent operations, or floating point heavy calculations. The AMD part is the better choice when single-thread speed and physics are the priority. The data does not show a clear winner for integer math, as the scores are nearly identical at 184884 versus 185421, a 0.3% difference that falls within noise.

Specification Differences

The two processors differ on several specification fields. The core count is 16 for Intel versus 12 for AMD. Thread count is 32 versus 24. Base clock is 3.20 GHz versus 4.70 GHz. Boost clock is 4.80 GHz versus 5.40 GHz. TDP is 180 W versus 350 W. Socket is Intel Socket 4710 versus AMD Socket sTR5. Architecture is Granite Rapids versus Zen 5. Process node is 5 nm versus 4 nm. Foundry is Intel versus TSMC. Die size is 598 mm² versus 2x 70.6 mm². Transistor count is not listed for Intel, while AMD lists 16,630 million.

L1 cache is 112 KB per core for Intel versus 64 KB per core for AMD. L2 cache is 2 MB per core for Intel versus 1 MB per core for AMD. L3 cache is 72 MB shared for Intel versus 64 MB for AMD. Memory bus is quad-channel for Intel versus eight-channel for AMD. Memory bandwidth is 204.8 GB/s for Intel versus 409.6 GB/s for AMD. PCIe is Gen 5 with 80 lanes for Intel versus Gen 5 with 128 lanes for AMD. Release date is February 2026 for Intel versus June 2025 for AMD. Part number is SA2DN for Intel versus 100-000000726 for AMD.

Both processors share several fields: DDR5 memory support, ECC memory support, no integrated graphics, server/workstation market segment, active production status, and unlocked multipliers.

FAQ

Q: Which processor has the higher single-thread benchmark score?

A: The AMD Ryzen Threadripper PRO 9945WX scores 4573 in the Passmark single-thread test, while the Intel Xeon 638 scores 3670. The AMD part leads by 19.7%.

Q: How do the Cinebench multicore scores compare?

A: The AMD part wins all three Cinebench multicore tests: 4871 versus 4757 in R15, 20296 versus 19824 in R20, and 48325 versus 47202 in R23. Each result shows a 2.3% advantage for AMD.

Q: Where does the Intel Xeon 638 have its largest advantage?

A: The largest Intel win is in floating point math, scoring 144757 versus 111566 for AMD, a 29.8% margin. Intel also leads in find prime numbers by 13.7% and data compression by 8%.

Q: What is the memory bandwidth difference between the two?

A: The AMD Ryzen Threadripper PRO 9945WX has 409.6 GB/s of memory bandwidth on an eight-channel bus. The Intel Xeon 638 has 204.8 GB/s on a quad-channel bus. The AMD part provides exactly double the bandwidth.

Q: Which processor has more cores and threads?

A: The Intel Xeon 638 has 16 cores and 32 threads. The AMD Ryzen Threadripper PRO 9945WX has 12 cores and 24 threads. Despite fewer cores, the AMD part wins the multicore Cinebench tests.

Q: How does the physics performance compare?

A: The AMD part scores 6118 in the Passmark physics test, while the Intel part scores 4704. The AMD processor leads by 23.1%, which is its second largest margin after single-thread performance.