AMD Ryzen 9 PRO 9945 vs Intel Xeon 6731P Comparison

FAQ

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

A: The AMD Ryzen 9 PRO 9945 scores 4619 in PassMark single-thread tests, while the Intel Xeon 6731P scores 2107. The AMD part leads by 119.2% in this metric.

Q: How do the two chips compare in multithreaded performance?

A: The Intel Xeon 6731P wins the PassMark multithread test with 52790 versus 48664 for the AMD Ryzen 9 PRO 9945, a 7.8% advantage for Intel.

Q: What is the biggest benchmark margin between the two?

A: The largest gap is in PassMark physics, where the Intel Xeon 6731P scores 7105 versus 2902 for AMD, a 59.2% difference in favor of Intel.

Q: Which processor has more cores and threads?

A: The Intel Xeon 6731P has 32 cores and 64 threads, compared to 12 cores and 24 threads on the AMD Ryzen 9 PRO 9945.

Q: What are the memory bandwidth figures for each?

A: The AMD Ryzen 9 PRO 9945 has 89.6 GB/s over a dual-channel DDR5 bus, while the Intel Xeon 6731P has 409.6 GB/s over an eight-channel DDR5 bus.

Q: Do both processors support ECC memory?

A: Yes, both the AMD Ryzen 9 PRO 9945 and the Intel Xeon 6731P support ECC memory.

Architecture Differences

The AMD Ryzen 9 PRO 9945 uses the Granite Ridge architecture based on Zen 5 cores, fabricated on a 4 nm process at TSMC. It is a dual-chiplet design with a die size of 2x 70.6 mm², containing 16,630 million transistors. The Intel Xeon 6731P uses the Granite Rapids architecture, built on Intel's 5 nm process, with a monolithic die measuring 598 mm².

Cache layouts differ substantially. The AMD part provides 80 KB of L1 per core, 1 MB of L2 per core, and 64 MB of shared L3 cache. The Intel part offers 112 KB of L1 per core, 2 MB of L2 per core, and a much larger 144 MB of shared L3 cache. This gives Intel a significant cache advantage for workloads with large working sets.

The memory controllers also diverge. AMD uses a dual-channel DDR5 interface with 89.6 GB/s bandwidth, while Intel uses an eight-channel DDR5 interface with 409.6 GB/s. That is 4.6 times the raw memory bandwidth for Intel, which matters for memory-bound server applications.

PCIe connectivity is another major differentiator. AMD provides Gen 5 with 24 lanes from the CPU, while Intel offers Gen 5 with 136 lanes. The Intel part also lacks integrated graphics, whereas the AMD Ryzen 9 PRO 9945 includes Radeon Graphics. Both are locked multipliers and target the server and workstation market segment.

The power envelopes are far apart: the AMD chip has a TDP of 65 watts, while the Intel chip draws 245 watts. That difference reflects the core count and clock strategy, with AMD leaning on a 3.40 GHz base and 5.40 GHz boost, versus Intel's 2.50 GHz base and 4.10 GHz boost.

Where Each One Wins

The AMD Ryzen 9 PRO 9945 wins exclusively in single-threaded testing. Its PassMark single-thread score of 4619 is 119.2% higher than the Intel Xeon 6731P's 2107. This gives AMD the edge for lightly threaded workloads, latency-sensitive tasks, and applications that depend on high per-core clock speeds. The 5.40 GHz boost clock, combined with the efficient 4 nm Zen 5 design, makes this chip suitable for workstation use where single-thread responsiveness is critical.

The Intel Xeon 6731P dominates every other recorded benchmark. Its wins include data compression, data encryption, extended instructions, prime number finding, floating-point math, integer math, multithreading, physics, and random string sorting. The largest advantages come in physics (59.2% ahead) and prime number finding (36.8% ahead), but the Intel part is consistently faster across all heavy compute workloads. The combination of 32 cores, 64 threads, 144 MB of L3 cache, and 409.6 GB/s memory bandwidth makes it the clear choice for parallel processing, large dataset manipulation, and high-throughput server tasks.

The data shows a clean split: AMD for single-thread dominance, Intel for almost everything else. Buyers should match the workload to the chip's strengths.

Specification Differences

The two processors differ on nearly every core specification. The AMD Ryzen 9 PRO 9945 has 12 cores and 24 threads, while the Intel Xeon 6731P has 32 cores and 64 threads. Clock speeds go the other way: AMD runs at 3.40 GHz base and 5.40 GHz boost, Intel at 2.50 GHz base and 4.10 GHz boost. The TDP gap is substantial, with AMD at 65 watts and Intel at 245 watts.

Socket and platform requirements are incompatible. AMD uses Socket AM5, Intel uses Socket 4710. The process node differs as well: AMD is on 4 nm from TSMC, Intel is on 5 nm at Intel. Die sizes are 2x 70.6 mm² for AMD versus 598 mm² for Intel. The transistor count is only listed for AMD at 16,630 million.

Cache hierarchies: AMD has 80 KB L1 per core, 1 MB L2 per core, and 64 MB L3. Intel has 112 KB L1 per core, 2 MB L2 per core, and 144 MB L3. Memory channels are dual-channel for AMD and eight-channel for Intel, with corresponding bandwidth of 89.6 GB/s versus 409.6 GB/s. PCIe lanes are 24 for AMD and 136 for Intel, both Gen 5.

Integrated graphics are present only on the AMD chip. Both support ECC memory. The AMD part released on 2025-09-15 and the Intel part on 2025-02-23. The Intel part has a launch MSRP of $2700, while no launch MSRP is recorded for the AMD chip.

Head-to-Head Benchmarks

The Intel Xeon 6731P wins 9 of the 11 recorded head-to-head comparisons. Its largest victory is in PassMark physics, where it scores 7105 against AMD's 2902, a 59.2% margin. This indicates significantly better performance in simulation and physics-heavy workloads, likely from the combination of more cores and the larger cache.

In PassMark find prime numbers, Intel scores 541 versus 342, a 36.8% advantage. Extended instructions show Intel at 65656 versus 44374, a 32.4% lead. Floating-point math follows the same pattern: Intel at 157330 versus 107340, a 31.8% margin. Random string sorting gives Intel 88019 versus 62458, a 29% lead. Data compression is Intel at 799474 versus 579972, a 27.5% edge. Data encryption shows Intel at 40087 versus 30450, a 24% difference. Integer math is closer, with Intel at 198761 versus 171175, a 13.9% margin. The multithread test is the narrowest Intel win: 52790 versus 48664, only 7.8% ahead.

The AMD Ryzen 9 PRO 9945 wins both single-thread tests by the same margin. Its score of 4619 doubles Intel's 2107, a 119.2% advantage. This is the single largest delta in either direction across the entire benchmark suite.

The average benchmark scores reflect the overall split. AMD averages 96083 with a percentile of 96 among all CPUs. Intel averages 87756, also at the 96th percentile. The nearest rivals data shows AMD's closest competitor is the AMD EPYC 4565P at 95764, just 0.3% behind, while Intel's nearest rival is the Intel Xeon 6736P at 87864, a 0.1% difference.

The Verdict

The benchmark data presents a clear choice based on workload profile. The AMD Ryzen 9 PRO 9945 is the pick for single-threaded performance. Its 119.2% lead in PassMark single-thread over the Intel Xeon 6731P is decisive. Workloads that rely on high clock speeds, low latency per core, and efficient power draw fit this chip. The 65 watt TDP also makes it far easier to cool and power in a workstation chassis.

The Intel Xeon 6731P is the pick for parallel and data-intensive workloads. It wins 9 of 11 benchmarks, with the multithread advantage of 7.8% being the smallest of its victories. The 32-core count, 64 threads, 144 MB of L3 cache, and 409.6 GB/s memory bandwidth deliver consistent wins across compression, encryption, math, and physics tests. The 245 watt TDP is a cost of entry for that capability.

For a single-socket server handling virtualization, database queries, or scientific computing where parallelism is the norm, the Intel part is the stronger choice. For a workstation where interactive responsiveness, single-thread application performance, or power efficiency are priorities, the AMD part leads.

The data does not indicate a universal winner. It shows two specialized tools. The AMD Ryzen 9 PRO 9945 and Intel Xeon 6731P both sit at the 96th percentile of all CPUs, but they achieve that rank through opposite strategies: extreme single-thread efficiency versus massive parallel throughput. Choose based on the workload, not the average score.