AMD Ryzen Threadripper PRO 9975WX vs Intel Xeon 6747P Comparison

Head-to-Head Benchmarks

The head-to-head data splits into two clear territories. The AMD Ryzen Threadripper PRO 9975WX takes the Cinebench family outright, while the Intel Xeon 6747P dominates most Passmark compute workloads.

In Cinebench R15, R20, and R23 multicore, the AMD part wins by the same margin each time: 7.6%. The scores climb from 9430 versus 8712 in R15, to 39292 versus 36301 in R20, and 93553 versus 86432 in R23. This consistency suggests the advantage is structural, not workload-specific. The Threadripper also wins Passmark multithread, though by a slimmer 3.1% (104902 versus 101685), and random string sorting by 4.1% (188014 versus 180382).

The single-thread story is even more lopsided. The AMD processor posts 4408 in Passmark single-thread against 3236 for the Intel, a 26.6% lead. That is the largest win for either side in the entire comparison. For workloads that scale poorly across cores, this gap will be decisive.

The Intel Xeon counters with a broad set of Passmark victories. Data compression goes to Intel by 11.5% (1833378 versus 1644573). Data encryption favors Intel by 6.8% (90789 versus 85035). Extended instructions, a proxy for AVX and similar vector workloads, favor Intel by 3.5% (142557 versus 137733). Floating point math shows a substantial 20% Intel advantage (365904 versus 304833). Integer math is nearly a tie, with Intel ahead just 1.5% (468518 versus 461724).

The two most dramatic deltas both belong to Intel. Find prime numbers shows an 85.6% Intel advantage (1151 versus 620), and physics simulation shows an 83.8% Intel lead (13398 versus 7288). These are not marginal differences; they indicate fundamental architectural strengths in specific instruction patterns. Overall, Intel wins 7 of the 14 head-to-head tests, and AMD wins 7, but the wins are not evenly distributed in importance. The AMD victories cluster in general-purpose and single-thread performance, while Intel's cluster in specialized math and encryption.

The average benchmark scores from the wider database put the Intel part at 238263 and the AMD part at 182700, a 23% gap in aggregate. However, that aggregate number is heavily influenced by the Intel part's Passmark wins. The Xeon 6747P sits at the 99th percentile of all CPUs in the database, while the Threadripper PRO 9975WX sits at the 98th percentile. Both are elite parts, but the Intel chip's closest rivals tell a different story: it trails the AMD EPYC 9634 by 2.5% and the Intel Xeon 6980P by 5.3%, while leading the AMD EPYC 9455P by 9.4% and the Intel Xeon w9-3595X by 13.5%. The AMD Threadripper's nearest rivals are closer: it trails the AMD EPYC 8534P by 1.3% and the Intel Xeon 6740E by 2.7%, while leading the AMD EPYC 7763 by 1.5% and the Intel Xeon 6740P by 3.7%. The Xeon 6747P competes in a higher absolute performance tier, despite losing the head-to-head Cinebench tests.

The Verdict

The data points to two different buyers. The Intel Xeon 6747P is the choice for database, compression, encryption, and floating-point-heavy server workloads. Its 7 head-to-head wins include the largest margins in the comparison, and its aggregate benchmark score of 238263 places it above the AMD part by a wide margin. The 99th percentile ranking reinforces that position.

The AMD Ryzen Threadripper PRO 9975WX is the choice for single-thread-bound applications, Cinebench-style rendering, and workloads that respond to raw clock speed. Its 26.6% single-thread lead is the standout number in this comparison. The 32-core part with a 4.00 GHz base clock and 5.40 GHz boost clock simply outruns the Intel part in lightly threaded tasks.

For mixed workloads, the Passmark multithread result (104902 versus 101685, a 3.1% AMD lead) and the Cinebench results suggest the AMD part handles general multithreaded productivity slightly better. But for sustained compute in encryption, compression, and physics, the Intel part's specialized wins make it the safer bet. The user who needs prime number search or physics simulation should not consider the AMD part at all; the 83.8% and 85.6% deficits are prohibitive.

FAQ

Q: Which processor wins more head-to-head benchmarks?

A: Both win 7 of the 14 head-to-head tests. The Intel Xeon 6747P wins data compression, data encryption, extended instructions, find prime numbers, floating point math, integer math, and physics. The AMD Ryzen Threadripper PRO 9975WX wins all three Cinebench multicore tests, Passmark multithread, random string sorting, and both single-thread tests.

Q: How large is the single-thread performance gap?

A: The AMD Ryzen Threadripper PRO 9975WX scores 4408 in Passmark single-thread, which is 26.6% higher than the Intel Xeon 6747P's 3236. This is the largest delta in the comparison.

Q: Does the Intel Xeon have any overwhelming wins?

A: Yes. The Intel part leads by 85.6% in find prime numbers (1151 versus 620) and by 83.8% in physics (13398 versus 7288). It also leads by 20% in floating point math (365904 versus 304833).

Q: How do the two parts compare on Cinebench R23 multicore?

A: The AMD part scores 93553, which is 7.6% higher than the Intel part's 86432. The same 7.6% delta appears in Cinebench R15 and R20 multicore.

Q: What is the aggregate benchmark score difference?

A: The Intel Xeon 6747P has an average benchmark score of 238263, while the AMD Ryzen Threadripper PRO 9975WX has 182700. The Intel part ranks at the 99th percentile of all CPUs, the AMD part at the 98th.

Q: Which processor has better data encryption performance?

A: The Intel Xeon 6747P scores 90789 in Passmark data encryption, which is 6.8% higher than the AMD part's 85035.

Specification Differences

The core counts differ substantially. The Intel Xeon 6747P has 48 cores and 96 threads, while the AMD Ryzen Threadripper PRO 9975WX has 32 cores and 64 threads. Clock speeds also diverge: the Intel part runs at a 2.70 GHz base and 3.90 GHz boost, while the AMD part runs at 4.00 GHz base and 5.40 GHz boost. The Intel part has a 330 TDP, the AMD part 350 TDP.

Sockets differ, with the Intel part on Intel Socket 4710 and the AMD part on AMD Socket sTR5. The Intel part uses a 5 nm process from Intel with a die size of 2x 598 mm². The AMD part uses a 4 nm process from TSMC with a die size of 4x 70.6 mm² and 33,260 million transistors. Cache configurations differ: the Intel part has 112 KB L1 per core, 2 MB L2 per core, and 288 MB shared L3. The AMD part has 64 KB L1 per core, 1 MB L2 per core, and 128 MB L3.

PCIe lane counts differ. The Intel part provides Gen 5 with 88 lanes (CPU only), while the AMD part provides Gen 5 with 128 lanes (CPU only). Memory support is identical in type and bus: both use DDR5 with an eight-channel bus and 409.6 GB/s bandwidth, and both support ECC memory. Neither has integrated graphics.

The multiplier is locked on the Intel part and unlocked on the AMD part. Release dates differ, with the Intel part launching on 2025-02-23 and the AMD part on 2025-07-22. The launch MSRP for the Intel Xeon 6747P is $6497. The launch MSRP for the AMD Ryzen Threadripper PRO 9975WX is $4099. The part numbers are SRVEZ for Intel and 100-000000723 for AMD.

Architecture Differences

The Intel Xeon 6747P uses Granite Rapids architecture, part of the Xeon 6 generation (Granite Rapids-SP). The AMD Ryzen Threadripper PRO 9975WX uses Zen 5 architecture, codenamed Shimada Peak, from the Ryzen Threadripper 9000 series.

The process nodes differ: Intel uses a 5 nm process at its own foundry, while AMD uses a 4 nm process at TSMC. The die configurations are dramatically different. The Intel part uses two large dies totaling 2x 598 mm², while the AMD part uses four small dies totaling 4x 70.6 mm². The AMD part has a transistor count of 33,260 million; no transistor count is recorded for the Intel part.

Cache hierarchy reflects the core count difference. The Intel part has larger per-core L1 (112 KB versus 64 KB) and L2 (2 MB versus 1 MB), and more than double the shared L3 (288 MB versus 128 MB). These cache advantages likely contribute to the Intel part's wins in data compression and encryption, which benefit from large working sets.

The AMD part compensates with much higher clock speeds. Its 4.00 GHz base clock exceeds the Intel part's 3.90 GHz boost clock. The 5.40 GHz boost clock is a substantial advantage in single-thread and lightly threaded workloads. The unlocked multiplier on the AMD part adds overclocking headroom, which the locked Intel part does not offer.

The PCIe difference is notable: 128 lanes on the AMD part versus 88 on the Intel part. This gives the AMD platform more room for expansion cards, NVMe storage, and accelerators. Both parts target server and workstation segments, and both are Active in production status.

Where Each One Wins

The Intel Xeon 6747P wins in data compression by 11.5%, data encryption by 6.8%, extended instructions by 3.5%, find prime numbers by 85.6%, floating point math by 20%, integer math by 1.5%, and physics by 83.8%. These are compute-heavy, often vectorized workloads. The large L3 cache and 48-core count make it the stronger choice for database workloads, compression pipelines, scientific computing with floating point math, and any task that involves prime number generation or physics simulation. Its 99th percentile ranking and aggregate score of 238263 place it in a higher overall performance tier.

The AMD Ryzen Threadripper PRO 9975WX wins in Cinebench R15, R20, and R23 multicore by 7.6% each, Passmark multithread by 3.1%, random string sorting by 4.1%, and single-thread by 26.6%. The single-thread result is the clearest signal: any workload that relies on one or a few cores will run substantially faster on the AMD part. The Cinebench wins suggest rendering and content creation workloads favor AMD. The 4.00 GHz base clock and 5.40 GHz boost clock, combined with the unlocked multiplier, make this the part for interactive workstation use, software compilation with many serial sections, and applications that do not scale to 48 cores. The 128 PCIe lanes also make it the better fit for systems with many expansion devices.

The split is clean. Intel for throughput in specialized math and data processing, AMD for frequency-sensitive and general-purpose multithreaded work. The 7-7 win split in head-to-head tests does not tell the whole story; the magnitude of Intel's wins in physics and prime numbers outweighs the AMD wins in Cinebench and single-thread, at least for the workloads where those benchmarks matter. The aggregate benchmark scores reinforce this: the Intel part leads by a significant margin in overall measured performance.