AMD Ryzen Threadripper PRO 5975WX vs Intel Xeon 6732P Comparison

The Intel Xeon 6732P and the AMD Ryzen Threadripper PRO 5975WX make for one of the closest matchups in the workstation database. Both are 32-core, 64-thread processors aimed at the same server and workstation segment, and both sit in the 97th percentile or higher of all recorded CPUs. Yet the recorded data shows a peculiar split: the Threadripper PRO 5975WX wins more individual tests, 11 to 6, while the Xeon 6732P posts some of the largest single-test margins on the page. The question worth investigating is which pattern matters more for a given workload.

FAQ

Q: Which CPU is faster overall?

A: The database shows a near-tie with a tilt toward AMD on breadth. The 5975WX wins 11 of 17 head-to-head tests, including every Cinebench test and the overall PassMark multithread score (75319 vs 74849). However, the Xeon 6732P posts the larger average benchmark score, 143444 vs 124171, because its wins in tests like physics and prime number computation come with enormous margins.

Q: How do they compare in single-core performance?

A: The picture is mixed. In Cinebench single-core tests the two are nearly identical, with the 5975WX ahead by roughly 0.5 percent in R15, R20, and R23 (9038 vs 8981 in R23). In PassMark single-thread the 5975WX is decisively ahead, 3323 vs 2506, a 24.6 percent gap in its favor.

Q: Which CPU has the memory bandwidth advantage?

A: The Xeon 6732P, by a wide margin. Its DDR5 eight-channel platform delivers 409.6 GB/s of bandwidth versus 204.8 GB/s on the DDR4-based 5975WX, exactly double.

Q: How do the PCIe platforms differ?

A: The Xeon 6732P offers Gen 5 with 136 CPU lanes, while the 5975WX offers Gen 4 with 128 CPU lanes. Both are aimed at GPU- and accelerator-heavy workstation builds.

Q: Which one uses less power?

A: The 5975WX has the lower TDP at 280 W versus 350 W for the Xeon 6732P.

Q: Are both CPUs still in production?

A: Yes, the database lists both as Active parts in the Server/Workstation segment, despite the 5975WX dating to a 2022-03-07 release and the Xeon to 2025-05-21.

Architecture Differences

These two processors come from different eras of silicon. The Xeon 6732P is a Granite Rapids-SP design, part of the Xeon 6 generation, built on Intel's 5 nm process. The Threadripper PRO 5975WX is a Zen 3 architecture part, codename Chagall PRO, from the 5000 series, fabricated on TSMC's 7 nm node. The 5975WX is a chiplet design documented as four dies of 81 mm² each with 16,600 million transistors; the database records no transistor or die-size figures for the Xeon.

The cache hierarchy differs substantially. The Xeon gives each core 112 KB of L1 and 2 MB of L2, with 144 MB of shared L3. The 5975WX pairs 64 KB of L1 and 512 KB of L2 per core with 128 MB of L3. The Xeon's approach front-loads far more cache per core, which raises an interesting question: does that larger per-core L2 partly explain its strength in sustained compute tests like physics, where it leads by 86 percent?

Platform connectivity is another divide. The Xeon runs DDR5 across eight channels at 409.6 GB/s; the 5975WX runs DDR4 across eight channels at 204.8 GB/s. Both support ECC memory. On expansion, the Xeon offers PCIe Gen 5 with 136 lanes against Gen 4 with 128 lanes on the 5975WX. Neither part has a multiplier unlocked, and neither ships with integrated graphics.

Clock behavior splits the difference. The Xeon has the higher base clock at 3.80 GHz versus 3.60 GHz, but the 5975WX boosts considerably higher, 4.50 GHz versus 4.10 GHz, which aligns neatly with its PassMark single-thread win.

Head-to-Head Benchmarks

The Cinebench suite is remarkably uniform: the 5975WX wins all six tests by 0.5 to 0.6 percent. That is R15 multi (6453 vs 6412), R15 single (910 vs 905), R20 multi (26888 vs 26720), R20 single (3795 vs 3772), R23 multi (64021 vs 63621), and R23 single (9038 vs 8981). In rendering, these two are functionally interchangeable.

The PassMark suite is where the data gets strange and interesting. The 5975WX dominates encryption at 80453 vs 63848, a 20.6 percent win, and single-thread at 3323 vs 2506, up 24.6 percent. It also takes integer math, 360156 vs 334340, by 7.2 percent, plus the overall multithread score, 75319 vs 74849.

The Xeon counters with some of the biggest margins recorded in this comparison. Physics is the outlier: 8109 vs 4360, an 86 percent lead. Find prime numbers goes to the Xeon 628 to 438, up 43.4 percent. Floating point math lands at 261703 vs 202363, a 29.3 percent Xeon win, and extended instructions at 106697 vs 82850, up 28.8 percent. Data compression favors the Xeon 1339480 to 1293784 by 3.5 percent, and string sorting 133467 vs 123527 by 8 percent.

What does that pattern imply? The Xeon's wins cluster in bandwidth-hungry and floating-point-heavy kernels, consistent with its doubled memory bandwidth and per-core cache. The 5975WX's wins cluster in lightly threaded, clock-sensitive, and encryption workloads, consistent with its higher boost clock. Neither sweeps the board; each exploits a different structural advantage.

Specification Differences

• Base clock: Xeon 6732P at 3.80 GHz, 5975WX at 3.60 GHz
• Boost clock: 5975WX at 4.50 GHz, Xeon at 4.10 GHz
• TDP: 5975WX at 280 W, Xeon at 350 W
• Process node: Xeon at 5 nm (Intel), 5975WX at 7 nm (TSMC)
• Architecture: Granite Rapids vs Zen 3 (Chagall PRO)
• Socket: Intel Socket 4710 vs AMD Socket WRX8
• L1 cache: 112 KB per core vs 64 KB per core
• L2 cache: 2 MB per core vs 512 KB per core
• L3 cache: 144 MB shared vs 128 MB
• Memory: DDR5 at 409.6 GB/s vs DDR4 at 204.8 GB/s, both eight-channel with ECC
• PCIe: Gen 5, 136 lanes vs Gen 4, 128 lanes
• Launch MSRP: $5295 for the Xeon 6732P, $3299 for the 5975WX
• Release: 2025-05-21 vs 2022-03-07

Cores and threads are identical at 32 and 64, both parts support ECC, and both are locked.

Where Each One Wins

The 5975WX is the pick where the data shows it winning: encryption-heavy pipelines (20.6 percent ahead), integer workloads (7.2 percent ahead), and any workload with meaningful single-threaded portions (24.6 percent ahead in PassMark single-thread). Rendering is a wash, but the 5975WX holds the slimmest of edges in every Cinebench variant, and it takes the aggregate PassMark multithread score while doing so at a lower TDP.

The Xeon 6732P owns the memory-bound and numeric-compute territory. Its 86 percent physics lead is the single largest gap in the dataset, reinforced by wins in floating point math (29.3 percent), prime number computation (43.4 percent), extended instructions (28.8 percent), compression (3.5 percent), and string sorting (8 percent). Add double the memory bandwidth, faster Gen 5 expansion with more lanes, and 144 MB of L3, and the platform case strengthens further for data-intensive builds.

Context from the rivals list reinforces how close these parts sit to the performance frontier: the Xeon's average score of 143444 places it within 1 percent of the AMD EPYC 7643P and within 0.6 percent of the Intel Xeon w9-3575X, while the 5975WX's 124171 lands it essentially level with the AMD EPYC 7642.

The Verdict

The data splits cleanly along workload lines. If the job mix is rendering, encryption, integer compute, or anything sensitive to per-thread responsiveness, the Threadripper PRO 5975WX delivers equal or better results across 11 of 17 recorded tests, with a lower TDP. If the job mix leans on floating point throughput, physics simulation, prime computation, or saturating memory bandwidth, the Xeon 6732P is clearly the stronger recorded performer, sometimes by enormous margins.

The Xeon is the newer platform by three years, with DDR5, Gen 5 PCIe, and 136 lanes, which matters for expansion-heavy systems. The 5975WX offers a proven platform with higher boost clocks. Curiously, the raw win count favors AMD while the largest single-test margins favor Intel, and that asymmetry is the real story: this is not a ranking question but a workload question, and the benchmark record answers it decisively once the workload is known.