Intel Xeon 6731P vs Intel Xeon 6736P Comparison

The Verdict

The data splits these two Granite Rapids Xeons into clear roles. The Intel Xeon 6736P wins exactly 3 of 17 head-to-head benchmarks, while the Intel Xeon 6731P takes 14. That is not a close contest on paper. But the 6736P's victories are not trivial — they are concentrated in encryption, integer math, and random string sorting, which points to specific workloads where its larger core count (36 vs 32) and higher thread count (72 vs 64) overpower the 6731P's clock advantage.

For most general multi-threaded server work, the 6731P is the stronger pick. It leads in every Cinebench test (R15, R20, R23) by roughly 5% in both single-core and multi-core runs, and it also wins PassMark multithread (52790 vs 50072) and physics (7105 vs 6531). If your job is rendering, physics simulation, or any task that scales with raw core throughput and per-thread speed, the 6731P delivers more performance per chip. Its 2.50 GHz base clock versus the 6736P's 2.00 GHz base clock is the obvious driver here, since both boost to 4.10 GHz and share the same 144 MB L3 cache.

The 6736P is the specialist. It wins PassMark data encryption by a decisive 15.3% (46236 vs 40087), random string sorting by 17.8% (103723 vs 88019), and integer math by 4.1% (206833 vs 198761). Those wins are not flukes — they align with the 6736P's extra 4 cores and 8 threads. If your workload is encryption-heavy, database-style sorting, or large integer calculations, the 6736P justifies its presence despite losing the overall benchmark count. It also carries a higher average benchmark score (87864 vs 87756), though the delta is just 0.1%, which is effectively a tie.

The verdict: choose the 6731P for balanced, high-frequency multi-core compute across Cinebench and general PassMark workloads. Choose the 6736P only when you know your primary bottleneck is encryption, sorting, or integer-heavy math, where its core advantage turns into real wins. For everything else, the 6731P's lower core count but higher base clock proves more useful.

Where Each One Wins

The 6731P dominates the synthetic render and compute suite. In Cinebench R23 multi-core, it scores 44871 against the 6736P's 42561 — a 5.1% lead. The same margin appears in R20 multi-core (18845 vs 17875) and R15 multi-core (4522 vs 4290). Single-core is consistent too: 6334 vs 6008 in R23, 2660 vs 2523 in R20, and 638 vs 605 in R15. The pattern is uniform: the 6731P is about 5% faster in every Cinebench test, both single and multi-threaded. PassMark multithread (52790 vs 50072) and physics (7105 vs 6531) follow the same trend, with the 6731P ahead by 5.1% and 8.1% respectively.

The 6736P's wins are narrower in count but larger in margin. Its biggest victory is random string sorting at 17.8% (103723 vs 88019), followed by data encryption at 15.3% (46236 vs 40087). Integer math is the only other win, at 4.1% (206833 vs 198761). These are not marginal edges; they are substantial double-digit gaps in two cases. The 6736P also stays competitive in data compression, losing only 0.4% (796658 vs 799474) — effectively a tie.

Notably, the 6731P wins extended instructions by 15.4% (65656 vs 55563) and find prime numbers by 27.5% (541 vs 392). Those are massive margins that suggest the 6731P's higher base clock and per-core efficiency matter more than raw core count in instruction-heavy and prime-number workloads. The 6736P cannot keep up there, despite having more cores.

Architecture Differences

Both chips are built on the same Granite Rapids architecture, Intel Socket 4710, and a 5 nm process with a 598 mm² die. They share identical cache hierarchies: 112 KB L1 per core, 2 MB L2 per core, and 144 MB shared L3. Memory support is also identical — DDR5, eight-channel, 409.6 GB/s bandwidth, with ECC enabled. No integrated graphics on either. Both launch on the same date, 2025-02-23, and both are active production parts.

The real differences are core count, clock speed, TDP, PCIe lanes, and price. The 6736P has 36 cores and 72 threads; the 6731P has 32 cores and 64 threads. That is a 4-core, 8-thread advantage for the 6736P. But the 6731P compensates with a 2.50 GHz base clock versus 2.00 GHz, while both boost to 4.10 GHz. The 6731P draws more power — 245 W TDP versus 205 W — and offers more PCIe lanes: 136 Gen 5 lanes (CPU only) versus 88. The 6736P has a higher launch MSRP of $3351 versus $2700 for the 6731P.

Neither chip has an unlocked multiplier, so overclocking is not a factor. The part numbers differ (SRVNW for the 6736P, SRVNR for the 6731P), but that is just a SKU identifier. In short: the 6736P trades higher power draw and more PCIe lanes for more cores, while the 6731P uses fewer cores but a much higher base clock to win most benchmarks.

FAQ

Q: Which CPU is faster in Cinebench R23 multi-core?

A: The Intel Xeon 6731P. It scores 44871 versus the 6736P's 42561, a 5.1% lead.

Q: Does the 6736P ever win any benchmark?

A: Yes, it wins 3 of 17 head-to-head tests: PassMark data encryption (46236 vs 40087, +15.3%), PassMark random string sorting (103723 vs 88019, +17.8%), and PassMark integer math (206833 vs 198761, +4.1%).

Q: Why does the 6731P win most benchmarks despite having fewer cores?

A: The 6731P has a 2.50 GHz base clock versus the 6736P's 2.00 GHz, and both boost to 4.10 GHz. That higher base clock gives it a per-core speed advantage that most benchmarks favor.

Q: Are these CPUs the same architecture?

A: Yes. Both are Granite Rapids, on a 5 nm process, with the same 598 mm² die, 144 MB L3 cache, and eight-channel DDR5 memory support.

Q: Which CPU has more PCIe lanes?

A: The 6731P has 136 Gen 5 lanes (CPU only), while the 6736P has 88 Gen 5 lanes. This is a significant difference for expansion-heavy servers.

Q: Is the 6736P worth more than the 6731P?

A: Based on the data, the 6736P has a higher launch MSRP ($3351 vs $2700) but wins only 3 of 17 benchmarks. Unless your workload is encryption, sorting, or integer math, the 6731P is the better performer.

Head-to-Head Benchmarks

The biggest win for the 6731P is PassMark find prime numbers, where it scores 541 against the 6736P's 392 — a massive 27.5% lead. That is not a small edge; it is a dominance in a specific computational task. The 6731P also crushes extended instructions by 15.4% (65656 vs 55563) and physics by 8.1% (7105 vs 6531). These are not marginal differences. The Cinebench suite is uniformly 5.1-5.2% in the 6731P's favor across all six tests, which confirms a consistent per-core advantage rather than a workload-specific fluke.

The 6736P's largest win is random string sorting at 17.8% (103723 vs 88019). That is a substantial margin for a memory-bandwidth-heavy operation, and it suggests the extra cores help when the workload is parallelizable sorting. Data encryption is close behind at 15.3% (46236 vs 40087), which is a common server workload — so this win matters in real deployments. Integer math is the smallest 6736P win at 4.1% (206833 vs 198761), but still a clear positive.

Data compression is the closest test overall: 799474 for the 6731P versus 796658 for the 6736P, a 0.4% difference that is effectively a tie. Floating point math goes to the 6731P by 4.8% (157330 vs 149770), and single-thread tests are consistently 3.9% in its favor (2107 vs 2024). The multithread score (52790 vs 50072) is a 5.1% win for the 6731P, mirroring the Cinebench results.

The overall benchmark count — 14 wins for the 6731P, 3 for the 6736P — tells the story. But the 6736P's wins are in categories that are often mission-critical for servers: encryption and sorting. If your workload is database-heavy or security-focused, those double-digit margins could outweigh the 5% losses everywhere else. For general compute, the 6731P is the clear winner.

Specification Differences

| Field | Intel Xeon 6736P | Intel Xeon 6731P |

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

| Cores | 36 | 32 |

| Threads | 72 | 64 |

| Base Clock | 2.00 GHz | 2.50 GHz |

| TDP | 205 W | 245 W |

| PCIe | Gen 5, 88 Lanes (CPU only) | Gen 5, 136 Lanes (CPU only) |

| Launch MSRP | $3351 | $2700 |

| Part Number | SRVNW | SRVNR |

Everything else — architecture (Granite Rapids), socket (Intel Socket 4710), process node (5 nm), die size (598 mm²), cache (112 KB L1/core, 2 MB L2/core, 144 MB L3), memory (DDR5, eight-channel, 409.6 GB/s), ECC support, boost clock (4.10 GHz), integrated graphics (N/A), market segment (Server/Workstation), production status (Active), release date (2025-02-23), and locked multiplier — is identical. The 6731P uses more power and has more PCIe lanes, while the 6736P has more cores but a lower base clock. The 6731P also costs less at launch, making its benchmark dominance even more notable from a selection standpoint.