AMD Ryzen 9 9850HX vs AMD Ryzen AI Embedded P164 Comparison

Head-to-Head Benchmarks

The recorded data shows a decisive sweep for the AMD Ryzen 9 9850HX across all 11 shared benchmark tests, with no wins recorded for the AMD Ryzen AI Embedded P164. The largest margin appears in prime number generation, where the 9850HX scores 323 versus 71, a 354.9% advantage. This is the kind of result that indicates a fundamental throughput gap, not a marginal clock advantage.

Multi-threaded workloads follow the same pattern. The 9850HX delivers 51,722 in PassMark multithread, while the P164 manages 25,889, a 99.8% difference. Integer math shows a 96.7% lead (172,943 versus 87,940), and floating-point math comes in at 106.2% ahead (115,062 versus 55,799). Data compression favors the 9850HX by 102%, with scores of 662,381 and 327,891, respectively. These figures align closely with the core and thread disparity between the two processors.

Extended instruction performance is also heavily skewed. The 9850HX posts 53,817, against 24,193 for the P164, a 122.4% delta. Encryption workloads show a 100.2% lead (32,139 versus 16,055), while random string sorting lands at 101.6% (70,175 versus 34,801). Physics simulation scores 3,054 versus 1,210, a 152.4% gap that suggests the 9850HX handles complex physical calculations with substantially more headroom.

The single-thread results are the closest comparison in the entire dataset. The 9850HX scores 4,461, and the P164 scores 4,029, a 10.7% lead for the former. This is a modest margin compared to the multi-thread tests, which indicates that per-core efficiency is comparable, but the 9850HX still holds the advantage. The average benchmark score reinforces the overall picture: 106,413 for the 9850HX against 52,901 for the P164, roughly a 101% difference.

Percentile placement confirms the gap. The 9850HX ranks in the 97th percentile among all CPUs, while the P164 sits in the 91st percentile. The nearest rivals for the 9850HX include the Intel Xeon w7-3555, which trails by 0.2%, and the Intel Xeon 6521P, which is 0.5% behind, while the Intel Xeon w7-2595X leads it by 2.1% and the AMD EPYC 8324P trails by 3%. For the P164, the AMD Ryzen 5 9500F is 0.1% behind, the Intel Xeon 634 is 0.1% ahead, the AMD EPYC 7313P is 0.6% ahead, and the AMD Ryzen 9 7900X is 0.7% ahead. These rival positions show that the 9850HX competes in workstation-class territory, while the P164 is aligned with mainstream desktop and embedded parts.

Architecture Differences

Both processors are built on TSMC's 4 nm node, but the silicon layouts diverge sharply. The 9850HX uses the Zen 5 architecture under the Fire Range codename, while the P164 uses a hybrid configuration described as Zen 5 / Zen 5c under the Gorgon Point codename. The 9850HX is a 12-core, 24-thread part with a base clock of 3.00 GHz and a boost clock of 5.20 GHz. The P164 is an 8-core, 16-thread part with a base clock of 2.00 GHz and a boost clock of 5.00 GHz.

Cache allocation is a major differentiator. The 9850HX carries 64 MB of L3 cache, while the P164 has only 8 MB. Both use 80 KB of L1 per core and 1 MB of L2 per core, so the L3 difference directly impacts workloads that benefit from large shared pools of data. The die size reflects this: the 9850HX uses a dual-chiplet design with 2x 70.6 mm² dies, while the P164 is a monolithic 233 mm² die. The 9850HX also lists 16,630 million transistors, while the P164 does not include a transistor count in the database.

The integrated graphics differ as well. The 9850HX uses the Radeon 610M, while the P164 uses the Radeon 880M. No benchmark scores are provided for either iGPU, so direct graphics performance cannot be quantified, but the model difference indicates a different GPU block. Memory support also separates the two: the 9850HX supports DDR5 only, while the P164 supports both DDR5 and LPDDR5X. Both run dual-channel memory at 89.6 GB/s and both support ECC memory.

PCIe connectivity is another divergence. The 9850HX offers Gen 5 with 28 lanes from the CPU, whereas the P164 offers Gen 4 with 16 lanes. This positions the 9850HX for higher-bandwidth expansion, while the P164 targets more constrained embedded or mobile layouts. The sockets are different: the 9850HX uses AMD Socket FL1, and the P164 uses AMD Socket FP8. The multiplier is unlocked on the 9850HX, but locked on the P164.

Where Each One Wins

The 9850HX wins every recorded benchmark, so the analysis is about the degree of advantage rather than distribution of wins. The largest leads appear in prime number generation and physics simulation, both above 150%. These are compute-heavy, latency-sensitive tasks that respond to the 9850HX's higher core count, larger L3 cache, and superior boost clock. Multi-threaded integer and floating-point workloads also favor the 9850HX by roughly 100%, making it the stronger choice for rendering, simulation, and data processing.

The P164's closest result is single-thread performance, where it trails by only 10.7%. This suggests that for lightly threaded tasks, the P164 remains competitive, though still behind. Its lower TDP of 28 watts, compared to 55 watts for the 9850HX, indicates an efficiency-oriented design. The P164 supports LPDDR5X memory in addition to DDR5, which can be relevant for compact or power-constrained systems. Its Radeon 880M integrated graphics may also offer a different feature set than the Radeon 610M, though no benchmark data is available to confirm a performance advantage.

The 9850HX, with its 64 MB L3 cache and 12 Zen 5 cores, is positioned for high-throughput mobile workstations. The P164, with 8 cores and 8 MB L3, is aimed at embedded applications where power draw and physical footprint matter more than raw throughput. The 28-lane Gen 5 PCIe support on the 9850HX further separates it from the P164's 16-lane Gen 4 interface.

Specification Differences

The core and thread counts are the most obvious difference: 12 cores and 24 threads on the 9850HX, versus 8 cores and 16 threads on the P164. Base clocks differ by 1.00 GHz (3.00 GHz versus 2.00 GHz), and boost clocks differ by 0.20 GHz (5.20 GHz versus 5.00 GHz). TDP is nearly halved on the P164: 28 watts versus 55 watts.

L3 cache is 64 MB on the 9850HX and 8 MB on the P164. Die size is 2x 70.6 mm² versus 233 mm². The 9850HX lists 16,630 million transistors, while the P164 has no recorded transistor count. Memory support is DDR5 for the 9850HX, and DDR5 plus LPDDR5X for the P164. PCIe is Gen 5 with 28 lanes on the 9850HX, and Gen 4 with 16 lanes on the P164. The integrated GPU is Radeon 610M on the 9850HX and Radeon 880M on the P164. Sockets are FL1 and FP8, respectively. The 9850HX has an unlocked multiplier, the P164 does not. Release dates are 2025-01-05 for the 9850HX and 2026-03-08 for the P164. Neither processor has a recorded launch MSRP in the database.

FAQ

Q: Which processor has more cores?

A: The AMD Ryzen 9 9850HX has 12 cores and 24 threads, while the AMD Ryzen AI Embedded P164 has 8 cores and 16 threads.

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

A: The 9850HX scores 4,461 in PassMark single-thread, and the P164 scores 4,029, a 10.7% advantage for the 9850HX.

Q: Do both processors support ECC memory?

A: Yes, both the 9850HX and the P164 support ECC memory.

Q: What is the L3 cache difference?

A: The 9850HX has 64 MB of L3 cache, while the P164 has 8 MB.

Q: Which processor supports LPDDR5X memory?

A: The AMD Ryzen AI Embedded P164 supports both DDR5 and LPDDR5X, while the 9850HX supports DDR5 only.

Q: How do the average benchmark scores compare?

A: The 9850HX has an average benchmark score of 106,413, and the P164 has an average of 52,901. The 9850HX ranks in the 97th percentile, while the P164 ranks in the 91st percentile.