AMD Ryzen 5 PRO 8640U vs AMD Ryzen 9 8945HS Comparison

The Verdict

The recorded data separates these two mobile processors clearly by workload profile. The AMD Ryzen 9 8945HS wins 12 of 15 head-to-head benchmark comparisons, while the AMD Ryzen 5 PRO 8640U takes 3. The Ryzen 9 leads in almost every throughput-oriented test, often by margins between 17.9% and 41%. However, the Ryzen 5 PRO wins the Cinebench R23 multicore and singlecore tests, including a 36% advantage in single-core performance. The database places the Ryzen 9 at the 82nd percentile among all CPUs, with an average benchmark score of 31074, while the Ryzen 5 PRO sits at the 81st percentile with an average of 30254. These percentile positions are nearly identical, meaning the overall performance class is similar despite the different core counts.

For a user prioritizing sustained multi-threaded workloads, data compression, encryption, or extended instruction throughput, the Ryzen 9 8945HS is the clear choice from the measurements. For workloads that favor Cinebench R23, particularly single-threaded responsiveness, the Ryzen 5 PRO 8640U shows a surprising edge. The Ryzen 5 PRO also supports ECC memory, a feature absent on the Ryzen 9, which matters for certain professional or reliability-focused use cases.

Where Each One Wins

The Ryzen 9 8945HS dominates the PassMark suite. It wins data compression by 36.6%, data encryption by 34.6%, extended instructions by 41%, floating point math by 36.6%, integer math by 35.2%, multithread by 30.6%, physics by 24.3%, random string sorting by 34.5%, and find prime numbers by 17.9%. It also takes the Cinebench R15 multicore test by 31.1%. These are all compute-heavy tasks where additional cores and higher boost clocks translate directly into throughput.

The Ryzen 5 PRO 8640U wins Cinebench R23 multicore by 15.9% and singlecore by 36%, plus a narrow 1.1% margin in Cinebench R15 singlecore. The R23 results are especially notable because they contradict the general trend of the Ryzen 9 leading multi-threaded tests. This suggests the Ryzen 5 PRO has a more efficient implementation under certain sustained loads, possibly due to its lower 28W TDP versus the Ryzen 9's 45W TDP, which may allow better thermal management in compact chassis.

The Ryzen 9 also edges out the Ryzen 5 PRO in PassMark single-thread by 3.2%, so the single-core story is not uniform. The Ryzen 5 PRO only wins in Cinebench's specific single-thread measurement, not in PassMark's equivalent test. Users should match the benchmark to the application: Cinebench R23 favors the Ryzen 5 PRO, but PassMark single-thread favors the Ryzen 9.

Architecture Differences

Both processors share the same fundamental architecture: Zen 4, codename Hawk Point, manufactured on TSMC's 4nm process. Both have 25,000 million transistors and a die size of 178 mm². The cache hierarchies are identical in structure: 64 KB L1 per core, 1 MB L2 per core, and 16 MB shared L3. Both support dual-channel DDR5 memory with 89.6 GB/s bandwidth and PCIe Gen 4 with 20 lanes from the CPU.

The core configuration differs significantly. The Ryzen 9 8945HS has 8 cores and 16 threads, while the Ryzen 5 PRO 8640U has 6 cores and 12 threads. Base clocks are 4.00 GHz versus 3.50 GHz, and boost clocks are 5.20 GHz versus 4.90 GHz. The TDP also differs: 45W for the Ryzen 9 versus 28W for the Ryzen 5 PRO.

Socket types diverge: the Ryzen 9 uses AMD Socket FP8, while the Ryzen 5 PRO uses AMD Socket FP7. This has practical implications for motherboard compatibility, as the two are not interchangeable. The integrated graphics differ as well: the Ryzen 9 features Radeon 780M, while the Ryzen 5 PRO features Radeon 760M. The Ryzen 5 PRO supports ECC memory, which the Ryzen 9 does not. Release dates are also different, with the Ryzen 9 launching on December 5, 2023, and the Ryzen 5 PRO on April 15, 2024.

FAQ

Q: Which processor has more cores and threads?

A: The AMD Ryzen 9 8945HS has 8 cores and 16 threads, while the AMD Ryzen 5 PRO 8640U has 6 cores and 12 threads.

Q: Why does the Ryzen 5 PRO win Cinebench R23 multicore despite having fewer cores?

A: The recorded Cinebench R23 multicore score for the Ryzen 5 PRO is 19982, which is 15.9% higher than the Ryzen 9's 16795. The data does not provide a direct cause, but the lower 28W TDP of the Ryzen 5 PRO may enable more consistent sustained clock behavior in this specific benchmark, whereas the Ryzen 9's 45W TDP could lead to thermal throttling under the same test conditions.

Q: Does the Ryzen 9 8945HS support ECC memory?

A: No, the Ryzen 9 8945HS does not support ECC memory. The Ryzen 5 PRO 8640U does support ECC memory, making it the only one of the two with that feature.

Q: Which processor has the higher boost clock?

A: The Ryzen 9 8945HS has a boost clock of 5.20 GHz, while the Ryzen 5 PRO 8640U has a boost clock of 4.90 GHz. The base clocks are 4.00 GHz and 3.50 GHz, respectively.

Q: Are these processors on the same socket?

A: No. The Ryzen 9 8945HS uses AMD Socket FP8, while the Ryzen 5 PRO 8640U uses AMD Socket FP7. They are not socket-compatible with each other.

Q: Which processor has the better integrated graphics?

A: The Ryzen 9 8945HS has Radeon 780M integrated graphics, while the Ryzen 5 PRO 8640U has Radeon 760M. The database does not include graphics benchmarks, so only the model names can be compared.

Head-to-Head Benchmarks

The largest win for the Ryzen 9 8945HS comes in PassMark extended instructions, where it scores 26964 against the Ryzen 5 PRO's 19130, a 41% advantage. This test likely stresses AVX or similar instruction sets, where the Ryzen 9's higher boost clock and extra cores provide a substantial lead.

Data compression and floating point math both show 36.6% margins in favor of the Ryzen 9. The scores are 356508 versus 260925 for compression, and 61821 versus 45265 for floating point. Integer math follows at 35.2% (101226 versus 74875). Data encryption shows a 34.6% edge (21323 versus 15843). Random string sorting is 34.5% higher for the Ryzen 9 (43202 versus 32114). The multithread PassMark test shows a 30.6% lead (29780 versus 22795). Cinebench R15 multicore shows a 31.1% margin (2640 versus 2014). Physics is 24.3% higher (1434 versus 1154). Find prime numbers is 17.9% higher (92 versus 78). PassMark single-thread shows a smaller 3.2% edge for the Ryzen 9 (3850 versus 3732).

The Ryzen 5 PRO's wins are concentrated in Cinebench R23. Its multicore score of 19982 beats the Ryzen 9's 16795 by 15.9%. The singlecore score of 2821 beats 1805 by 36%, which is the largest single margin in either direction across all tests. Cinebench R15 singlecore is a narrow 1.1% win for the Ryzen 5 PRO (284 versus 281).

These results create an unusual pattern: the Ryzen 9 wins every PassMark test and the older Cinebench R15 multicore, but the Ryzen 5 PRO dominates the newer Cinebench R23 versions. This suggests the Ryzen 5 PRO's architecture, despite fewer cores and lower clocks, may have better sustained performance in Cinebench's specific rendering workload, possibly due to thermal behavior at its 28W TDP.

Specification Differences

| Specification | AMD Ryzen 9 8945HS | AMD Ryzen 5 PRO 8640U |

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

| Cores | 8 | 6 |

| Threads | 16 | 12 |

| Base Clock | 4.00 GHz | 3.50 GHz |

| Boost Clock | 5.20 GHz | 4.90 GHz |

| TDP | 45W | 28W |

| Socket | AMD Socket FP8 | AMD Socket FP7 |

| ECC Memory | No | Yes |

| Integrated Graphics | Radeon 780M | Radeon 760M |

| Release Date | December 5, 2023 | April 15, 2024 |

| Part Number | 100-000001319(FP7r2), 100-000001383(FP7), 100-000001309(FP8) | 100-000001318(FP7r2), 100-000001378(FP7) |

The shared specifications include architecture (Zen 4), codename (Hawk Point), process node (4nm), foundry (TSMC), transistors (25,000 million), die size (178 mm²), L1 cache (64 KB per core), L2 cache (1 MB per core), L3 cache (16 MB shared), memory support (DDR5), memory bus (dual-channel), memory bandwidth (89.6 GB/s), PCIe (Gen 4, 20 lanes), market segment (mobile), production status (active), and unlocked multiplier (false for both). These commonalities mean the performance differences observed in benchmarks stem primarily from core count, clock speeds, TDP, and the specific silicon binning, not from fundamental architectural changes.