AMD Ryzen AI 9 HX 370 vs AMD Ryzen AI Max 385 Comparison

AMD
AMD

AMD Ryzen AI 9 HX 370

CORE STATE Strix Point
CORE SPECS 12 Cores / 24 Threads
CLOCK SPEED 2 Base / 5.1 GHz Turbo
CACHE 16 MB
MAX TDP 28W
ARCHITECTURE Zen 5
nm
PROCESS 4 nm
LAUNCH DATE 2024
VS
AMD
AMD

Ryzen AI Max 385

CORE STATE Strix Halo
CORE SPECS 8 Cores / 16 Threads
CLOCK SPEED 3.6 Base / 5 GHz Turbo
CACHE 32 MB (shared)
MAX TDP 55W
ARCHITECTURE Zen 5
nm
PROCESS 4 nm
LAUNCH DATE 2025

PERFORMANCE BENCHMARKS

3dmark_16_threads
9,365
N/A
3dmark_2_threads
2,179
N/A
3dmark_4_threads
3,970
N/A
3dmark_8_threads
6,992
N/A
3dmark_max_threads
9,725
N/A
3dmark_single_thread
1,153
N/A
cinebench_cinebench_r15_multicore
3,409.5
1,579
cinebench_cinebench_r15_singlecore
306.5
222
cinebench_cinebench_r23_multicore
21,761
15,674
cinebench_cinebench_r23_singlecore
2,018
2,212
geekbench_multicore
15,242
N/A
geekbench_singlecore
2,407
N/A
passmark_data_compression
445,719
406,505
passmark_data_encryption
22,252
19,926
passmark_extended_instructions
31,653
33,873
passmark_find_prime_numbers
126
165
passmark_floating_point_math
76,892
71,105
passmark_integer_math
122,933
107,046
passmark_multithread
35,148
33,705
passmark_physics
1,881
1,889
passmark_random_string_sorting
48,723
43,725
passmark_single_thread
3,973
4,060
passmark_singlethread
3,973
4,060
cinebench_cinebench_r20_multicore
N/A
6,583
cinebench_cinebench_r20_singlecore
N/A
929

Analysis: AMD Ryzen AI 9 HX 370 vs AMD Ryzen AI Max 385

Head-to-Head Benchmarks

The recorded head-to-head data presents a clear split between two very different processor designs. The AMD Ryzen AI 9 HX 370 wins 9 of the 15 direct comparisons, while the AMD Ryzen AI Max 385 takes 6. The magnitude of those wins tells a more nuanced story than the raw win count.

The largest single victory belongs to the HX 370 in Cinebench R15 multi-core. It scores 3409.5 against the Max 385's 1579, a delta of 115.9%. That is more than double the older chip's output in this legacy multi-threaded workload. The HX 370 also leads by 38.8% in Cinebench R23 multi-core, scoring 21761 versus 15674. These are substantial margins that point to a decisive advantage in heavily threaded rendering tasks.

Single-core results are closer but still favor the HX 370 in the older R15 test. It scores 306.5 versus 222, a 38.1% lead. However, the situation reverses in Cinebench R23 single-core. There, the Max 385 scores 2212 against 2018, a 8.8% advantage for the newer chip. This suggests the Max 385's higher base clock of 3.60 GHz (versus 2.00 GHz) helps it in certain single-threaded scenarios, even though its boost clock is slightly lower at 5.00 GHz versus 5.10 GHz.

PassMark tests reveal a similar pattern of trade-offs. The HX 370 wins integer math by 14.8% (122933 versus 107046) and floating-point math by 8.1% (76892 versus 71105). It also leads in data compression by 9.6% (445719 versus 406505) and data encryption by 11.7% (22252 versus 19926). Random string sorting goes to the HX 370 by 11.4% (48723 versus 43725). The HX 370's PassMark multi-thread score is 35148 against 33705, a 4.3% edge.

The Max 385 counters in other PassMark workloads. It wins extended instructions by 6.6% (33873 versus 31653) and find prime numbers by 23.6% (165 versus 126). The latter is its largest single victory in absolute percentage terms. The Max 385 also edges out the HX 370 in PassMark single-thread, scoring 4060 versus 3973, a 2.1% lead. The physics test is nearly a tie, with the Max 385 at 1889 and the HX 370 at 1881, a 0.4% difference that is effectively noise.

The overall benchmark averages in the database reflect this mixed picture. The HX 370 has an average benchmark score of 37904, while the Max 385 sits at 44309. That is a 16.9% higher average for the Max 385 across all recorded tests, even though the HX 370 wins more individual head-to-head comparisons. The Max 385 also holds a higher percentile versus all CPUs at 88, compared to the HX 370's 86.

Where Each One Wins

The HX 370 is the clear winner for multi-core productivity workloads. Its 12 cores and 24 threads give it a structural advantage over the Max 385's 8 cores and 16 threads. The Cinebench R23 multi-core result of 21761 versus 15674 confirms this. Any workload that scales with thread count, such as video encoding, 3D rendering, or software compilation, will favor the HX 370 based on these measurements.

The HX 370 also dominates integer-heavy and memory-intensive tasks. PassMark integer math, floating-point math, data compression, encryption, and random string sorting all show the HX 370 ahead by margins ranging from 4.3% to 14.8%. These tests indicate strong general-purpose compute capability that benefits from the higher core count.

The Max 385 wins in specific single-threaded and specialized instruction scenarios. Its Cinebench R23 single-core score of 2212 is 8.8% higher than the HX 370's 2018. PassMark single-thread also favors the Max 385 by 2.1%. The extended instructions test shows a 6.6% advantage for the Max 385, which suggests better handling of SIMD or specialized instruction sets despite fewer cores. The find prime numbers result is striking: the Max 385 scores 165 versus 126, a 23.6% lead. This particular workload often responds to clock speed and cache efficiency, and the Max 385's larger 32 MB L3 cache may play a role there.

For workloads that are latency-sensitive or rely on a single thread, the Max 385 appears better suited. The physics test is essentially a draw, so neither chip claims that territory.

Architecture Differences

The two processors share the same Zen 5 architecture and 4 nm TSMC process node, but they diverge sharply in implementation. The HX 370 uses the Strix Point codename and belongs to the Ryzen AI 300 generation, while the Max 385 uses the Strix Halo codename and belongs to the Ryzen AI Max generation.

Core counts differ significantly. The HX 370 has 12 cores and 24 threads, while the Max 385 has 8 cores and 16 threads. Both use an 80 KB L1 cache per core and a 1 MB L2 cache per core. The L3 cache differs: the HX 370 has 16 MB, while the Max 385 has 32 MB shared. That 16 MB difference likely contributes to the Max 385's wins in cache-sensitive single-threaded tests.

The sockets are different. The HX 370 uses AMD Socket FP8, while the Max 385 uses AMD Socket FP11. Base clocks are 2.00 GHz for the HX 370 and 3.60 GHz for the Max 385. Boost clocks are 5.10 GHz and 5.00 GHz respectively. The Max 385 has a much higher base clock, which explains its strength in single-threaded tests despite a lower boost ceiling.

Memory support is another major divider. The HX 370 supports DDR5 and LPDDR5X over a dual-channel bus with 89.6 GB/s bandwidth. The Max 385 supports only LPDDR5X but over a quad-channel bus with 256.0 GB/s bandwidth. That 2.86x bandwidth advantage for the Max 385 is substantial and may influence memory-bound workloads.

The Max 385 also supports ECC memory, while the HX 370 does not. Both use PCIe Gen 4 with 16 lanes (CPU only). The integrated graphics differ: the HX 370 has Radeon 890M, while the Max 385 has Radeon 8050S. Die sizes are 233 mm² for the HX 370 versus 2x 70.6 mm² for the Max 385, indicating a chiplet design for the latter.

The Max 385 has a higher TDP of 55 W versus 28 W for the HX 370. That higher power envelope likely enables its higher base clock and larger cache configuration.

FAQ

Q: Which processor has a higher average benchmark score in the database?

A: The AMD Ryzen AI Max 385 has an average benchmark score of 44309, compared to 37904 for the AMD Ryzen AI 9 HX 370. The Max 385 also holds a higher percentile versus all CPUs at 88, versus 86 for the HX 370.

Q: Why does the HX 370 win more head-to-head tests despite a lower average score?

A: The HX 370 wins 9 of 15 direct comparisons, mostly in multi-core and integer-heavy workloads. The Max 385 wins 6 tests, but several of its wins, such as find prime numbers and Cinebench R23 single-core, show large percentage advantages. The Max 385's higher average score is driven by the overall database distribution of its recorded benchmarks, which include tests not directly compared in the head-to-head set.

Q: What is the largest single head-to-head margin recorded?

A: The Cinebench R15 multi-core test shows the HX 370 scoring 3409.5 against the Max 385's 1579, a delta of 115.9%. This is the largest percentage difference in the head-to-head data.

Q: Does the Max 385 win any multi-core tests?

A: No. In Cinebench R15 multi-core, Cinebench R23 multi-core, and PassMark multithread, the HX 370 wins by margins of 115.9%, 38.8%, and 4.3% respectively. The Max 385 does not claim victory in any multi-core comparison.

Q: How do the memory systems differ?

A: The HX 370 uses dual-channel memory with 89.6 GB/s bandwidth and supports DDR5 and LPDDR5X. The Max 385 uses quad-channel memory with 256.0 GB/s bandwidth and supports only LPDDR5X. The Max 385 also supports ECC memory, which the HX 370 does not.

Q: Which processor has a higher boost clock?

A: The HX 370 has a boost clock of 5.10 GHz, slightly higher than the Max 385's 5.00 GHz. However, the Max 385 has a much higher base clock of 3.60 GHz versus 2.00 GHz for the HX 370.

Specification Differences

| Specification | AMD Ryzen AI 9 HX 370 | AMD Ryzen AI Max 385 |

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

| Cores | 12 | 8 |

| Threads | 24 | 16 |

| Base Clock | 2.00 GHz | 3.60 GHz |

| Boost Clock | 5.10 GHz | 5.00 GHz |

| TDP | 28 W | 55 W |

| Socket | AMD Socket FP8 | AMD Socket FP11 |

| Codename | Strix Point | Strix Halo |

| Generation | Ryzen AI 300 (Zen 5 / Zen 5c) | Ryzen AI Max (Zen 5 (Strix Halo)) |

| Die Size | 233 mm² | 2x 70.6 mm² |

| L3 Cache | 16 MB | 32 MB (shared) |

| Memory Support | DDR5, LPDDR5X | LPDDR5X |

| Memory Bus | Dual-channel | Quad-channel |

| Memory Bandwidth | 89.6 GB/s | 256.0 GB/s |

| ECC Memory | No | Yes |

| Integrated Graphics | Radeon 890M | Radeon 8050S |

| Release Date | 2024-06-30 | 2025-01-05 |

| Part Number | 100-000000994 | 100-000001424 |

The Verdict

The data indicates two distinct usage profiles. The AMD Ryzen AI 9 HX 370 is the choice for multi-threaded compute. Its 12 cores and 24 threads deliver decisive wins in Cinebench R23 multi-core (21761 versus 15674) and Cinebench R15 multi-core (3409.5 versus 1579). It also leads in PassMark integer math, floating-point math, data compression, encryption, and random string sorting. Users running rendering, encoding, or heavy parallel processing will see the HX 370 outperform the Max 385 by significant margins.

The AMD Ryzen AI Max 385 is the choice for single-threaded responsiveness and specialized workloads. Its 3.60 GHz base clock and 32 MB L3 cache contribute to wins in Cinebench R23 single-core (2212 versus 2018), PassMark single-thread (4060 versus 3973), extended instructions (33873 versus 31653), and find prime numbers (165 versus 126). The Max 385 also offers quad-channel memory with 256.0 GB/s bandwidth and ECC support, which matter for memory-intensive or reliability-focused tasks.

The Max 385's higher average benchmark score of 44309 versus 37904 suggests it performs better across the full recorded benchmark suite, even though the HX 370 wins more direct head-to-head comparisons. The HX 370's wins are larger in aggregate, but the Max 385's overall consistency pushes its average higher. The Max 385 also sits at the 88th percentile versus all CPUs, two points above the HX 370.

For a laptop user who values multi-core throughput above all else, the HX 370's 38.8% lead in Cinebench R23 multi-core is decisive. For a user who prioritizes single-thread performance, memory bandwidth, or ECC support, the Max 385's profile is more compelling. The two chips are not direct substitutes; they serve different segments within the mobile market.

DETAILED SPECIFICATIONS

SPECIFICATION
AI 9 HX 370
AI Max 385
Core Specs
Cores
12
8 -33.3%
Threads
24
16 -33.3%
Base Clock (GHz)
2
3.6 +80.0%
Boost Clock (GHz)
5.1
5 -2.0%
Frequency (GHz)
2
3.6 +80.0%
Turbo Clock (GHz)
5.1
5 -2.0%
Multiplier
20
36 +80.0%
SMP CPUs
1
1 0.0%
Cache
L1 Cache
80 KB (per core)
80 KB (per core)
L2 Cache
1 MB (per core)
1 MB (per core)
L3 Cache
16 MB
32 MB (shared)
Power
TDP (W)
28
55 +96.4%
Configurable TDP
15-54 W
45-120 W
Architecture
Architecture
Zen 5
Zen 5
Codename
Strix Point
Strix Halo
Generation
Ryzen AI 300 (Zen 5 / Zen 5c)
Ryzen AI Max (Zen 5 (Strix Halo))
Process Size
4 nm
4 nm
Die Size
233 mm²
2x 70.6 mm²
Foundry
TSMC
TSMC
Memory
Memory Support
DDR5, LPDDR5X
LPDDR5X
Memory Bus
Dual-channel
Quad-channel
Memory Bandwidth
89.6 GB/s
256.0 GB/s
ECC Memory
No
Yes
Platform
Socket
AMD Socket FP8
AMD Socket FP11
PCIe
Gen 4, 16 Lanes(CPU only)
Gen 4, 16 Lanes(CPU only)
Intel Hybrid
Hybrid Cores
4 + 8
—
E-Core Frequency
1400 MHz up to 3.3 GHz
—
AI/NPU
NPU
Yes / 50 TOPS
Yes / 50 TOPS
Graphics
Integrated Graphics
Radeon 890M
Radeon 8050S
Other
Market
Mobile
Mobile
Production Status
Active
Active
Part Number
100-000000994
100-000001424
Package
FP8
FC-BGA
Tj Max
100°C
100°C
View Ryzen AI 9 HX 370 Details View Ryzen AI Max 385 Details