AMD Ryzen AI Max 390 vs Intel Arc G3 EXTREME Comparison

AMD
AMD

AMD Ryzen AI Max 390

CORE STATE Strix Halo
CORE SPECS 12 Cores / 24 Threads
CLOCK SPEED 3.2 Base / 5 GHz Turbo
CACHE 64 MB (shared)
MAX TDP 55W
ARCHITECTURE Zen 5
nm
PROCESS 4 nm
LAUNCH DATE 2025
VS
Intel
INTEL

Arc G3 EXTREME

CORE STATE Panther Lake
CORE SPECS 14 Cores / 14 Threads
CLOCK SPEED 1.9 Base / 4.7 GHz Turbo
CACHE 18 MB (shared)
MAX TDP 25W
ARCHITECTURE Panther Lake
nm
PROCESS 3 nm
LAUNCH DATE 2026

PERFORMANCE BENCHMARKS

cinebench_cinebench_r15_multicore
3,635
2,192
cinebench_cinebench_r15_singlecore
513
267
cinebench_cinebench_r20_multicore
15,146
10,945
cinebench_cinebench_r20_singlecore
2,138
1,545
cinebench_cinebench_r23_multicore
36,064
14,655
cinebench_cinebench_r23_singlecore
5,091
1,862
passmark_data_compression
487,145
307,094
passmark_data_encryption
25,097
23,881
passmark_extended_instructions
38,716
24,017
passmark_find_prime_numbers
316
248
passmark_floating_point_math
90,594
86,929
passmark_integer_math
146,519
71,164
passmark_multithread
41,737
30,659
passmark_physics
2,761
2,515
passmark_random_string_sorting
53,113
37,331
passmark_single_thread
4,028
4,293
passmark_singlethread
4,028
4,293

Analysis: AMD Ryzen AI Max 390 vs Intel Arc G3 EXTREME

Head-to-Head Benchmarks

The benchmark data records a decisive overall victory for the AMD Ryzen AI Max 390, which wins 15 of the 17 head-to-head comparisons. The average benchmark score for the AMD part is 56273, placing it in the 91st percentile of all CPUs, while the Intel Arc G3 EXTREME averages 36699, placing it in the 85th percentile. The most dramatic separation occurs in Cinebench R23, where the AMD processor scores 36064 in multi-core and 5091 in single-core, versus 14655 and 1862 for the Intel part. These are margins of 146.1% and 173.4% respectively, the two largest deltas in the entire comparison set.

Single-core performance reveals a split between the two suites. In Cinebench R15 single-core, the AMD Ryzen AI Max 390 scores 513 against 267 for the Intel Arc G3 EXTREME, a 92.1% advantage. Cinebench R20 single-core shows the AMD leading 2138 to 1545, a 38.4% gap. However, the PassMark single-thread test tells a different story: the Intel part scores 4293 against 4028 for the AMD chip, giving Intel a 6.2% lead. This is one of only two recorded wins for the Intel processor, the other being the duplicate PassMark single-thread test result.

Multi-core workloads heavily favor the AMD processor. Cinebench R15 multi-core shows 3635 versus 2192, a 65.8% margin. Cinebench R20 multi-core records 15146 against 10945, a 38.4% advantage. PassMark multithread scores 41737 for AMD versus 30659 for Intel, a 36.1% gap. The integer math test produces the second-largest delta of the entire set: 146519 for AMD against 71164 for Intel, a 105.9% difference. Data compression also shows a substantial gap, with AMD scoring 487145 versus 307094, a 58.6% margin.

The closest contests reveal where the Intel part remains competitive. PassMark floating point math shows AMD at 90594 versus Intel at 86929, a modest 4.2% lead. Data encryption is nearly tied, with AMD at 25097 and Intel at 23881, a 5.1% difference. Physics simulation favors AMD by 9.8%, with scores of 2761 and 2515. The remaining wins for AMD are more lopsided: extended instructions show a 61.2% margin (38716 versus 24017), random string sorting a 42.3% margin (53113 versus 37331), and find prime numbers a 27.4% margin (316 versus 248).

Where Each One Wins

The AMD Ryzen AI Max 390 dominates across rendering, compression, encryption, and integer-heavy tasks. Cinebench R23 multi-core, with a 146.1% lead, indicates a strong fit for CPU-based rendering workloads. The 105.9% advantage in integer math suggests superior performance in database operations, financial modeling, and general productivity software. Data compression, with a 58.6% lead, points to advantages in archiving, backup, and file-transfer scenarios. The 61.2% margin in extended instructions indicates better execution of vectorized and specialized instruction sets, which benefits scientific computing and media encoding.

The Intel Arc G3 EXTREME claims only the PassMark single-thread test, with a 6.2% advantage. This suggests a narrow but real strength in lightly threaded, latency-sensitive tasks that depend on high per-core throughput in specific instruction patterns. The Intel part also shows relative competitiveness in floating point math, trailing by only 4.2%, and in data encryption, trailing by 5.1%. These are the workloads where the Intel chip comes closest to matching the AMD processor, though it does not win either test.

The AMD part also wins the Cinebench R15 and R20 single-core tests by 92.1% and 38.4% respectively, which contradicts the PassMark single-thread result. The discrepancy likely stems from different test methodologies, but the recorded data shows that the AMD processor holds a decisive lead in most single-core scenarios despite losing the PassMark single-thread test.

The Verdict

The recorded data supports a clear verdict: the AMD Ryzen AI Max 390 is the stronger processor for the vast majority of workloads. It wins 15 of 17 head-to-head tests, including every multi-core benchmark and all but one single-core benchmark. The Cinebench R23 multi-core margin of 146.1% is the largest single delta, and the integer math margin of 105.9% reinforces the pattern. The AMD part also holds a higher percentile ranking (91st versus 85th) and a higher average benchmark score (56273 versus 36699).

The Intel Arc G3 EXTREME is the choice only when the PassMark single-thread result is the primary consideration, where its 6.2% lead over the AMD chip could matter for specific lightly threaded applications. The Intel part also offers a lower TDP of 25 watts against 55 watts for the AMD chip, which may be relevant for power-constrained mobile designs. However, the performance data shows that the AMD processor delivers substantially higher scores in nearly every category, often by double-digit or triple-digit percentages.

For users prioritizing multi-threaded performance, rendering, compression, or integer processing, the AMD Ryzen AI Max 390 is the clear pick. For users who require the fastest possible PassMark single-thread score and can accept lower multi-core performance, the Intel Arc G3 EXTREME has a narrow but legitimate claim. The overall data, however, favors the AMD processor by a wide margin.

FAQ

Q: Which processor wins the most head-to-head benchmarks?

A: The AMD Ryzen AI Max 390 wins 15 of 17 head-to-head tests, while the Intel Arc G3 EXTREME wins 2 (the PassMark single-thread test and its duplicate).

Q: What is the largest performance gap between the two processors?

A: The largest gap is in Cinebench R23 single-core, where the AMD Ryzen AI Max 390 scores 5091 against 1862 for the Intel Arc G3 EXTREME, a 173.4% difference.

Q: Does the Intel Arc G3 EXTREME win any benchmark?

A: Yes, the Intel part wins PassMark single-thread with a score of 4293 versus 4028 for the AMD processor, a 6.2% margin. This result appears twice in the benchmark list.

Q: How do the processors compare in multi-core performance?

A: The AMD Ryzen AI Max 390 leads in all multi-core tests. Examples include Cinebench R23 multi-core (36064 versus 14655, a 146.1% margin) and PassMark multithread (41737 versus 30659, a 36.1% margin).

Q: What are the average benchmark scores for each processor?

A: The AMD Ryzen AI Max 390 has an average benchmark score of 56273, while the Intel Arc G3 EXTREME averages 36699. The AMD part ranks in the 91st percentile of all CPUs, and the Intel part ranks in the 85th percentile.

Q: Which workloads show the closest results between the two?

A: The closest results are in PassMark floating point math (90594 versus 86929, a 4.2% margin), PassMark data encryption (25097 versus 23881, a 5.1% margin), and PassMark physics (2761 versus 2515, a 9.8% margin).

Architecture Differences

The AMD Ryzen AI Max 390 uses the Zen 5 architecture under the Strix Halo codename, built on a 4 nm process at TSMC. It features 12 cores and 24 threads. The Intel Arc G3 EXTREME uses the Panther Lake codename, built on a 3 nm process at Intel. It features 14 cores and 14 threads, meaning it has no hyperthreading support. The AMD part offers a larger L3 cache at 64 MB shared, while the Intel part has 18 MB shared. Per-core cache also differs: the AMD part has 80 KB L1 and 1 MB L2 per core, while the Intel part has 192 KB L1 and 2.5 MB L2 per core.

Memory architecture diverges significantly. The AMD Ryzen AI Max 390 supports LPDDR5X with a quad-channel memory bus and a memory bandwidth of 256.0 GB/s. The Intel Arc G3 EXTREME also supports LPDDR5X but uses a dual-channel bus with a memory bandwidth of 136.5 GB/s. The AMD part supports ECC memory, while the Intel part does not. PCIe capabilities differ as well: the AMD processor provides Gen 4 with 16 lanes (CPU only), while the Intel processor provides Gen 5 with 4 lanes (CPU only).

The integrated graphics differ by name and architecture family. The AMD Ryzen AI Max 390 includes the Radeon 8050S, while the Intel Arc G3 EXTREME includes the Arc B390. The die size for the AMD part is recorded as 2x 70.6 mm², while no die size is recorded for the Intel part. The production status for both is active, but the release dates differ: the AMD part was released on 2025-01-05, and the Intel part on 2026-05-27.

Specification Differences

The core and thread counts differ: the AMD Ryzen AI Max 390 has 12 cores and 24 threads, while the Intel Arc G3 EXTREME has 14 cores and 14 threads. The base clock of the AMD part is 3.20 GHz, versus 1.90 GHz for the Intel part. The boost clock is 5.00 GHz for AMD and 4.70 GHz for Intel. TDP also differs, with the AMD processor rated at 55 watts and the Intel processor at 25 watts.

The socket types are different: the AMD Ryzen AI Max 390 uses AMD Socket FP11, while the Intel Arc G3 EXTREME uses Intel BGA 2540. The memory bus width differs, with the AMD part using quad-channel and the Intel part using dual-channel. Memory bandwidth is 256.0 GB/s for AMD and 136.5 GB/s for Intel. ECC support is present on the AMD part but absent on the Intel part. PCIe generation and lane count differ: Gen 4 with 16 lanes for AMD, Gen 5 with 4 lanes for Intel.

Process node and foundry differ: the AMD part uses a 4 nm process at TSMC, while the Intel part uses a 3 nm process at Intel. The part numbers are 100-000001423 for AMD and SA4R3 for Intel. Neither processor has an unlocked multiplier, and neither has a recorded launch MSRP. The cache hierarchy differs in size at every level, with the AMD part having a larger L3 cache (64 MB versus 18 MB) but smaller per-core L1 and L2 caches compared to the Intel part.

DETAILED SPECIFICATIONS

SPECIFICATION
AI Max 390
G3 EXTREME
Core Specs
Cores
12
14 +16.7%
Threads
24
14 -41.7%
Base Clock (GHz)
3.2
1.9 -40.6%
Boost Clock (GHz)
5
4.7 -6.0%
Frequency (GHz)
3.2
1.9 -40.6%
Turbo Clock (GHz)
5
4.7 -6.0%
Multiplier
32
19 -40.6%
SMP CPUs
1
1 0.0%
Cache
L1 Cache
80 KB (per core)
192 KB (per core)
L2 Cache
1 MB (per core)
2.5 MB (per core)
L3 Cache
64 MB (shared)
18 MB (shared)
Power
TDP (W)
55
25 -54.5%
Configurable TDP
45-120 W
15-45 W
Architecture
Architecture
Zen 5
—
Codename
Strix Halo
Panther Lake
Generation
Ryzen AI Max (Zen 5 (Strix Halo))
Arc G3 (Panther Lake)
Process Size
4 nm
3 nm
Die Size
2x 70.6 mm²
—
Foundry
TSMC
Intel
Memory
Memory Support
LPDDR5X
LPDDR5X
Memory Bus
Quad-channel
Dual-channel
Memory Bandwidth
256.0 GB/s
136.5 GB/s
ECC Memory
Yes
No
Platform
Socket
AMD Socket FP11
Intel BGA 2540
PCIe
Gen 4, 16 Lanes(CPU only)
Gen 5, 4 Lanes(CPU only)
Intel Hybrid
Hybrid Cores
—
P-Cores: 2 E-Cores: 12
E-Core Frequency
—
1500 MHz up to 3.4 GHz
LP E-Cores
—
4
AI/NPU
NPU
Yes / 50 TOPS
Yes / 46 TOPS
Graphics
Integrated Graphics
Radeon 8050S
Arc B390
Other
Market
Mobile
Mobile
Production Status
Active
Active
Part Number
100-000001423
SA4R3
Package
FC-BGA
FC-BGA
Tj Max
100°C
100°C
View Ryzen AI Max 390 Details View Arc G3 EXTREME Details