Feex·CPU
AMD · Ryzen AI · Laptop

AMD® Ryzen™ AI Z2 Extreme

The AMD Ryzen AI Z2 Extreme is a laptop processor released in 2025 using a 4 nm process. It has 8 cores and 16 threads, a base clock of 2.00 GHz and 16 MB of L3 cache.

Cores
8
Threads
16
Base Clock
2.0 GHz
Process Node
4 nm
L3 Cache
16 MB
Multi-Core Performance
22,245
Single-Core Performance
3,581
Core layout
PPPPPPPP
8 Performance 16 Threads

Best for

High-FPS gaming & low-latency tasksDaily productivity & development

Overview

The AMD Ryzen AI Z2 Extreme is a laptop processor with 8 cores and 16 threads (4 nm process). It launched in 2025. A 16-MB L3 cache backs the cores (3.1 MB per core).

Performance

Overall CPU Performance computed
8,925
36% of reference
Reference ceiling 25,000
Multi-Core Performance PassMark rating
22,245
19% of reference
Reference ceiling 120,000
Single-Core Performance PassMark rating
3,581
65% of reference
Reference ceiling 5,500
Productivity Performance computed
19,473
12% of reference
Reference ceiling 160,000
Creator Performance computed
18,681
13% of reference
Reference ceiling 140,000
Engineering Performance computed
22,801
11% of reference
Reference ceiling 200,000
Gaming Performance computed
16,304
74% of reference
Reference ceiling 22,000
Performance per Core computed
2,781
13% of reference
Reference ceiling 21,053
Performance per Thread computed
1,390
13% of reference
Reference ceiling 10,527
Performance index computed
111
Intel Core i5-12400 = 100

The AMD Ryzen AI Z2 Extreme scores 3,581 in single-core performance and 22,245 in multi-core performance (PassMark CPU Mark ratings). From those two ratings Feex computes 8,925 for overall performance, 19,473 for productivity performance, 18,681 for creator performance, 22,801 for engineering performance and 16,304 for gaming performance. Its multi-core throughput works out to 2,781 per core and 1,390 per thread. Its performance index is 111, so its overall performance is 11% above the Intel Core i5-12400 (index 100). Each bar shows the score as a share of a reference ceiling, which is the larger of the 99th-percentile score across the Feex CPU database and a fixed baseline.

How these scores are computed

Single-core and multi-core performance are PassMark CPU Mark single-thread and multi-thread ratings. Every other figure on this page's performance section is computed by Feex from those two ratings plus core count, thread count and TDP, so it is a derived value and not a separate benchmark run.

  • Overall performance is the geometric mean of the two ratings: the square root of single-core times multi-core.
  • Productivity performance is 0.65 times multi-core plus 1.4 times single-core.
  • Creator performance is 0.55 times multi-core plus 1.8 times single-core.
  • Engineering performance is multi-core times the smaller of 1.4 and (0.9 plus cores divided by 64).
  • Gaming performance is 3 times single-core plus 0.25 times multi-core.
  • Performance per core, per thread and per watt divide the multi-core rating by physical cores, logical threads and TDP in watts.
  • The efficiency index is the square root of single-core times multi-core, divided by the square root of TDP.
  • The performance index is overall performance divided by the overall performance of the Intel Core i5-12400 (8,069), times 100.

Workload Benchmarks

Scores for individual PassMark CPU Mark test categories. Higher is better. These are measured test scores, not computed from the single-core and multi-core ratings.

Integer Math Throughput
77,210
12% of reference
Reference ceiling 617,811
31% above the Intel Core i5-12400 (58,893)
Raw integer arithmetic throughput across cores.
Floating-Point Math
48,172
8% of reference
Reference ceiling 600,000
5% above the Intel Core i5-12400 (45,899)
Heavy floating-point math, mixed precision.
Compression Performance
266,882
12% of reference
Reference ceiling 2,309,868
17% above the Intel Core i5-12400 (228,054)
Throughput on lossless compression workloads.
Cryptography Performance
13,509
10% of reference
Reference ceiling 137,106
18% above the Intel Core i5-12400 (11,489)
Symmetric encryption and hashing throughput.
Physics Simulation
1,301
11% of reference
Reference ceiling 12,291
19% above the Intel Core i5-12400 (1,090)
Iterative physics & rigid-body simulation.
Vector / SIMD Throughput
18,885
11% of reference
Reference ceiling 172,975
22% above the Intel Core i5-12400 (15,509)
Wide-vector SIMD intrinsics performance.
Prime Sieve
83
8% of reference
Reference ceiling 1,067
24% above the Intel Core i5-12400 (67)
Searching for prime numbers (branch-heavy work).
Sort Performance
28,897
11% of reference
Reference ceiling 260,323
28% above the Intel Core i5-12400 (22,652)
Random string sorting throughput.

In the individual workload tests the AMD Ryzen AI Z2 Extreme scores 77,210 in integer math, 48,172 in floating-point math, 266,882 in compression, 13,509 in cryptography, 1,301 in physics simulation, 18,885 in vector SIMD, 83 in the prime sieve and 28,897 in sorting. Higher is better. Compared with the Intel Core i5-12400, it is 31% above in integer math, 5% above in floating-point math, 17% above in compression, 18% above in cryptography, 19% above in physics simulation, 22% above in vector SIMD, 24% above in the prime sieve and 28% above in sorting. Each bar shows the score as a share of that workload's reference ceiling, the larger of the 99th-percentile score across CPUs with a recorded value and a fixed baseline.

Core Configuration

Cores
8
Threads
16
Threads per Core
2
Base Clock
2.0 GHz
Boost Clock
Not recorded

The AMD Ryzen AI Z2 Extreme has 8 physical cores and runs 16 logical threads. It runs 2 threads per core. It runs at a base clock of 2.00 GHz.

Architecture & Generation

Manufacturer
AMD
Family
Ryzen AI
Segment
Laptop
Generation
Not recorded
Architecture
Not recorded
Process Node
4 nm
Lithography
4 nm
Release Date
2025
Release quarter
Q4 2025

The AMD Ryzen AI Z2 Extreme is a laptop processor, belongs to the AMD Ryzen AI family and is manufactured on a 4 nm process. It launched in 2025.

Cache

L1 Cache (Total)
640 KB
L2 Cache (Total)
8 MB
L2 Cache (per Core)
1 MB
L3 Cache
16 MB
Cache per Core
3.1 MB
Total Cache
24.6 MB

The AMD Ryzen AI Z2 Extreme has 640 KB of L1 cache, 8 MB of L2 cache and 16 MB of L3 cache. That adds up to about 24.6 MB of cache, or 3.08 MB per core. Each core has 1 MB of private L2 cache.

Cache Hierarchy

L1L2L3
L1 640 KBL2 8 MBL3 16 MB

The AMD Ryzen AI Z2 Extreme has a 3-level cache hierarchy: L1 at 640 KB, L2 at 8 MB and L3 at 16 MB. L3 is the largest level and makes up about 65% of the cache listed here.

Platform Capabilities

Instruction sets, security, virtualization, AI acceleration and OS readiness. Features are listed as supported, explicitly not supported, or not recorded; a feature that is not recorded is not the same as unsupported.

Instruction Sets

0 supported
Not recorded
AES-NIAVX-512AVX2AVXAVX-VNNICET

Security and Reliability

0 supported
Not recorded
AMD SEVAMD SMEControl-flow Enforcement (CET)Execute Disable (NX)Memory EncryptionSMAP

Virtualization

0 supported
Not recorded
AMD SVMAMD-V (Hardware Virtualization)NPT / RVIIOMMUSecond-Level Address Translation

OS Compatibility

1 supported
Supported
Windows 11 eligible

Feex has no record for AES-NI, AVX-512, AVX2, AVX, AVX-VNNI and CET, which is different from being unsupported.

Feex has no record for AMD SEV, AMD SME, Control-flow Enforcement (CET), Execute Disable (NX), Memory Encryption and SMAP, which is different from being unsupported.

Feex has no record for AMD SVM, AMD-V (Hardware Virtualization), NPT / RVI, IOMMU and Second-Level Address Translation, which is different from being unsupported.

Feex records the AMD Ryzen AI Z2 Extreme as Windows 11 eligible.

Physical & Manufacturing

Foundry
TSMC

The AMD Ryzen AI Z2 Extreme is manufactured by TSMC.

Naming & Aliases

Display Name
AMD® Ryzen™ AI Z2 Extreme
Normalized Name
AMD® Ryzen™ AI Z2 Extreme
Product line brand
Ryzen AI

Real-world compute in the Feex fleet

No Feex fleet machines with the AMD Ryzen AI Z2 Extreme have reported compute telemetry yet, so real-world compute for this processor is not recorded.

How fleet compute is measured · Fleet compute leaderboard · JSON

Data Coverage

Feex has recorded 35% of the specifications it tracks for the AMD Ryzen AI Z2 Extreme. Values Feex does not hold are shown as "Not recorded" rather than estimated.

Not recorded: Core Configuration: Performance Cores, Efficiency Cores, Hybrid Architecture, All-Core Boost and Unlocked Multiplier. Architecture & Generation: Codename, ISA Family and Market Status. Cache: L4 Cache. Memory Support: Max Memory, Memory Channels, Memory Bandwidth, Memory Types, DDR5 Support, DDR4 Support, LPDDR Support, ECC Memory and Max Memory Speed. Graphics: Integrated Graphics, Graphics Model, Graphics Base Clock, Graphics Max Clock, Execution Units, Display Outputs and Max Video Memory. Platform & Compatibility: Socket, Package, Package Size, PCIe Generation and PCIe Lanes. Power & Thermals: TDP, Base Power, Maximum Turbo Power, Configurable TDP (Min), Configurable TDP (Max), Max Junction Temperature and Tcase. Physical & Manufacturing: Transistor Count, Die Size, Die Count and Launch Price. Naming & Aliases: SKU, Model Number, Processor Number and Part Number.

Common questions

How many cores and threads does the AMD Ryzen AI Z2 Extreme have?

The AMD Ryzen AI Z2 Extreme has 8 cores and 16 threads.

How does the AMD Ryzen AI Z2 Extreme perform?

In the Feex normalized performance dataset, the AMD Ryzen AI Z2 Extreme scores 22,245 multi-core and 3,581 single-core.

What is the AMD Ryzen AI Z2 Extreme best for?

Based on its specifications, the AMD Ryzen AI Z2 Extreme is well suited for High-FPS gaming & low-latency tasks, Daily productivity & development.

How much cache does the AMD Ryzen AI Z2 Extreme have?

The AMD Ryzen AI Z2 Extreme has 640 KB of L1 cache, 8192 KB of L2 cache and 16 MB of L3 cache.

How does the AMD Ryzen AI Z2 Extreme compare with the Intel Core i5-12400?

The AMD Ryzen AI Z2 Extreme has a computed performance index of 111, which is 11% above the Intel Core i5-12400 (index 100). The index is derived from the single-core and multi-core ratings.