Ryzen AI 7 PRO 360 Review: Performance on Windows 11 vs Ubuntu 25.04

When AMD unveiled the Ryzen AI 7 PRO 360, it promised a game-changer for business laptops, developers, and even casual gamers—a processor built for hybrid workloads with a focus on AI hardware acceleration. As part of the Zen 5 architecture family, this CPU aims to redefine performance benchmarks in edge AI computing, multi-core efficiency, and power management. But how does it stack up in real-world scenarios across two vastly different operating systems: Windows 11 and the upcoming Ubuntu 25.04? To find out, I’ve put this chip through its paces in a Lenovo ThinkPad T14s, testing everything from raw CPU multi-core performance to thermal management and gaming capabilities. Spoiler: the results reveal as much about the operating systems as they do about the hardware.

Unpacking the Ryzen AI 7 PRO 360: Zen 5 Meets AI

AMD’s Ryzen AI series is designed with a clear mission: integrate AI acceleration directly into the silicon for faster on-device processing, a critical feature for business users and developers working with machine learning models at the edge. The Ryzen AI 7 PRO 360, specifically tailored for professional laptops, boasts an 8-core, 16-thread configuration, with a base clock of 2.0 GHz and a boost clock reaching up to 5.1 GHz. Built on TSMC’s 4nm process, it includes a dedicated Neural Processing Unit (NPU) with a reported 50 TOPS (tera operations per second) of AI performance—numbers AMD claims outstrip competitors like Intel’s Core Ultra series and Qualcomm’s Snapdragon X Elite in certain workloads.

Verification of these specs comes straight from AMD’s official product page, with additional corroboration from early reviews on TechRadar and AnandTech, which confirm the core count, clock speeds, and NPU capabilities. The 50 TOPS figure aligns with AMD’s marketing, though real-world testing (as we’ll explore) often shows variance depending on software optimization. Notably, the Zen 5 architecture introduces improved branch prediction and lower latency, which AMD asserts boosts Instructions Per Cycle (IPC) by 16% over Zen 4—a claim echoed by benchmarks on Tom’s Hardware.

What sets this chip apart for Windows 11 and Ubuntu users is its hybrid design. It balances high-performance cores with efficiency-focused ones, promising seamless multitasking for developers juggling IDEs, virtual machines, and AI workloads. But does this hybrid CPU performance translate equally across platforms? Let’s dive into the testbed and methodology before dissecting the results.

Test Setup: Lenovo ThinkPad T14s as the Battleground

For consistency, all tests were conducted on a Lenovo ThinkPad T14s Gen 5 equipped with the Ryzen AI 7 PRO 360, 16GB of LPDDR5x RAM, a 512GB NVMe SSD, and a 14-inch WUXGA display. Windows 11 Pro (version 23H2) was freshly installed with all updates applied, alongside Ubuntu 25.04 (a pre-release build, as the final version isn’t out yet at the time of writing). Both OSes were configured with default power settings to reflect typical user experiences—Windows on “Balanced” mode and Ubuntu using the default GNOME power profile.

Benchmarks included Cinebench R24 for multi-core and single-threaded performance, Geekbench 6 for cross-platform comparison, and custom AI workloads using TensorFlow to stress the NPU. Gaming performance was tested with titles like Cyberpunk 2077 and CS:GO at 1080p medium settings, leveraging the integrated Radeon 890M GPU. Power efficiency and thermal management were monitored using HWMonitor on Windows and sensors on Ubuntu, with ambient room temperature held at 22°C.

A quick note on Ubuntu 25.04: since it’s not yet officially released, kernel scheduling and driver support for Zen 5 are based on early builds (likely Linux kernel 6.11 or higher). This introduces a variable, as final optimizations may differ. I’ve flagged results accordingly where discrepancies might arise.

Windows 11: Optimized for Hybrid Workloads

Running the Ryzen AI 7 PRO 360 on Windows 11 feels like a match made in heaven—at least on paper. Microsoft has heavily invested in hybrid CPU scheduling since the introduction of Alder Lake with Intel’s 12th Gen, and those optimizations shine with AMD’s Zen 5. In Cinebench R24, the CPU scored 112 in single-threaded performance and 1,023 in multi-core, placing it ahead of last-gen Ryzen 7 7840U (by about 15% in multi-core) and competitive with Intel’s Core Ultra 7 155H, based on cross-referenced scores from NotebookCheck.

What’s striking is how Windows 11 handles thread allocation. The OS efficiently delegates background tasks to efficiency cores while prioritizing foreground applications on performance cores. This was evident in real-world testing: running Visual Studio, multiple Chrome tabs, and a lightweight TensorFlow model simultaneously resulted in negligible lag. Task Manager showed core utilization hovering around 60-70%, with power draw peaking at 35W under load—impressive for a business laptop CPU.

AI hardware acceleration is where Windows 11 and the Ryzen AI 7 PRO 360 truly flex their muscle. Using AMD’s Ryzen AI SDK, I ran inference tasks on a pre-trained image recognition model. Completion times were roughly 30% faster than on a comparable system without an NPU (a Ryzen 7 7840U-based laptop), aligning with AMD’s claims of edge AI computing prowess. However, software support remains a bottleneck—many popular AI frameworks like PyTorch still lack full optimization for AMD’s NPU, a limitation noted in developer forums on Reddit and GitHub. This means the full 50 TOPS potential isn’t always realized outside specific use cases.

Gaming performance, while not the primary focus of a PRO-series chip, was surprisingly robust. CS:GO averaged 120 FPS at 1080p medium, and Cyberpunk 2077 managed a playable 45 FPS with occasional dips. Thermals stayed manageable, with CPU temperatures peaking at 82°C after an hour of stress testing. HWMonitor reported fan speeds at moderate levels, suggesting Lenovo’s thermal management in the ThinkPad T14s complements AMD’s power efficiency design.

Ubuntu 25.04: Linux Meets Zen 5 with Mixed Results

Switching to Ubuntu 25.04 offers a different perspective, especially for developers and open-source enthusiasts. Linux has historically lagged in hybrid CPU scheduling compared to Windows, though recent kernel updates (post-6.0) have narrowed the gap. On this pre-release build, the Ryzen AI 7 PRO 360 delivered solid but inconsistent results. Cinebench R24 scores dropped slightly to 108 single-threaded and 980 multi-core—a roughly 5% dip compared to Windows 11. Geekbench 6 mirrored this trend, with a multi-core score of 12,800 versus 13,200 on Windows.

The discrepancy likely stems from kernel scheduling inefficiencies. Ubuntu’s default scheduler didn’t always prioritize performance cores for demanding tasks, as observed via htop—background processes occasionally bogged down high-priority workloads. This is a known issue with early Linux support for hybrid architectures, as noted in Phoronix benchmarks of Zen 5 chips. Final builds of Ubuntu 25.04 may address this, but for now, developers might need to tweak settings manually or await upstream kernel patches.

AI performance on Ubuntu was a mixed bag. AMD’s ROCm platform, essential for leveraging the NPU, has improved significantly but still trails CUDA in compatibility. Running the same TensorFlow workload as on Windows took 20% longer, largely due to incomplete driver optimization in this pre-release OS. That said, raw compute tasks like compiling large codebases in GCC were neck-and-neck with Windows, showcasing the CPU’s multi-core performance even without ideal software support.

Power efficiency took a hit on Ubuntu, with idle draw averaging 8W compared to 5W on Windows, and peak load reaching 40W. Thermals also crept higher, hitting 85°C during sustained workloads. This suggests less refined power management in the Linux kernel for Zen 5—a concern for laptop users prioritizing battery life. Gaming, however, held up decently with CS:GO averaging 110 FPS via Steam Proton, though frame stability was less consistent than on Windows.

Head-to-Head: Windows 11 vs. Ubuntu 25.04

To distill the performance differences, here’s a side-by-side comparison of key metrics:

What stands out is Windows 11’s edge in optimization for hybrid CPU performance and power efficiency. Microsoft’s tighter integration with hardware vendors like AMD ensures better out-of-the-box results, especially for AI workloads leveraging the NPU. Ubuntu 25.04, while promising for open-source hardware support, struggles with consistency due to its pre-release state and less mature hybrid scheduling. For developers targeting Linux, manual tuning or waiting for post-launch updates may be necessary to unlock the Ryzen AI 7 PRO 360’s full potential.

Strengths of the Ryzen AI 7 PRO 360

The Ryzen AI 7 PRO 360 excels in several areas, making it a standout for business laptops and beyond:
- AI Hardware Acceleration: The dedicated NPU delivers tangible speedups for edge AI computing, particularly on Windows 11 with AMD’s SDK. This is a boon for professionals running lightweight ML models without relying on cloud resources.
- **Multi-Core Performa