iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
AMD’s Strix Halo is now sold as Ryzen AI Max, with the Ryzen AI Max+ 395 at the top of the range. Its radical feature is a unified memory architecture that lets the Zen 5 CPU, RDNA 3.5 GPU, and NPU share up to 128GB of LPDDR5x memory—giving compact systems far more graphics-accessible capacity than typical integrated graphics and many mobile discrete GPUs.
Strix Halo is AMD’s former codename for the Ryzen AI Max 300 Series, led by the Ryzen AI Max+ 395. Its defining feature is not simply a large Zen 5 CPU or an unusually capable integrated Radeon GPU. It is the combination of 16 Zen 5 cores, a 40-compute-unit RDNA 3.5 graphics engine, an AI NPU, and up to 128GB of fast unified LPDDR5x memory that all three processors can access.
That design gives compact laptops and workstations a capability usually associated with a CPU paired with a discrete GPU: a large memory pool for graphics, rendering, and local AI models without a separate VRAM subsystem. The trade-off is equally important: this is still integrated graphics using shared system memory, the RAM is soldered, and actual performance depends heavily on the device’s power limit, cooling, firmware, drivers, and software.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11What happened to Strix Halo?
AMD introduced Strix Halo during CES 2025 on January 6, 2025. Strix Halo was the codename used in launch coverage; AMD’s current retail identity is the Ryzen AI Max 300 Series. The flagship is the Ryzen AI Max+ 395, while professional systems use related Ryzen AI Max PRO branding.
The processor is designed for laptops, compact desktops, and workstations. It uses AMD’s FP11 platform and soldered LPDDR5x memory, so it is not a conventional desktop processor that a buyer installs into a replaceable motherboard socket. A system manufacturer determines the memory capacity, cooling solution, power envelope, display, battery, and other parts of the final product.
Ryzen AI Max+ 395 specifications at a glance
| Part | Ryzen AI Max+ 395 specification |
|---|---|
| CPU architecture | Zen 5 |
| CPU cores and threads | 16 cores, 32 threads |
| Maximum boost clock | Up to 5.1GHz |
| L3 cache | 64MB |
| Integrated graphics | Radeon 8060S, RDNA 3.5 |
| Graphics compute units | 40 CUs |
| Maximum graphics frequency | Up to 2.9GHz |
| NPU performance | Up to 50 NPU TOPS |
| Total advertised AI performance | Up to 126 overall TOPS |
| Memory | Up to 128GB LPDDR5x-8000 |
| Memory interface | 256-bit unified interface |
| Theoretical memory bandwidth | Approximately 256GB/s |
| Default TDP | 55W |
| Configurable TDP range | 45W to 120W |
| Package | Three-die package |
TOPS and theoretical bandwidth are specifications, not direct measures of application speed. A laptop running the processor at a sustained 55W will not necessarily perform like a thick workstation configured near 120W, and a supported AI model can behave very differently from a game or a CPU renderer.
The important innovation: one large memory pool
Most gaming and workstation laptops divide memory into two separate resources. The CPU uses system RAM, while a discrete GPU has its own high-speed VRAM. Moving data between those pools costs time and often becomes a capacity problem: a laptop may have plenty of system memory but only 8GB or 16GB of graphics memory.
Ryzen AI Max takes a different approach. AMD specifies a 256-bit LPDDR5x interface, support for LPDDR5x-8000, and configurations up to 128GB. At the maximum memory speed, the theoretical bandwidth calculation is:
8,000MT/s × 256 bits ÷ 8 = approximately 256GB/s
The CPU, GPU, and NPU share a coherent address space. That means a workload does not always need to make an explicit CPU-to-GPU copy before the graphics engine can work with data already in memory. The shared design can reduce duplication and, more importantly, gives the integrated GPU access to much more capacity than a typical integrated graphics implementation.
What AMD’s 96GB graphics-memory figure really means
A 128GB system can reserve up to 96GB through AMD’s Variable Graphics Memory allocation. This is the reason some launch material described Strix Halo as offering up to 96GB of graphics memory.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Fix the driver behind crashes, sound loss and screen glitchesFind Drivers →It is not the same as a discrete GPU with 96GB of dedicated GDDR or HBM. The reserved graphics-oriented pool is still part of the system’s unified LPDDR5x memory. The GPU can also access the broader pool, but memory access, contention with the CPU and NPU, latency, and sustained bandwidth are not identical to those of a discrete graphics card with physically attached VRAM.
The practical benefit is capacity and convenience. A large model or graphics dataset that would not fit inside an 8GB or 16GB mobile GPU’s VRAM may fit in a suitably configured Ryzen AI Max system. The practical limitation is that fitting is not the same as running quickly. Once a workload competes for bandwidth with the CPU, operating system, or another accelerator, performance can change substantially.
Rank #2
- 【Incredible performance】: Equipped with an AMD Ryzen 5 processor and 512GB SSD, this laptop is designed to provide an ultrafast and smooth experience
- 【Fast charging battery】: ASUS fast-charge technology can recharge the battery up to 50% capacity in just 30 minutes, allowing you to quickly top it up without interrupting your workflow
- 【Extra toughness and durability】: This laptop stays cool in all situations thanks to ASUS IceCool thermal technology, and meets US military-grade standards for longevity and sustainability
- 【Effortless typing experience】: The precisely measured and fine-tuned ErgoSense keyboard design reduces strain on your hands and wrists
- 【Smooth video call experience】: AI Noise-Canceling Technology isolates unwanted noise for smooth communications
Four-channel memory and MALL cache
AMD’s workstation material describes the design as a four-channel memory architecture with up to 32MB of MALL cache. The cache and wide memory interface are intended to keep the CPU, GPU, and NPU supplied with data while avoiding the capacity ceiling typical of mobile discrete GPUs.
This is why memory capacity is central to Strix Halo’s identity. The processor is not merely an ordinary laptop APU with a larger specification sheet; its design is aimed at workloads that can use a large shared working set. However, the memory is soldered LPDDR5x. Buyers cannot plan to purchase a 32GB model and upgrade it later to 128GB.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →How the CPU, GPU, and NPU fit together
Zen 5 CPU
The Ryzen AI Max+ 395 has 16 Zen 5 CPU cores and 32 threads, with a maximum boost clock of 5.1GHz and 64MB of L3 cache. That is a substantial CPU configuration for a thin or compact system, particularly for compiling code, CPU rendering, virtual machines, photo and video work, and other heavily threaded tasks.
The 55W default TDP should not be mistaken for a fixed operating power. AMD allows system designers to configure the processor from 45W to 120W. A thin tablet and a larger workstation may therefore use the same processor name while delivering different sustained performance.
Radeon 8060S integrated graphics
The flagship’s graphics engine is called Radeon 8060S. It contains 40 RDNA 3.5 compute units and has a listed graphics frequency of up to 2.9GHz.
Launch coverage sometimes called this a 40-core GPU. The more accurate term is 40 compute units. Compute units are GPU building blocks and are not equivalent to CPU-style cores. The Radeon 8060S is integrated into the processor package and uses the unified LPDDR5x memory rather than a separate bank of dedicated VRAM.
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errorsThe large integrated GPU is useful for gaming, GPU rendering, image generation, video work, and compute workloads in a compact chassis. But whether it can maintain its rated speed depends on cooling and the system’s power configuration. The same 40-CU GPU can perform differently in a well-cooled workstation and a thin 2-in-1.
AI NPU
The Ryzen AI Max+ 395 includes an NPU rated at up to 50 TOPS, while AMD lists up to 126 overall TOPS for the processor. The NPU is intended for efficient AI inference and compatible operating-system or application features. It does not automatically accelerate every local-AI workload; many large language-model and image-generation applications may use the GPU, CPU, or a combination depending on their software support.
Why AMD uses three dies
AMD lists the flagship as a three-die package. Launch descriptions explain that arrangement as a large central I/O die containing the GPU and NPU alongside two smaller CPU dies. This chiplet-style construction helps AMD combine a large integrated graphics engine with as many as 16 Zen 5 CPU cores in a compact processor package.
Rank #3
- CPU: AMD Ryzen 7 7730U 8-Core (Base Frequency 2.0 GHz, Up to 4.5 GHz with AMD Turbo Core Technology, 16 MB AMD L3 cache, 8 cores (P + E), 16 threads)
- Screen and Graphics: Brilliant 15.6" Touchscreen LED-backlit FHD (1920 x 1080) IPS 300 nits Anti-glare, 45% NTSC, 60Hz, Thin Bezel LCD Display powered by AMD Radeon Graphics
- RAM & Storage: 16GB DDR4-3200 MHz ; 512GB PCI-E NVMe Solid State Drive
- Ports and Slots:1 x USB 3.2 Gen 1 Type-C support Data Transfer / 5Gbps, 2 x USB 3.2 Gen 1 Type-A support 5Gbps, 1 x HDMI v1.4, 1 x Headphone/Microphone Combo Jack
The Ryzen AI Max family is not one identical chip
It is a mistake to assume that every Ryzen AI Max system has the flagship’s 40 compute units. The launch family includes different CPU and GPU configurations:
Free tools Windows power users keep installed
One-click scans. No signup required.
| Processor | CPU configuration | GPU configuration | What to watch for |
|---|---|---|---|
| Ryzen AI Max+ 395 | 16 Zen 5 cores | 40 CUs | Flagship configuration; Radeon 8060S |
| Ryzen AI Max 390 | 12 Zen 5 cores | 32 CUs | Fewer CPU cores but still a large integrated GPU |
| Ryzen AI Max 385 | 8 Zen 5 cores | 32 CUs | GPU count does not fall with the CPU count in the same way |
| Ryzen AI Max 380 | 6 Zen 5 cores | 16 CUs | Lower graphics configuration |
| Ryzen AI Max PRO variants | Varies by model | Varies by model | Generally found in business and workstation systems |
| OEM-only configuration reported at launch | 16 Zen 5 cores | 16 CUs | Check the exact OEM specification rather than relying on the CPU name |
Memory capacity also varies by system. The processor family can appear in 32GB, 64GB, and 128GB configurations, and those differences may matter more to a local-AI buyer than a modest CPU-clock difference.
What the launch performance claims prove—and what they do not
AMD’s CES launch material made ambitious comparisons. AMD claimed up to 1.4 times the gaming performance of an Intel Core Ultra 9 288V system, up to 84% faster rendering than an Apple MacBook Pro with an M4 Pro, and up to 2.2 times the performance of an RTX 4090 in a specified AI workload at a lower TDP.
Those statements were AMD claims from specified test configurations, not independent test results and not universal rankings. They should be read with the processor power limit, cooling configuration, operating system, drivers, application version, and benchmark workload in mind.
At launch, AMD supplied synthetic 3DMark gaming results rather than a broad set of real-world game tests. A reference board can also provide substantially better cooling than a thin commercial laptop. That makes it unsafe to convert a launch slide into a promise that every Ryzen AI Max+ 395 laptop will beat every RTX 4070 or RTX 4090 system.
CPU results also require care. Reported Cinebench 2024 results put the chip only marginally ahead of a 12-core Apple M4 Pro and behind a 14-core M4 Pro in the cited comparison. That does not make the Ryzen AI Max+ 395 a weak CPU; it demonstrates why one benchmark or one vendor comparison cannot establish an across-the-board performance crown.
A better interpretation is that Strix Halo offers an unusually high performance ceiling for a unified-memory compact system. Whether it reaches that ceiling depends on the product around the processor.
Why Strix Halo matters for local AI
The strongest case for the architecture is memory capacity. A local language or image model may fail to load when it exceeds a GPU’s available VRAM, even if the computer has ample system RAM. A Ryzen AI Max system with 64GB or 128GB of unified memory can provide a much larger graphics-accessible working space.
AMD’s own workstation material uses Stable Diffusion 3.5 Large and Llama 3.1 70B-Q4 as examples of workloads that can benefit from the larger graphics-memory capacity. These are capability examples, not guarantees. Whether a model fits depends on the exact quantization, context length, runtime, weights, cache, operating system, and other memory allocations. A model loading successfully also says nothing by itself about tokens per second or image-generation time.
Rank #4
- 【RAM & Storage】This computer comes with 16GB RAM | 512GB SSD
- 【 AMD Ryzen 7 7730U Processor】It is a processor for thin and light laptops based on the Cezanne generation. It is part of the second "Barcelo" refresh in early 2023 and offers the same specifications as the old Ryzen 7 5825U. The CPU integrates all eight cores based on the Zen 3 microarchitecture clocked at 2 (guaranteed base clock) to 4.5 GHz (Turbo) and support SMT / Hyperthreading (16 threads). The chip is manufactured in 7nm at TSMC.
- 【15.6 " (1920x1080) Full HD Display】With its narrow bezel design the 15.6" 1080p Full HD display is perfect for casual web browsing and watching movies or streaming, allowing for a sharp, detailed view of what's in front of you. And with Acer BluelightShield, lower the levels of blue light to lessen the negative effects of blue light exposure.
- 【Other features】AMD Radeon Graphics,Copilot in Windows - your AI Assistant,Windows 11 Home OS.
- 【 Bundle with Portable Laptop Stand】Bundled with a pair portable laptop stand.Propping your laptop up on a stand keeps it elevated from the surface of your workstation, protecting it from any accidental spills. As you'll be using an external mouse and keyboard you'll also limit the amount of dirt being transferred onto the laptop, keeping it in good working order for longer
AMD’s March 2025 testing used an ASUS ROG Flow Z13 with 64GB of unified memory and reported up to 2.2 times the token throughput of Intel Arc 140V in selected LM Studio tests. AMD also reported large time-to-first-token advantages on selected models. Those results came from AMD’s chosen system, software stack, models, and test conditions, so they should be treated as vendor testing rather than independent evidence.
For AI developers, software support may matter as much as hardware. AMD’s Ryzen AI Halo developer platform is positioned around Windows or Linux, ROCm support, and a 128GB unified-memory configuration. AMD advertises local operation of models up to 200 billion parameters and up to 60 FP16 TFLOPS of graphics performance. Those figures describe the platform’s advertised capability and should not be treated as a speed guarantee for every model or runtime.
AMD also identifies a future configuration using a Ryzen AI Max+ PRO 495 processor with support for 192GB of memory as coming soon. That should not be treated as generally available until AMD or a system vendor confirms final availability, pricing, and specifications.
Where you can actually find the platform
ASUS ROG Flow Z13 (2025)
The most recognizable consumer implementation is the ASUS ROG Flow Z13 2025 Ryzen AI Max+ 395. It is a 13-inch 2-in-1 gaming tablet rather than a conventional clamshell laptop. ASUS lists the Ryzen AI Max+ 395, Radeon 8060S graphics, unified quad-channel memory, a 2.5K 180Hz display, and a 70Wh battery among the platform’s headline features. ASUS announced U.S. preorders in February 2025.
Recommended Free Tools
For a consumer who wants to try the Strix Halo concept in a portable gaming and productivity device, the Flow Z13 is the clearest starting point. Before buying, verify the exact CPU and RAM configuration: the product family includes materially different Ryzen AI Max processors and memory capacities. A listing that says ROG Flow Z13 does not, by itself, guarantee the 395, 40-CU graphics, or 128GB of memory.
HP ZBook Ultra G1a
The HP ZBook Ultra G1a Ryzen AI Max PRO workstation is a better fit for professional creators, developers, and workstation users than for buyers seeking a gaming tablet. HP lists Ryzen AI Max PRO configurations and up to 128GB of soldered LPDDR5X memory.
The ZBook’s value is its workstation positioning, business support, and professional configuration options. It is not a substitute for checking application certification, driver support, sustained power behavior, and the exact memory configuration required by your workflow.
AMD Ryzen AI Halo Developer Platform
The AMD Ryzen AI Halo developer platform is a specialized developer system, not a conventional laptop alternative. Its advertised 128GB unified-memory configuration, Windows and Linux support, ROCm integration, and compact developer-oriented setup make it more relevant to local-AI researchers and software developers than to ordinary office buyers.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Availability and pricing should be checked directly before treating it as a purchase recommendation. Its purpose is to provide a prepared development environment and a large memory pool, not necessarily the battery life, display, keyboard, or portability expected from a notebook.
GPD and One-Netbook systems
GPD has promoted a Ryzen AI Max+ 395 handheld, while One-Netbook has publicized a Ryzen AI Max+ 395 gaming 2-in-1. These products are interesting enthusiast options, but availability, final memory configuration, warranty coverage, regional support, and retailer reliability can vary. Verify those details before importing or preordering one.
Which type of buyer benefits most?
- Local-AI developers: Choose a 64GB or 128GB configuration when model capacity is the priority, and confirm that your intended runtime supports the GPU or ROCm path you plan to use.
- Mobile workstation users: The high CPU-core count and large shared memory pool can suit compilation, rendering, media work, simulation, and multitasking, but check the application’s GPU and driver support.
- Portable gamers: A Radeon 8060S system can offer a powerful integrated-graphics experience in a compact design, but compare the device’s sustained power limit and cooling rather than comparing processor names alone.
- Advanced general users: The platform can reduce the need to carry a separate GPU-equipped machine when workloads alternate between CPU, GPU, and AI tasks.
Who should be cautious?
- Anyone who expects upgradeable RAM: LPDDR5x memory is soldered and selected at purchase.
- Buyers who need maximum discrete-GPU performance: Unified memory provides capacity, but it does not recreate the behavior of a high-end discrete GPU with dedicated GDDR or HBM.
- People comparing only model numbers: A Ryzen AI Max 380, 385, 390, and 395 can have different CPU and GPU configurations, while OEM power limits can further separate real-world results.
- Users with a narrowly optimized software stack: Confirm application, driver, and accelerator compatibility before assuming that a large memory pool will translate into faster work.
- Desktop builders: This is not a socketed desktop CPU for a normal upgradeable DIY platform. Look for a complete compact system or workstation.
How to choose a Ryzen AI Max system
- Start with the workload. Decide whether your priority is gaming, CPU rendering, local language models, image generation, software development, or a mixture. The ideal memory capacity and cooling solution differ by workload.
- Confirm the exact processor. Look for Ryzen AI Max+ 395 if you specifically need the 16-core CPU and 40-CU Radeon 8060S. Do not assume that a product described merely as Ryzen AI Max has the flagship graphics configuration.
- Choose memory before appearance. Compare 32GB, 64GB, and 128GB versions carefully. The RAM is soldered, so the decision is effectively permanent.
- Check graphics-memory settings. Review the manufacturer’s documentation for Variable Graphics Memory and the maximum graphics allocation. Remember that a reserved graphics pool is still unified system memory.
- Check the sustained power design. Look for the device’s configured TDP or performance modes, cooling system, charger, and noise expectations. The 45W-to-120W range allows very different implementations.
- Verify the software path. For local AI, check the intended model runtime, GPU backend, driver version, and ROCm or other accelerator support. For professional applications, check certification and vendor support.
- Compare the complete system. Display quality, battery size, ports, storage, keyboard, warranty, and repair policy may matter more than a small difference in peak clock speed.
Windows driver maintenance after setup
Strix Halo does not require a third-party driver utility to work. AMD, ASUS, HP, and Windows support pages should remain the primary sources for BIOS, chipset, graphics, firmware, and device drivers.
Best Value
- 256 GB SSD of storage.
- Multitasking is easy with 16GB of RAM
- Equipped with a blazing fast AMD Ryzen 7 Pro 2.30 GHz processor.
If Windows has been reinstalled or Device Manager shows an unknown device, a narrowly bounded convenience option is Outbyte Driver Updater, which can scan for outdated or missing drivers and recommend updates. Use it as a diagnostic and maintenance aid—not as a way to unlock performance the hardware does not have, and not as a replacement for the OEM’s validated driver package. Create a restore point or backup first, and compare any recommendation against the laptop or workstation manufacturer’s support page before installing it.
What Strix Halo is—and is not
| Accurate description | Misleading description |
|---|---|
| A Ryzen AI Max processor with Zen 5, RDNA 3.5 graphics, and an NPU | A standalone socketed desktop CPU |
| Up to 128GB of coherent unified LPDDR5x memory | 128GB of upgradeable laptop RAM |
| Up to 96GB reserved for graphics in a 128GB configuration | 96GB of dedicated discrete-GPU VRAM |
| A flagship Radeon 8060S with 40 compute units | A 40-core CPU-style GPU |
| A platform whose results vary with power, cooling, software, and drivers | A processor that universally beats every RTX 4070, RTX 4090, or Apple silicon system |
Bottom line
AMD’s Strix Halo concept is compelling because it attacks a real limitation in compact computers: the separation between a relatively small pool of system RAM and an even smaller pool of mobile GPU memory. Ryzen AI Max+ 395 combines a 16-core Zen 5 CPU, a 40-CU Radeon 8060S, an NPU, and up to 128GB of shared LPDDR5x memory in one package.
That makes it especially interesting for local AI and mixed CPU/GPU workloads where model or dataset capacity matters. It can also enable powerful compact gaming and workstation systems. But the headline numbers need context. Unified memory is not dedicated VRAM, 96GB is a reservation rather than a separate graphics-memory bank, lower Max models have fewer compute units, and a 120W-capable chip will not perform the same in every chassis.
The right purchase decision therefore depends on the complete system: exact Ryzen AI Max model, soldered memory capacity, sustained power limit, cooling, software support, and warranty. For a consumer tablet-style system, start by checking the ASUS ROG Flow Z13 configuration. For professional work, examine the HP ZBook Ultra G1a. For local-AI development, consider the specialized Ryzen AI Halo platform if its availability and software environment match your needs.
Frequently Asked Questions
What is Strix Halo?
Strix Halo is AMD’s former codename for the Ryzen AI Max 300 Series. The flagship retail processor is the Ryzen AI Max+ 395.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Does Ryzen AI Max+ 395 have 96GB of dedicated VRAM?
No. A 128GB system can reserve up to 96GB for graphics through AMD’s Variable Graphics Memory feature, but that memory remains part of the unified LPDDR5x system pool. It is not equivalent to 96GB of dedicated GDDR or HBM VRAM.
Can you upgrade the RAM in a Strix Halo laptop?
No. Ryzen AI Max systems use soldered LPDDR5x memory, and the manufacturer determines the available capacity. Choose the required 32GB, 64GB, or 128GB configuration when purchasing.
Does every Ryzen AI Max processor have 40 GPU compute units?
The Ryzen AI Max+ 395 has 40 RDNA 3.5 compute units in its Radeon 8060S graphics engine. Other Ryzen AI Max models have fewer: the 390 and 385 have 32 CUs, while the 380 has 16 CUs.
Which Strix Halo system should you consider?
The ASUS ROG Flow Z13 is the clearest consumer gaming implementation, the HP ZBook Ultra G1a targets professional workstation users, and AMD Ryzen AI Halo is a specialized developer platform for local AI. Availability and exact configurations vary by region and vendor.
Recommended Free Tools
The Bottom Line
Bottom line: Strix Halo’s breakthrough is the combination of high-end integrated graphics and a large, coherent memory pool—not 96GB of dedicated VRAM. The Ryzen AI Max+ 395 is most compelling in 64GB or 128GB systems for local AI, rendering, and compact workstations, provided the device has adequate cooling and software support.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
