Latest news with #AMDRyzenAIMax
Tom's Guide
2 days ago
- Tom's Guide
I tested the Framework Desktop — and it made me rethink everything I knew about mini PCs
The Framework Desktop is a powerful and highly customizable mini PC that's equally well-suited for work, gaming and local AI workloads. But we are talking about Framework, so it's much more than that. Just like with Framework's laptops, it's incredibly easy to build and repair thanks to its replaceable parts. However, unlike with other mini PCs, it's also very customizable since you can use the company's Expansion Cards to configure its front ports to your liking and its magnetic front panel can be personalized using a number of different swappable tiles. The only components you can't swap out are its AMD Ryzen AI Max processor and LPDDR5x memory, which is soldered to the motherboard due to this chip's limitations. You do get to pick from three CPU and memory configurations when putting together your Framework Desktop though. Likewise, you can either pay extra for storage and the computer's operating system at checkout or you can bring your own to save a bit of cash. My Framework Desktop review will help you decide if this is the best mini PC for your needs and workflow or if you'd be better off with something smaller, cheaper and less customizable. For me though, this computer is the small form factor desktop I've always wanted. Here's why. Max 385 (32GB) Max+ 395 (64GB) (as tested) Max+ 395 (128GB) Price $1,099 1,599 $1,999 CPU AMD Ryzen AI Max 385 AMD Ryzen AI Max+ 395 AMD Ryzen AI Max+ 395 GPU Radeon 8050S Radeon 8060S Radeon 8060S NPU Up to 50 TOPS Up to 50 TOPS Up to 50 TOPS RAM 32GB LPDDR5x 64GB LPDDR5x 128GB LPDDR5x Storage 2 x M.2 SSD from 500GB to 8TB 2 x M.2 SSD from 500GB to 8TB 2 x M.2 SSD from 500GB to 8TB Ports 2 x Expansion Cards (Front), 2 x USB-A, 2 x USB-C, 1 x HDMI 2.1, 2 x DisplayPort 2.1, 1 x 5 GbE Ethernet (rear) 2 x Expansion Cards (Front), 2 x USB-A, 2 x USB-C, 1 x HDMI 2.1, 2 x DisplayPort 2.1, 1 x 5 GbE Ethernet (rear) 2 x Expansion Cards (Front), 2 x USB-A, 2 x USB-C, 1 x HDMI 2.1, 2 x DisplayPort 2.1, 1 x 5 GbE Ethernet (rear) Connectivity Bluetooth 5.4, Wi-Fi 7 Bluetooth 5.4, Wi-Fi 7 Bluetooth 5.4, Wi-Fi 7 OS Windows, Linux Windows, Linux Windows, Linux Size 3.81 × 8.09 × 8.9 inches 3.81 × 8.09 × 8.9 inches 3.81 × 8.09 × 8.9 inches Weight 6.83 pounds (3.1kg) 6.83 pounds (3.1kg) 6.83 pounds (3.1kg) The Framework Desktop builds on the company's experience with customizable and easily repairable laptops but in a mini PC form factor. Everything about this mini PC from its unique case and customization options, to the fact that you have to build it yourself make it stand on its own. However, it's the extra power under the hood from that AMD Ryzen AI Max processor that makes the Framework Desktop such a versatile desktop or laptop replacement. If you've ever built a PC before, you're in for a familiar but much simpler experience with the Framework Desktop. This is because its motherboard, CPU, RAM and power supply come pre-installed. In fact, you don't even have to plug in any cables either. Putting the Framework Desktop together starts with opening its case. From there, you need to install your M.2 SSD and the CPU fan before closing everything up and loading either Windows or Linux onto your new PC. Get instant access to breaking news, the hottest reviews, great deals and helpful tips. In total, it took me around 15 minutes to put everything together, though I did stop and take pictures and videos along the way. I found the whole process to be quite refreshing compared to building a traditional PC where simple mistakes can not only be time consuming but also quite expensive. I did have to download Windows 11 and install the operating system from scratch but this way, you don't have to worry about any of the bloatware you find on other pre-built PCs or laptops. When configuring your Framework Desktop, the parts you pick can give this mini PC quite a different look once everything's put together. For instance, I went with a sleek, all-black design with more of a business-focused look to it. However, I could have easily chose a CPU fan with RGB lighting and a translucent side panel for something that looks more like one of the best gaming PCs than a sleeper build. At the time of writing, there are actually three CPU fan options to choose from but you can also get a fan mounting kit and use your own fan instead. Given the all-black look I was going for, I went with the Cooler Master Mobius 120 fan, though Framework also sent over one for Noctua for me to check out as well. Given that the Framework Desktop's side panel slides right off, swapping out one fan for another later on would be quite easy to do. What excited me most about the Framework Desktop when it was first announced was its customizable front panel. It has 21 slots for plastic tiles that you can easily snap in with your thumb. You can mix and match colors, or choose tiles with logos like Framework or AMD. Framework even sent one with the Tom's Guide logo, and I expect more custom options will come as the system grows. Behind the front panel, there are two open slots for Framework's Expansion Cards, which let you choose what ports you want on the front of the mini PC. These cards come in different colors and finishes to match the look of your setup. For my review unit, I chose a USB-A and an SD card reader, but I also really liked the translucent pink USB-C card. At 3.81 × 8.09 × 8.9 inches, the Framework Desktop is definitely on the large size for a mini PC. I'd say it straddles the line between a more cube-shaped mini PC like the Geekom A6 and a true small form factor desktop PC case due to its size. Still, it's small enough to slip into a backpack and the optional carrying handle makes it very easy to move this computer from one room to another. Even though it's compact, the Framework Desktop is powerful thanks to its Ryzen AI Max chip. There are two CPU options — one with 8 cores and 32 graphics cores, and another with 16 cores and 40 graphics cores. You can also choose between 32GB, 64GB, or 128GB of memory. I tested the mid-range setup with an AMD Ryzen AI Max+ 395, 64GB of RAM, and a 1TB SSD. With 16 cores and 32 threads, it was more than I needed for my usual tasks, which mostly involve running several Chrome windows with lots of tabs and doing some photo editing in GIMP. Although there aren't any extra Expansion Card slots around back, you still get a nice selection of rear ports which include two USB-A, two USB 4, an HDMI 2.1 port, dual DisplayPort 2.1 connections, a 5 GbE Ethernet port and a headphone jack. While I wish there were a few more USB-A ports for my older devices, one thing I really appreciate about the Framework Desktop is that as it has an internal power supply, you can use a standard PC power cable. Of the mini PCs I've tested only the Mac mini M4 has an internal power supply while the rest use bulkier external ones. Framework Desktop AtomMan G7 Ti MSI Cubi NUC Geekbench 6.4 single-core 2966 2864 2712 Geekbench 6.4 multi-core 17574 15651 8740 25GB file copy test (MBps) 2976.46 1445.75 1337.70 Handbrake (Mins:Secs) 2:43 3:50 9:24 I'd say the base version of the Framework Desktop will be more than enough for most people but you could upgrade to the version with 64GB of RAM for future proofing or for playing games (more on that later). Meanwhile, the 128GB version of this mini PC is best suited for those that want to run AI models like Llama locally. Another thing that really surprised me about the Framework Desktop was just how fast my download speeds were when using it with one of the best Wi-Fi 7 routers. It comes equipped with both Wi-Fi 7 and a 5 GbE Ethernet port for fast wireless and wired connectivity. For instance, when running a speed test over Wi-Fi, I saw download speeds of 1.52 gigabits per second (Gbps). Then when I plugged it into my Eero Max 7 mesh router using an Ethernet cable, I was able to take almost full advantage of the bandwidth offered by my 10 gigabit fiber connection. All in all, the Framework Desktop delivers and then some both on the performance and the connectivity front. As I mentioned before, the Framework Desktop is a very versatile mini PC and as such, it can be whatever kind of computer you need it to be at the moment. While I spent a lot of time using it as a work machine during the day, at night, it effortlessly transformed into a gaming PC. Based on the gaming benchmarks ran in our testing lab, you'd think that the Framework Desktop is best suited for 1080p or even 1440p gaming. However, in my testing on the MSI MPG 321URXW QD-OLED 4K gaming monitor at home, I walked away quite impressed after playing several more demanding PC games like Final Fantasy 7 Rebirth and Ratchet & Clank: Rift Apart. Framework Desktop AtomMan G7 Ti Asus ROG NUC 970 Assassin's Creed Mirage 85 103 97 Metro Exodus 56.56 91.63 91.78 Cyberpunk 2077 22.09 39.90 39.33 Red Dead Redemption 2 56.57 73 62.94 While the latest entry in the FF7 remake series doesn't support AMD's FSR tech, I still managed to get a smooth frame rate of 60 fps when I enabled Radeon Super Resolution in the pre-installed AMD Software: Adrenaline Edition app. The same was true in Ratchet & Clank: Rift Apart which does come with support for FSR 3.1 Obviously upscaling and frame-gen tech aren't for everyone and in that case, you'll get excellent results at 1440p and especially at 1080p with it turned off on the Framework Desktop. Still, being able to play newer and more demanding titles in 4K using an integrated GPU is quite impressive. Likewise, with two USB 4 ports around back, you could always plug in an external graphics card too. It's also worth noting that if you opt for a Framework Desktop with more memory, you can allocate extra RAM to this mini PC's Ryzen AI Max chip for improved graphics performance. For instance, in AMD's app, I was able to split my review unit's memory in half with 32GB going to the system and the other 32GB going to its graphics. The Framework Desktop is an impressive debut into the world of mini PCs by a company known for its repairable and upgradeable laptops. However, it does have its downsides too which include this PC's higher price as you are buying into an ecosystem after all and the fact that you can't upgrade its RAM or processor. The Framework Desktop makes building a PC easier, but it's not as flexible as a true barebones setup. You only get two CPU options and a few RAM configurations. The Ryzen AI Max is a strong processor, so it makes sense for Framework's first desktop — but the downside is that the RAM is soldered in and can't be upgraded later if your needs change. You also don't save much money compared to other pre-built PCs, since the price doesn't include a CPU fan, storage, Expansion Cards, or a Windows license if you don't want to use Linux. Overall, I really like the Framework Desktop and have had a wonderful experience building and using it for the past two weeks. Your money could go further with a full-size desktop PC especially when you factor in sales on components. However, that would defeat the purpose of the Framework Desktop. At the same time, as someone who has built small form factor PCs in the past, you're always going to pay a premium for smaller parts like mini ITX motherboards, SFX or SFX-L power supplies and of course, low-profile graphics cards. As such, if you're thinking about buying your own Framework Desktop, then you're definitely going to want to think long and hard about your needs and workflow before committing to a specific configuration. While the base configuration will likely meet the needs of most people, opting for the middle of the road one like my review unit or even the fully equipped configuration if you have the budget for it might make more sense for futureproofing. The Framework Desktop stands out from typical mini PCs, and that's one of its biggest strengths. Instead of a fully pre-built design, it lets you customize the look and features with its tiled front panel and Expansion Cards. You can also choose different side panels and CPU fans — or even use your own — making it a much more personal and flexible experience than most other mini PCs. While the soldered RAM and being limited to two variations of one processor might not be for everyone, this decision is what allows the Framework Desktop to be readily mass produced instead of a more niche, boutique-style mini PC. Likewise, some PC gamers might scoff at the idea of gaming on an integrated GPU but the Ryzen AI Max's Radeon 8050S or 8060S graphics get the job done while allowing you to play a wide variety of both newer and older games. The Framework Desktop could be a new way forward for mini PCs as well as small form factor desktop overall. Only time will tell but so far, it has rightfully earned its place among the best mini PCs you can get today.
Techday NZ
31-07-2025
- Business
- Techday NZ
AMD brings 128B LLMs to Windows PCs with Ryzen AI Max+ 395
AMD has announced a free software update enabling 128 billion parameter Large Language Models (LLMs) to be run locally on Windows PCs powered by AMD Ryzen AI Max+ 395 128GB processors, a capability previously only accessible through cloud infrastructure. With this update, AMD is allowing users to access and deploy advanced AI models locally, bypassing the need for third-party infrastructure, which can provide greater control, lower ongoing costs, and improved privacy. The company says this shift addresses growing demand for scalable and private AI processing at the client device level. Previously, models of this scale, such as those approaching the size of ChatGPT 3.0, were operable only within large-scale data centres. The new functionality comes through an upgrade to AMD Variable Graphics Memory, included with the upcoming Adrenalin Edition 25.8.1 WHQL drivers. This upgrade leverages the 96GB Variable Graphics Memory available on the Ryzen AI Max+ 395 128GB machine, supporting the execution of memory-intensive LLM workloads directly on Windows PCs. A broader deployment This update also marks the AMD Ryzen AI Max+ 395 (128GB) as the first Windows AI PC processor to run Meta's Llama 4 Scout 109B model - specifically with full vision and multi-call processing (MCP) support. The processor can manage all 109 billion parameters in memory, although the mixture-of-experts (MoE) architecture means only 17 billion parameters are active at any given time. The company reports output rates of up to 15 tokens per second for this model. According to AMD, the ability to handle such large models locally is important for users who require high-capacity AI assistants on-the-go. The system also supports flexible quantisation and can facilitate a range of LLMs, from compact 1B parameter models to Mistral Large, using the GGUF format. This isn't just about bringing cloud-scale compute to the desktop; it's about expanding the range of options for how AI can be used, built, and deployed locally. The company further states that performance in MoE models like Llama 4 Scout correlates with the number of active parameters, while dense models depend on the total parameter count. The memory capacity of the AMD Ryzen AI Max+ platform allows users to opt for higher-precision models, supporting up to 16-bit models through when trade-offs between quality and performance are warranted. Context and workflow AMD also highlights the importance of context size when working with LLMs. The AMD Ryzen AI Max+ 395 (128GB), equipped with the new driver, can run Meta's Llama 4 Scout at a context length of 256,000 (with Flash Attention ON and KV Cache Q8), significantly exceeding the standard 4,096 token window default in many applications. Examples provided include demonstrations where an LLM summarises extensive documents, such as an SEC EDGAR filing, requiring over 19,000 tokens to be held in context. Another example cited the summarisation of a research paper from the ARXIV database, needing more than 21,000 tokens from query initiation to final output. AMD notes that more complex workflows might require even greater context capacity, particularly for multi-tool and agentic scenarios. AMD states that while occasional users may manage with a context length of 32,000 tokens and a lightweight model, more demanding use cases will benefit from hardware and software that support expansive contexts, as offered by the AMD Ryzen AI Max+ 395 128GB. Looking ahead, AMD points to an expanding set of agentic workflows as LLMs and AI agents become more widely adopted for local inferencing. Industry trends indicate that model developers, including Meta, Google, and Mistral, are increasingly integrating tool-calling capabilities into their training runs to facilitate local personal assistant use cases. AMD also provides guidance on maintaining caution when enabling tool access for large language models, noting the potential for unpredictable system behaviour and outcomes. Users are advised to install LLM implementations only from trusted sources. The AMD Ryzen AI Max+ 395 (128GB) is now positioned to support most models available through and other tools, offering flexible deployment and model selection options for users with high-performance local AI requirements.
