When you hear “Framework”, if you’re familiar with the laptop vendor, you probably think of modular components and repairable PCs. Indeed, the company allowed popular repair guide website iFixit early access to the very exciting Framework Desktop, and the machine is as easy as expected to tear down given that it is, of course, a desktop PC. The RAM remains soldered, though, and it seems like that’s simply going to be the case for the foreseeable future due to signal integrity concerns.
Having a wide bus isn’t enough, though. GPUs typically use super hot-clocked GDDR6, GDDR6X, or GDDR7 memory that runs at speeds upwards of 20 gigabits per second (20,000 MT/s). It does this by sacrificng latency in favor of transfer rate, which is fine on a GPU that can mask the latency with architectural design. General purpose CPUs don’t work that way, and need as low memory latency as possible. To that end, Ryzen AI Max still uses LPDDR5X RAM, which only clocks up to around 8000 MT/s (or possibly as high as 9600 MT/s, soon.) The double-wide memory bus allows the GPU to remain fed while running games at relatively high resolutions and frame rates.
The upshot of all this is that Ryzen AI Max requires high-speed LPDDR5X memory, and that kind of RAM doesn’t come in a socketed form. Framework CEO Nirav Patel explained to Linus Sebastian that the company worked closely with AMD to try and resolve this issue. AMD even purportedly assigned an engineer specifically to this project. Patel explained:
So we did actually ask AMD about this the first time they told us about Strix Halo. It was literally our first question: ‘How do we get modular memory? We are Framework after all.’ And they didn’t say no actually, they did assign one of their technical architects to really really go deep on this—they ran simulations, they ran studies, and they just determined it’s not possible with Strix Halo to do LPCAMM. The signal integrity doesn’t work out because of how that memory’s fanning out over the 256-bit bus.
There’s a lot of skepticism around the Framework desktop because of both its relatively high price at the entry-level—entry-level Mac Minis are both more potent and cheaper—and also because of its relatively poor upgradeability compared to typical desktop systems. These are fair complaints; you will get more for your money in a traditional desktop with replaceable CPU, RAM, and GPU than you will with the Ryzen AI Max 385-based machine.
It’s really the top-end Ryzen AI Max+ 395 models that justify this product, and the price there can’t be beat, especially considering the unique capabilities of the Ryzen AI Max+ processor. This machine really couldn’t exist without the repairability compromises that Framework had to make, and kudos to them for choosing to bring this killer product to market.