My recent look at Alienware's awesome Area51-m Extreme, replete with 3.2GHz Extreme Edition P4 and swappable graphics modules, made this article attractive. I'm not usually one to cover developments in the mobile space, but given the advent of swappable graphics in recent notebook designs from a few vendors and HEXUS' continued interest in consumer graphics, a look at MXM was definitely on the cards.
MXM, short for Mobile PCI-Express Module, is NVIDIA's, with the help of major notebook ODMs and other notebook GPU vendors, attempt to standardise the physical graphics interface for notebook PCs.
The current state of play is that notebook ODMs use completely custom physical packaging for their notebook GPUs. Those custom packaging designs are often model specific, never mind range specific, so GPU packaing designs aren't transferable to new laptop designs. This proprietary way of doing things, which even extends to something like Alienware's swappable modules, since they are Area51 specific and aren't used by other notebook vendors, has a couple of appreciable knock-on effects for mobile graphics.
The cost and time of designing completely new GPU packages for every new notebook design means that time to market for notebooks using a new GPU is often in the range of 12+ months. That can be 12+ months after that GPU tapes out. The added cost of new designs each time also means higher cost for the consumer, as the ODM passes on R&D work costs to you.
Much like VLB, ISA, MCA, PCI and AGP et al, before it, the PCI-Express standard seeks to offer GPU makers the ability to standardise on packaging and interface for new GPU designs. The benefits are pretty obvious. Time to market for significantly new GPUs is reduced, costs are less, thermal considerations are well known and understood. The list goes on.
At its heart, MXM is simply a slightly customised PEG16X interface for mobile graphics, with standardised module sizes and thermal designs, so that packaging an MXM design is easy and set in stone.
The Physical Interface
The connector is a wrapper for a full PEG16X implementation and the display output from the GPU, so that the PCI Express 16X interface and the display output can be routed in a standard manner.
NVIDIA's documentation for MXM states that much like a desktop PC with existing onboard graphics, inserting an MXM module into a notebook with existing onboard graphics will simply disable them and route the display output of the MXM module correctly.
The ModulesThe MXM standard specs three module packaging configurations, imaginatively named MXM-I, MXM-II and MXM-III. As expected, each is a different size, mainly offering the ability for increased memory densities the larger the module gets. NVIDIA's press documentation shows a 128MB GeForce 6 Go on an MXM-I module.
Each MXM module size uses the same base mounting holes, so that MXM-I modules can fit in a board capable of MXM-II modules, and MXM-I and MXM-II modules will fit in a board that can house MXM-III hardware. Since the GPU on any MXM design will sit in the same place, the thermal solution for all MXM types is the same. NVIDIA have a pair of reference cooler designs but partners are free to spec whatever works for the ultimate form factor that they're using.
What it means for consumersWell, in theory quite a bit. Standard form factor, standard physical connector and some standard on thermal design at first seems to vindicate NVIDIA's initial assertions that MXM will reduce costs and time to market. You'll see most coverage of MXM today regurgitate that same press-pushed mantra.
In practice however, it doesn't seem to mean too much at all, at least initially. It's all down to overall mobile form factors. While the standardisation of a mobile graphics form factor like MXM is a good thing, the fact that mobile designs like notebooks, sub-notebooks, desktop replacements are such a moving target in terms of size and their own form factor, that augmenting MXM to a large majority of those designs is going to be just as difficult as a proprietary module.
Upgradability of the GPU in a mobile device isn't the attractive panacea that drives desktop PCs. Think of fixed graphics devices like Nintendo's GameCube or an Xbox and you're most of the way towards understanding why swapping the GPU in a notebook doesn't matter to the average man on the street.
While 3D graphics continues to drive technology forward in desktop PCs, especially now where the newest GPUs will sit and wait for the CPU in a vast array of rendering setups, the challenges of getting that same scenario in a portable device that isn't a desktop replacement is a lot harder.
Ultimately it's a good idea and NVIDIA are to be applauded for being the driving force behind MXM. MXM isn't an NVIDIA GPU-only solution either. ATI are commited to providing MXM form factor modules for any ODM customer that requests them. But MXM is just one form factor in a sea of them in the mobile space, and that's discounting the massive number of proprietary designs that are still being churned out.
Parting notesWe here from some sources that MXM isn't without its fair share of problems. Signalling and shielding at the connector being just one set of issues that MXM implementors will need to overcome. Here's hoping they are minor bumps along the road.
NVIDIA's biggest chance of success with MXM comes from the Go variant of their new GeForce 6 series of GPUs and uptake from vendors that also wish to push non-NVIDIA GPUs onto the form factor. Strong performance and a solution based around it that's attractive to ODMs (cost and support wise), taking away some of the shine from ATI and M2x, their next-generation FLEXFIT (ATI's own V-MAP style spec for pin-compatible GPU replacement) PCI-Express parts, will help drive acceptance of MXM into the mobile space.
However, given that mobile form factors often favour form over function, creating a giant moving target of designs to shoehorn MXM into, it's not going to be easy.
We should have an MXM chassis and a couple of MXM modules for functional testing soon, look for an article when that happens.