Intel’s Arc graphics cards have been seriously underwhelming since they hit the desktop – with a China-only release – but there was a brief flurry of excitement surrounding a rumor that Team Blue might have an ace up its sleeve. Unfortunately, though, it wasn’t meant to be.
The ace in question was multi-GPU support, meaning that multiple graphics cards could be installed on the same PC and work together to boost gaming performance levels – which, if Intel could get it right, would be a huge advantage against AMD and Nvidia. (Team Red’s CrossFire is dead, as is Team Green’s SLI effectively, though it’s still kicking over the top in a way).
News that Intel was planning a multi-GPU innovation – which was supposed to be showcased at SIGGRAPH 2022 but didn’t make it to the cut – came from TweakTown (opens in new tab)but the tech site later reported that after getting the story out, Intel reached out to clarify that support for multiple boards was only happening to a very limited extent.
Intel told TweakTown: “Intel showed a rendering demo of Blender Cycles in SIGGRAPH with Intel Arc graphics. Multi-GPU rendering support for Intel Arc and Intel Arc Pro graphics cards through oneAPI is supported as of Blender 3.3. Intel Arc graphics does not support multi-GPU for gaming.”
This effectively debunks the idea that PC gamers could benefit from accelerated frame rates with more than one Arc GPU on their machine.
Analysis: Okay, so not now – but maybe in the future?
We can see why people jumped on it and got so excited. The entire desktop release of Arc saw discrete GPUs coming out of the gate, and the bad news about drivers and stability was compounded by more concerns about whether the entire project should be cancelled. The latter is something Intel has vehemently denied, mind.
Either way, we’re all very disappointed in the state of Arc Alchemist right now, so a glimmer of hope of something unexpected would inevitably be seized upon. Especially since in the early days, Intel talked about going the multi-GPU route with gaming, and as we noted at the time, this seemed like a viable way for Team Blue to significantly differentiate their desktop graphics cards from AMD and Nvidia.
Of course, just because this isn’t happening with Alchemist doesn’t mean Intel isn’t going to look at a gaming implementation on a future generation of GPUs. If we get that far, but of course the concern is that with the driver side of the equation already being seriously problematic for Intel, further complicating things with a multi-card implementation – a notoriously tricky business, as gamers have already know from previous experience with SLI and the like – doesn’t seem like a sensible idea.