Rumor Claims Intel Will Forcibly Disable AVX-512 On 12th Gen Alder Lake Chips But Why?
The saga of AVX-512 on Alder Lake has been an interesting one. Originally, folks reasonably assumed that the 12th-Gen Core processors would support AVX-512 extensions. After all, it was supported on the last-generation Rocket Lake chips, to say nothing of Ice Lake before that. Then we found out about Alder Lake's hybrid CPU core configuration. And shortly after, Intel accidentally confirmed that AVX-512 would not work on Alder Lake because its Gracemont Efficiency cores simply don't support AVX-512. If you're not familiar, AVX-512 is a set of instructions that can accelerate workloads "such as scientific simulations, financial analytics, artificial intelligence (AI)/deep learning, 3D modeling and analysis, image and audio/video processing, cryptography and data compression."
More recently, Intel released a somewhat controversial Alder Lake developer guide, stating again that AVX-512 would in fact work—as long as you disabled the E-cores first. Intel immediately disavowed the guide, but of course, as these things go, it was proven correct in the end. Developers around the world still disagree as to the value of the on-again-off-again instruction set support for desktop users, but as just one example, gamers playing on console system emulators like RPCS3 and Yuzu can see a nice little performance boost with AVX-512 enabled.
Still, not everyone is willing to install what could be a clumsily-modified firmware in their motherboard from an unknown source. We shouldn't have to explain how that could open up a whole can of worms for both stability and security. Folks fearful of funky firmware, as well as regular users who don't know the difference, are left with two options: install the latest firmware and lose out on one of the capabilities of their processor, or stick with an older version and miss out on improvements in later updates—particularly improvements in DDR5 memory support.
For its part, if we asked, Intel would surely say that AVX-512 was never officially supported on Alder Lake consumer CPUs to begin with, and that the folks fiddling around to make it happen are scoundrels and villains. (Maybe not that last part.) It's still strange to see Intel moving in this direction, though. After all, AVX-512 is a feature that adds value for some users; while it's fairly unlikely anyone's decision to purchase a 12th-gen Core processor hinged on AVX-512 support, there's certainly a small subset of users reaping the benefits of their new CPU's wide vector math capability, and disabling it retroactively reduces the value of their processor.
It's possible that Intel simply wants to avoid warranty claims from users whose AVX-512 units are defective. Ensuring that the processor is forced into an "official" state isn't great (especially on an "unlocked" K-series CPU), but it's at least not totally without reason. It's also possible that Intel is concerned that its upcoming P-Core-only Alder Lake CPUs will gain undocumented AVX-512 support. Support for the high-end feature only on lower-end CPUs would put the company's product stack in a strange, upside-down position, although it's not the first time such a thing has happened—note the case of ECC support on Pentium and Core i3 CPUs where it is absent on Core i5 and Core i7 models.
Alternatively, perhaps this is just market segmentation, and Intel simply wants to prevent its desktop Core-family processors from cannibalizing the sales of desktop Xeon processors based on the same processor die. Whatever the case, if you're interested in using AVX-512 on your 12th-Gen Core CPU, you'd better be watching the patch notes on your motherboard's BIOS updates like a hawk.
