Why are AMD GPUs causing CPU Overclocking to its Graphics drivers?

In September 2021, AMD added support for simple CPU overclocking to its graphics drivers. If you had a Ryzen 5000-series CPU and wanted to benefit from the extra performance, this auto-overclocking function could save you from needing to download the more complex Ryzen Master utility. The overclock would also be conservative enough that it probably wouldn’t cause system instability or other issues.

 

Reason behind the CPU Overclocking

 

The problem for some users is that this auto-overclocking feature has become too automated—that is, it’s changing systems’ overclocking settings whether users want it to or not. According to one of the AMD representatives, an issue in the AMD software suite caused the feature to begin adjusting certain AMD processor settings for some users. Because the CPU overclocking feature is actually changing settings in your system’s BIOS, that means it can change overclocking settings that users have changed themselves and apply an overclock where there was no overclock before.

That second bit could be especially problematic since overclocking processors generally voids AMD’s CPU warranty, even when you’re using AMD-provided tools like Ryzen Master or using AMD-advertised features like Precision Boost Overdrive (though, anecdotally, this policy isn’t consistently enforced).

 

AMD GPUs with Intel CPUs

 

PCs using AMD GPUs with Intel CPUs won’t be affected by this bug since AMD’s software won’t overclock Intel’s processors. According to testing by Igor’s Lab and by users on the AMD subreddit, the bug appears to be triggered when loading a GPU overclocking profile created using older versions of the software. Those profiles were created before CPU overclocking was added to the software, but the software isn’t aware of that and attempts to modify the CPU settings anyway.

AMD says that it is “investigating” the software issue causing the problem and that the company will “share more information as soon as we’re able.” The issue appears to have been introduced in version 22.3.1 of AMD’s Adrenalin driver package, so holding off on an update or rolling back to a previous driver version may prevent or resolve the issue.

 

Also Read: Will YouTube roll out its ‘YouTube comment spam’ to tackle spammy remarks?

How was Intel’s removal of Alder Lake AVX-512 support a good decision?

The fluctuating saga of 12th Gen AVX-512 support appears to be off for good this time. According to an Intel spokesperson: “Although AVX-512 was not fuse-disabled on certain early Alder Lake desktop products, Intel plans to fuse off AVX-512 on Alder Lake products going forward.

 

The AVX-512

 

When Intel launched its 12th Gen range back in November, the AVX-512 was not enabled and the fiasco was accepted by a majority of the tech press. As per Intel, it wasn’t enabled due to the inclusion of two different architectures—the E cores did not support it although the P cores did.

Motherboard manufacturers used this bit of knowledge to cheekily allow users to enable AVX-512 after disabling the E cores. As a result, Intel issued new BIOS microcode to close off this loophole. However, that wasn’t the end of it, as MSI released a BIOS that got around the block. That leads us to Intel’s decision to fuse off AVX-512 in silicon.

 

Irrelevant to Users in Certain Apps

 

It’s the nuclear option, and it means no more workarounds to enable AVX-512 support. Users will still have the option if they are using an earlier batch 12th Gen CPU and don’t update their BIOS, but in a month or two or three from now, all 12th Gen stock will have AVX-512 blocked off completely.

AVX-512 instructions aren’t really relevant to gamers but in certain applications, they can result in dramatic performance gains. Some, like Linux creator Linus Torvalds, hate it. Its use leads to excessive power consumption and heat generation, and perhaps Intel decided that the PR hit from nuclear reactor memes wasn’t worth the trouble anymore. But that might not be the end of AVX-512 for consumer CPUs. There are rumors that AMD will include it in its upcoming Zen 4 CPUs. That would be a twist.

 

Steady Decision to Remove Feature

 

According to a motherboard manufacturer, Intel’s decision to remove the feature was made late in the design stage. That upset some in the industry as motherboards were designed with steep AVX-512 power requirements in mind. That doesn’t matter with premium boards but it adds expense to lower-tier models which are already more expensive due to various supply chain issues.

AVX-512 support isn’t the only thing Intel is unhappy with. There’s also non-K overclocking, though the feature is so far limited to expensive motherboards that are unlikely to be paired with cheap CPUs in meaningful numbers. If a vendor were to release a cheaper B660 DDR4 motherboard that could do it, it would make me and a lot of gamers very happy, but Intel would likely be stirred into action again.

 

Also Read: Nvidia hackers target Samsung, release 190GB of sensitive data