Intel has released more details about the flaws in the 13th and 14th generation Core processors of its Raptor Lake family and the 0x129 microcode that should prevent further damage.

The chipmaker previously disclosed that the so-called instability issue plaguing many Raptor Lake chips stems from an elevated operating voltage set by the processor itself. While relatively high voltage is crucial for maintaining stability at high clock speeds, there is a limit to what a processor can tolerate. Setting the voltage to dangerously high values ​​can permanently damage the processor, causing crashes at what should be stable frequencies or even burning the thing out altogether, as Intel customers have discovered.

Previously, it was unclear what Intel considered “elevated,” but thanks to new guidance from Chipzilla, we finally have an answer: anything above 1.55 volts.

In case you’re wondering, 1.55 volts – which presumably only occurs during single- and lightly-threaded workloads – is high for a desktop CPU in 2024, let alone anything above that. For comparison, Intel’s older 12th-gen CPUs based on Alder Lake tended to max out at around 1.4 to maybe as high as 1.45 volts. AMD’s Ryzen 7000 CPUs are also in the same neighborhood, and the latest Ryzen 9000 chips likely will as well.

Not entirely surprisingly, 13th and 14th Gen Core CPUs require this voltage level to hit their maximum boost frequencies (again, on one or two cores), which are the highest for desktop processors. While AMD has only been able to hit 5.7GHz and Alder Lake topped out at 5.5GHz, the fastest Raptor Lake CPUs range from 6 to 6.2GHz.

It’s unclear what the operating voltage was before the introduction of the 0x129 microcode, but apparently 1.55V is the ideal value to prevent damage while still ensuring high clock speeds. According to Intel’s internal tests, the performance of Raptor Lake CPUs has not changed, except in a few cases. Hardware benchmarks will likely have the final say on this, though.

Overclocking is still an option for owners of K-class chips, but given the circumstances, pushing Raptor Lake processors might not be such a good idea.

Intel also says that affected processors that aren’t completely dead will have higher minimum operating voltages across multiple cores. We’ve reached out to Intel for more details on the implications of this.

Motherboard vendors begin delivering BIOS updates

This announcement comes just after Intel’s motherboard partners began releasing BIOS patches with the new microcode for their LGA 1700 motherboards. MSI has promised to update all of its 600- and 700-series motherboards by the end of the month, and has already started doing so by releasing beta BIOSes for its highest-end Z790 boards. ASRock, meanwhile, has quietly been pushing out updates for all of its 700-series motherboards.

ASUS says it will begin releasing new BIOS systems next week, while in a statement to The registerGigabyte has only committed to completing the patches for its motherboards by the second week of September. Dell told us, “Dell recently received the Intel BIOS update and our engineers are working quickly to validate it with our systems. We expect to make the validated updates available to our customers next week.”

However, it’s not yet clear how BIOS updates will play out for owners of pre-built OEM systems. The register has also reached out to HP to ask how they will be handling the updates for their 13th and 14th generation desktops, but we have not yet received a response.

Meanwhile, Intel has postponed its Innovation event, scheduled for mid-September.

Side note

Intel rival AMD released two security advisories this week detailing some interesting but not too scary flaws in its chips. One of them, dubbed Sinkclose, also known as CVE-2023-31315, allows bad actors who have gone so far as to compromise a machine’s kernel, on ring 0, to then execute code in System Management Mode, infecting the computer in a way that is difficult or impossible to remove. See AMD’s write-up for details.

Another advisory, for CVE-2024-21978, CVE-2024-21980, and CVE-2023-31355, warns that malicious hypervisors can access guest memory that should actually be protected by AMD’s Secure Encrypted Virtualization. Again, see the writeup for next steps.