Intel tells mobo makers to go easy on the BIOS settings amid CPU instability reports
- Reference: 1714407307
- News link: https://www.theregister.co.uk/2024/04/29/intel_cpu_instability_motherboards/
- Source link:
"Intel requests system and motherboard manufacturers to provide end users with a default BIOS profile that matches Intel recommended settings," Intel said in an email to motherboard vendors published by [1]Igor's Lab .
The chipmaker provides a default list of specifications for its CPUs, also known as stock settings. However, many motherboard manufacturers configure their boards with different settings by default, and on higher-end motherboards these differences usually include higher power limits, higher current limits, and other settings that are intended to increase out-of-the-box performance.
[2]
MSI, for instance, last month released BIOS updates that allow users to disable [3]Current Excursion Protection for even more performance. Additionally, this Taiwan-based manufacturer sets exceptionally high default power limits – up to 4,096 watts and 512 amps – on some of its high-end motherboards.
[4]Alibaba Yitian 710 rated fastest Arm server CPU in the cloud (for now)
[5]Intel excited by PC sales pop and GPU prospects, but investors aren't because the outlook is poor
[6]TSMC says first 1.6nm chips coming in 2026
[7]Intel Foundry ticks another box in quest to fab mil-spec chips for US DoD
Obviously, no consumer-grade Intel CPU will ever hit that much power or current, but Intel says most reports are coming from users who have motherboards loaded with these kinds of settings by default. In addition to using its default specifications, Intel also asks motherboard vendors to add warnings to performance-enhancing BIOS options so that users are aware that enabling such settings could result in system instability.
However, Intel has not yet determined or disclosed the precise reasons why these settings lead to issues. "Intel has observed that this issue may be related to out of specification operating conditions resulting in sustained high voltage and frequency during periods of elevated heat," Intel said in a statement. "Analysis of affected processors shows some parts experience shifts in minimum operating voltages which may be related to operation outside of Intel specified operating conditions."
[8]
Intel's explanation indicates that the performance-boosting settings that are prone to causing [9]stability issues may require chips to use more voltage to work correctly. However, it's not clear if instability is caused by a lack of voltage, something that can easily affect chips, particularly with an overclock, or by an excessive amount of voltage, which could theoretically cause physical damage.
Intel plans to make a public announcement regarding its findings in May. We've asked Intel to comment further and will let you know if the company gets back to us. ®
Get our [10]Tech Resources
[1] https://www.igorslab.de/en/intel-releases-the-13th-and-14th-generation-k-sku-processor-instability-issue-update/
[2] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2ZjAYf4RiKVa8OBPKasl6fAAAAEU&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[3] https://www.msi.com/news/detail/Stay-Cool-and-Perform-Better--MSI-BIOS-Update-Allows-to-Disable-CEP-Function-for-Intel-14th-Gen-Processors-143359
[4] https://www.theregister.com/2024/04/29/arm_cloud_server_database_performance/
[5] https://www.theregister.com/2024/04/26/intel_q1_2024/
[6] https://www.theregister.com/2024/04/25/tsmc_first_16nm_chips/
[7] https://www.theregister.com/2024/04/23/intel_foundry_military_approval_nears/
[8] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_onprem/systems&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44ZjAYf4RiKVa8OBPKasl6fAAAAEU&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[9] https://www.theregister.com/2024/02/23/intel_core_13th_crash/
[10] https://whitepapers.theregister.com/
Re: Pushing the envelope
Even nVidia had to come out with a statement saying... it's not us... It's intels fault
Re: Pushing the envelope
Organalog?
There's a reason for those data sheet timing diagrams...
I've been following this via a few techtubers I follow... The blame lies squarely with intel on this one... they greenlit the board partners to basically do what they wanted to push the CPU's beyond spec whilst making it seem like it was the 'deffault' settings... and they did so purely to be able to reclaim ground they were losing to AMD Ryzen CPU's.
But over time... normal silicon degradation has rendered them unstable... So there was no choice but to try and blame board partners for pushing beyond the intel defaults.... because intel classes even using XMP profiles as overclocking... Seriously, a few years back they tried denying warranty claims by asking consumers making claims if they used XMP... and people who said yes, had their warranty denied. Loads of people advising consumers to just say 'what's that' if asked.
So now... the avg CPU that has to be default to the 'real' defaults instead of the overclocked defaults... has lost... what... about 10% of the performance they thought they had running it at those previous defaults?
My days of overclocking are long gone... the days of getting 30-40% more out of a CPU have been dead for a decade or more. My old T-Bird 1.2Ghz that ran around 1.57Ghz, or my Phenom II 955 3.2ghz that ran at 3.8ghz for years. Or evn the old AMD Barton core 2500 that a simple FSB bump from 166mhz to 200mhz turned it into a 3200 CPU
OK... I'm old and I miss those days of tweaking and fiddling with settings to get a few more frames... These days CPU's are so on the cutting edge that getting more out of them isn't worth it.
I spend more of my time underclocking and undervolting my stuff now.
Pushing the envelope
My son was experiencing issues with his (gaming) PC recently and it was interesting watching all the chatter back and forth between support, various forums and so on. I, in my true identity as a "person who spent quite a bit of a lifetime designing logic" just dismissed the problem as "you need to turn the clock speed down a bit" but in my other identity as a parent I knew nothing and so needed to be ignored.
Needless to say, after a couple of weeks messing around the solution was triumphantly announced -- clock speed needed adjusting.
Its really difficult to explain to the masses (!) the statistical nature of logic, how you're not working with gates and logic levels but something that's much more organic (analog?) when you run it near the edge. Normally everyone keeps timing windows conservative so users don't experience weirdness but in the world of competitive motherboards there's a tendency to turn the wick up until things break and then just back it off a bit. It "kinda, sorta" works but isn't really a viable strategy. Things won't necessarily just break but there will be strange and random system errors that will point the finger at all matter of irrelevant components (like the son's failure manifest itself as a memory bank problem).