Intel has confirmed an in-the-wild erratum affecting its latest Skylake processor architecture and causing hangs during certain computationally-intensive operations, with its OEM partners to roll out a hotfix by way of BIOS update sharpish.
Errata in semiconductors are common: by the time the chip is ready for testing, the design is baked in and cannot be changed - meaning anything discovered can only be fixed in the next design iteration. Increased use of simulation has reduced the number of serious flaws that make it into production, but each release of any sufficiently complicated chip brings with it a list of errata that have been noted and need to be worked around in software. For Intel, its most famous was the FDIV bug in which the company's Pentium processors would occasionally return the wrong result in floating-point operations - a flaw which, despite the company's downplaying of its severity, led to a $475 million recall programme.
Intel's Skylake architecture is far removed from the old Pentium, but the company has once again let a breaking bug hit shipping parts. Users of the Great Internet Mersenne Prime Search began reporting problems on the forum back in November last year, claiming that the Prime95 software would crash on certain systems. Investigations began to narrow down the problem, and eventually found that only one processor family was affected: Intel's Skylake. Narrowed down, the problem was reported to Intel.
The company has now confirmed that the flaw spotted by the Mersenne hunters is indeed genuine and tied to changes in the Skylake architecture, but as with the FDIV bug before it is claiming the flaw is far from serious. 'This issue only occurs under certain complex workload conditions, like those that may be encountered when running applications like Prime95,' a company representative confirmed in reply to the support thread. 'In those cases, the processor may hang or cause unpredictable system behaviour. Intel has identified and released a fix and is working with external business partners to get the fix deployed through BIOS.'
By issuing a microcode update via the BIOS, Intel should be able to avoid a product recall - just as it attempted to do by offering software-based workarounds for the original FDIV bug, only caving in and recalling the faulty processors in the face of major consumer and media backlash. When installed in a system with a BIOS featuring the fix, the erratum should be resolved independent of any client software including operating system. Thus far, no manufacturers have come forward to state when they will release BIOS updates including Intel's workaround for the flaw.
Errata in semiconductors are common: by the time the chip is ready for testing, the design is baked in and cannot be changed - meaning anything discovered can only be fixed in the next design iteration. Increased use of simulation has reduced the number of serious flaws that make it into production, but each release of any sufficiently complicated chip brings with it a list of errata that have been noted and need to be worked around in software. For Intel, its most famous was the FDIV bug in which the company's Pentium processors would occasionally return the wrong result in floating-point operations - a flaw which, despite the company's downplaying of its severity, led to a $475 million recall programme.
Intel's Skylake architecture is far removed from the old Pentium, but the company has once again let a breaking bug hit shipping parts. Users of the Great Internet Mersenne Prime Search began reporting problems on the forum back in November last year, claiming that the Prime95 software would crash on certain systems. Investigations began to narrow down the problem, and eventually found that only one processor family was affected: Intel's Skylake. Narrowed down, the problem was reported to Intel.
The company has now confirmed that the flaw spotted by the Mersenne hunters is indeed genuine and tied to changes in the Skylake architecture, but as with the FDIV bug before it is claiming the flaw is far from serious. 'This issue only occurs under certain complex workload conditions, like those that may be encountered when running applications like Prime95,' a company representative confirmed in reply to the support thread. 'In those cases, the processor may hang or cause unpredictable system behaviour. Intel has identified and released a fix and is working with external business partners to get the fix deployed through BIOS.'
By issuing a microcode update via the BIOS, Intel should be able to avoid a product recall - just as it attempted to do by offering software-based workarounds for the original FDIV bug, only caving in and recalling the faulty processors in the face of major consumer and media backlash. When installed in a system with a BIOS featuring the fix, the erratum should be resolved independent of any client software including operating system. Thus far, no manufacturers have come forward to state when they will release BIOS updates including Intel's workaround for the flaw.
MSI MPG Velox 100R Chassis Review
October 14 2021 | 15:04
Want to comment? Please log in.