4 ms·
Curious Undisclosed Skylake Bug
- saagarjha 4y agoWonder if breaking the dependency would be a workaround?
- moonchild 4y agoYes; reuse the source, or use something it depends upon, or zero the destination ahead of time.
- j16sdiz 4y agoDoes false dependency count as bug?
- vardump 4y agoKinda. Not like it'd produce a wrong result, but you could say it's a bug (presumably) in the out-of-order scheduler.
- kevincox 4y agoExactly: It's a performance bug not a correctness bug.
- DannyBee 4y agoIntel will list them as errata if you beat them up enough. They rarely fix them. POPCNT false dependency on sky lake was never fixed, we just ended up with compilers inserting useless xor's before popcnt to break dependencies.
- DannyBee 4y agoThis is not the first one. Skylake (and a few others) have a "well-known" false dependency bug in popcnt. It's listed in the skylake erratum as "POPCNT Instruction May Take Longer to Execute than Expected".
- gjsman-1000 4y agoSkylake, according to rumors, is absolutely infested with bugs. True or not, an ex-Intel engineer said the number of bugs actually tipped Apple over the edge [1], and I personally suspect that Windows 11 requiring Intel 8th Gen could, partially, be a move against having to maintain Skylake compatibility (or at least a move against claiming that it is "supported"). See also macOS 13 cutting off Skylake support for a minimum of Kaby Lake (2017 and newer) [3]. See also Windows 7 announcing in 2017 they would not be adding Skylake support because it was too much work [4]. [1] https://www.pcgamer.com/intel-skylake-why-apple-left/ https://www.pcgamer.com/intel-skylake-why-apple-left/ [2] https://www.google.com/search?client=safari&rls=en&q=skylake+full+of+bugs&ie=UTF-8&oe=UTF-8#ip=1 https://www.google.com/search?client=safari&rls=en&q=skylake... [3] https://support.apple.com/en-us/HT213264 https://support.apple.com/en-us/HT213264 [4] https://www.computerworld.com/article/3025399/microsoft-wont-support-most-skylake-pcs-without-windows-10.html https://www.computerworld.com/article/3025399/microsoft-wont...
- ls65536 4y ago> On a system which has the bug, it should print two large numbers and two small numbers; on a system which does not have the bug, it should print one large number and three small numbers. I built and ran the provided code on an AMD Zen 4 CPU, and it printed "two large numbers and two small numbers", so could this bug also be present in other microarchitectures implementing AVX-512 other than Skylake(-X)?
- moonchild 4y agoInteresting; certainly sounds plausible. What happens if you replace '#define COMP "vcompressps"' with '#define COMP "vexpandps"'?
- ls65536 4y agoWith that change, no meaningful difference in the results from what I can tell: the program still prints two large integers (around 420M) and two smaller ones (around 85M).
- mrlonglong 4y agoThat's interesting. Which Intel gen is Skylake?
- colejohnson66 4y ago6000 series, but the core architecture itself was used through the 10000 series