AMD confirms FMA3 workloads crash Ryzen, promises microcode fix
PC Perspective offers some background info and speculates the hard lock may be caused because these type of workloads lean heavily upon the FPU and may require more power than the chip's Control Fabric can deliver. Some overclocked Ryzen chips with much higher voltages can finish the test successfully so this seems to confirm that it's indeed a power issue:
The interesting thing about this FMA3 finding is that it is seen to not be present in some overclocked Ryzen chips. To me this indicates that it could be a power delivery issue with the chip. A particular workload that heavily leans upon the FPU could require more power than the chip’s Control Fabric can deliver, therefore causing a hard lock. Several tested overclocked chips with much more power being pushed to them seems as though enough power is being applied to the specific area of the chip to allow the operation to be completed successfully.AMD acknowledges the issue and says it will provide a CPU microcode update, this will be distributed to the motherboard makers and they have to implement it via a motherboard firmware update. It's unknown how this errata will be fixed but PC Perspective speculates AMD will tweak Ryzen's voltage control system to push more power to the affected areas of the Ryzen cores. The site speculates this may result in a burst speed hit to keep Ryzen's TDP within its specified envelope.
http://forum.hwbot.org/show... "The issue with Flops was found and fixed in the beginning of february. The current µcode version dates to 01/27/2017, so the fix is obviously not included yet (due to the time required for validation). Flops is only affected when the SMT is enabled, so disabling the SMT can be used as a temporary work-around (until the actual fix arrives)."
Reply