xu_zh

Close as V3 bpu is re-written

I had originally planned to use hardwareError, but since `object ExceptionNO` didn't have that encoding at the time, and access fault was acceptable according to ISA, I didn't implement it...

I've drafted a hardwareError implementation in Frontend V3, please check out: - #4770 I guess you can merge this PR if you planned to.

> If this pr is merged into kmhv2(master) is there any additional work required and new bugs that may be introduced? Sorry I think I need to clarify that by...

SPEC 06 (0.3 coverage) tests show overall performance essentially unchanged (~0.03% increase), with ±1% fluctuations at individual test points (GemsFDTD -1.11%; zeusmp +0.74%). I guess this is acceptable.

> > I don't think it's a good idea to limit includePaths in .scalafmt.conf, it's a project level configuration file and the IDE uses it too. > > If you...

Thanks for reporting this issue, we do have a bug here, but it's probably not quite what you're thinking of: In the current version, PrefetchPipe is responsible for querying ITLB,...

Well, after this modification, `icache_bubble_s0_wayLookup` looks better, but topdown still looks weird, maybe has other bugs. Seems we have never maintained topdown since #2085, any idea? @eastonman CoreMark 2 iter...

> there is nothing else but ITLB miss would cause PrefetchPipe stall I got it wrong, redirect / cache refill will cause `s0_valid && !fromWayLookup.valid` to be `true` too. And,...

We're re-writing perf counters on kunminghu-v3 so this should not be a bug there. And as said before, we won't fix this on master, if you need this fix, please...