Reply 40 of 84, by kjliew
gaffa2002 wrote on 2021-06-15, 13:37:
You know, there is a middle ground between being awfully nice and being a jerk. Your negative feedback towards PCEm could be given in so many better ways if you wanted to.
Plus, what is the point of me showing proof that PCEm runs great on M1? PCEm can be slow AND you can have horrible communication skills, one thing has no relation to the other.
Wouldn't that also due to one's conspiracy theorized delusions for being nice or jerk?
I am fine with your quote and you just proved that it can be done in just 1 minute. I hope the Moderator would do the same since he once said he could do it, too.
It is OK if you do not feel the need to prove anything, you don't have to do it. It is your choice.
I am more than happy to share my technical assessments on how to make PCem great on M1. You don't have to take my words or even believe in what I said. Anyone who has the motivation to revive the project can take note. The M1 is a monster on ST performance and it rarely thermal throttled. If unfairly compared its performance to desktop-class CPUs, then Core i9-9900K or Ryzen 9 5950X, it couldn't win but not very far. For laptops, nothing beats the M1 yet, because almost all Intel/AMD x86 CPUs would thermal throttled. Giving the benefit of doubt to PCem that its ARM64 recompiler is as good as QEMU TCG and with QEMU losing its advantage in virtualization, M1 will be a great CPU for PCem. Unfortunately, if it still insists on software rendered 3D acceleration, then it will never be as good as QEMU at GPU dominant workload, such as 3DMark2000. That is my take.
If someone claimed, just an example, that PCem on Rosetta 2 translation on M1 could beat QEMU native performance, then I would just laugh. For me, it is a joke and I said it. How wrong could that be? Fine, anyone who does not think this is a joke, then just prove it. That's the best way to make anyone swallow one's own joke and be shut up once and for all. It is a fair game, isn't it? Did Intel ever laugh when Apple claimed M1 Rosetta 2 would offer equivalent performance to Intel Macs?
Allow me to quote from PCem forums that I can no longer provide the link as evidence since they are now closed, but anyone frequently visited there could testify if I was making it up. Perhaps if anyone decided to reopen the forums or keep them as read-only, it won't matter to me if it remained or got deleted. "... Any pass-through/virtualization evangelism would be massively missing the point and often introduce variables of stability and bloat that won't hold in the long term (TM) ..." I cordially made up the (TM) in the end 😁.
If this was their joke, then I am fine with it.