PCEM seems to faster than real Hardware?

Support and general discussion.
Post Reply
-Scallywag-
Posts: 3
Joined: Thu 10 Nov, 2016 4:14 pm

PCEM seems to faster than real Hardware?

Post by -Scallywag- » Thu 17 Nov, 2016 10:10 pm

Hi.

I was playing around with different configurations and run a few benchmarks.

In Norton Systeminfo 6 and Landmark 2.0 PCEM seems to be much faster than real hardware.

For Comparsion:

Landmark:
PCEM Award 286/8 configuraition shows a 8,240Mhz CPU as fast as a 14Mhz AT (286)

Noton SI 6:
PCEM IBM XT 8088/4,77 vs 8088/4,77 = 1,6 times faster (should be 1,0x)
Award 286/8 vs vs 8088/4,77 = 8,6 times faster (should be 4,4x)

MIPS 1.2:
PCEM Award 286/8 configuraition is 1,12x faster as a real 286/8 (this benchmark shows only little differences)

Is this a problem with my computer/setup or isnt PCEM not that accurate?

Thanks.

Mario

User avatar
SarahWalker
Site Admin
Posts: 1541
Joined: Thu 24 Apr, 2014 4:18 pm

Re: PCEM seems to faster than real Hardware?

Post by SarahWalker » Fri 18 Nov, 2016 7:45 pm

It's not that accurate. 286 and 386 are particularly bad for this.

-Scallywag-
Posts: 3
Joined: Thu 10 Nov, 2016 4:14 pm

Re: PCEM seems to faster than real Hardware?

Post by -Scallywag- » Fri 18 Nov, 2016 9:35 pm

Is it possible to gain nearly 100% accuracy? Maybe in upcomming realeases?

I think the main reason for using pcem over other emulators is the emulation of "machines"´. So accuracy is very important to emulate exact working machines.
Is it a main goal to achieving this?

User avatar
SarahWalker
Site Admin
Posts: 1541
Joined: Thu 24 Apr, 2014 4:18 pm

Re: PCEM seems to faster than real Hardware?

Post by SarahWalker » Fri 18 Nov, 2016 10:06 pm

It should be possible to get closer in future releases. 100% accuracy probably won't happen though.

User avatar
resle
Posts: 50
Joined: Mon 23 Feb, 2015 8:00 am

Re: PCEM seems to faster than real Hardware?

Post by resle » Sat 19 Nov, 2016 3:54 am

Frankly I seem to recall that all machines I had before the Pentium line, having highly fluctuating performances based on... no one knew what.

Benchmarking software from those times, also provided sometimes wildly different results on the same exact machine. I even remember clearly when our school ordered a batch of NEC 80286 PCs and a bunch of them were performing much worse than other from the very first day, with only Lotus 1-2-3 installed (Dos aside).

Would there even be a proper sample to compare against for the sake of the "100% accuracy"? I doubt.

Post Reply