3 Jan
2013
3 Jan
'13
8:35 p.m.
On 03/01/13 10:05 PM, Jonathan Katz wrote:
On Jan 3, 2013, at 9:52 PM, Jan-Benedict
Glaw<jbglaw at lug-owl.de> wrote:
What ever happened to the original Dhrystone tests? Since those were designed for VAXen
(era) systems, running them on a modern emulator may give some apples to apples
comparisons between the real big iron and our virtual machines. Especially if it's
done with a vintage OS (like BSD 4.3-Something) and its original compilers which probably
don't optimize as well as gcc (although, optimization for VAX in a modern compiler?
That's another can of worms.) But running a circa 1984 benchmark on an OS written in
1986 and compare that to known big iron (The VAX 11/780 was clocked with a Dhrystone of
1757 per Wikipedia) to simh running on one of our desktops.
Dhrystone was never a great benchmark; it was a poor workload mix and
manufacturers famously gamed it, tweaking their runtimes or even their
hardware for it.
This one is better, and figures are published for quite old machines:
http://homepages.cwi.nl/~tromp/c4/fhour.html
It does however take a few megabytes of RAM.
--Toby
It also
was surprisingly fast. No idea how SIMH on a 2.7 GHz Athlon compares
to real VAX metal speed-wise, though.
Some guy started to hack support for Qemu. I don't know the innards of
Qemu, but IIRC he told me that it's driving something like a
I think I know what I'm going to stay up late working on.