home *** CD-ROM | disk | FTP | other *** search
- > > I have now written down the result from memspeed and I'm not
- > > 100% sure that I trust this program. :-)
- ...
- > > You see, I can get different results from time to time in the
- > > cache. It changes from 500MB/s to 0.0 MB/s. :-)
-
- Perhaps things are happening too fast to give an accurate timing?
-
- The data cache bandwidth should be limited to 4 * 32e6 = 128 Mbyte/s, since
- the '040 can't run any instruction in less than one cycle as far as I know
-
- > I started my own profiling program last night, so I could send up a preview if
- > you
- > want. It's still just a fancy memspeed so don't expect too much. The relative
- > performance figures (shown as %) are just nonsense ATM, but the timing figures
- > are accurate.
-
- :-)
-
- I've been doing a lot of image processing, compiling, assembling, drawing etc
- tests on my non-accelerated Falcon now and will very soon be able to compare
- the timings with an '040 installed. :-)
-
- > It does show various things in MIPS, MB/sec and Megaflops. I haven't
- > quite got round to the SpecINT standards yet....
-
- Why bother with SpecINT when you've got _MIPS_? ;-)
-
- I'll see if I can get the BYTEmark suit compiled.
-
- --
- Chalmers University | Why are these | e-mail: rand@cd.chalmers.se
- of Technology | .signatures | johan@rand.thn.htu.se
- | so hard to do | WWW/ftp: rand.thn.htu.se
- Gothenburg, Sweden | well? | (MGIFv5, QLem, BAD MOOD)
-
-