lkoop@pnet01.cts.com (Lamonte Koop) (11/20/90)
I'm hoping perhaps someone can come up with a few ideas about a strange
oddity that I've come across with a program of mine and AmigaOS2.0.
I've written a program (and put it out generally) known as Amiga Intuition
Based Benchmarks (AIBB). The program performs 6 benchmark tests (Writepixel,
Sieve, Savage, Sort, Matrix, and Dhrystone) with an Intuition front end
(click a test gadget and go). The problem is this: Running the tests under
AmigaOS 1.3 is just fine...results come up generally as expected. AmigaOS 2.0
is a completely different story. The tests are running _slower_! And not
just a bit slower, but a great deal so. At first, I thought it was a strange
quirk with the A3000 doing this, but 'tis not the case. The same thing shows
up on an A2000 with an A2620 card running a current 2.01. What I want to know
is why???? Something is quite odd here, to be sure. The tests are nothing
special, and I see no reason why 2.0 should be causing such strange results.
Anyone have a simple 'well, this is what's up...' answer?
regards,
LaMonte
UUCP: {hplabs!hp-sdd ucsd nosc}!crash!pnet01!lkoop
ARPA: crash!pnet01!lkoop@nosc.mil
INET: lkoop@pnet01.cts.comjesup@cbmvax.commodore.com (Randell Jesup) (11/21/90)
In article <5748@crash.cts.com> lkoop@pnet01.cts.com (Lamonte Koop) writes: >(click a test gadget and go). The problem is this: Running the tests under >AmigaOS 1.3 is just fine...results come up generally as expected. AmigaOS 2.0 >is a completely different story. The tests are running _slower_! And not >just a bit slower, but a great deal so. At first, I thought it was a strange There are a number of possibilities. First, background processes (such as commodities like screenblankers) may be eating some CPU time. Second, some calls might not be quite as fast due to relaxing certain limits on them (like the longest line that can be drawn: was ~1024, now it's much larger). A good way to remove most extraneous variables is to turn off the startup-sequence (hold both mouse buttons down, and disable it when the BootMenu comes up). Then run your benchmark. -- Randell Jesup, Keeper of AmigaDos, Commodore Engineering. {uunet|rutgers}!cbmvax!jesup, jesup@cbmvax.commodore.com BIX: rjesup Thus spake the Master Ninjei: "If your application does not run correctly, do not blame the operating system." (From "The Zen of Programming") ;-)
es1@cunixb.cc.columbia.edu (Ethan Solomita) (11/21/90)
In article <15987@cbmvax.commodore.com> jesup@cbmvax.commodore.com (Randell Jesup) writes: >In article <5748@crash.cts.com> lkoop@pnet01.cts.com (Lamonte Koop) writes: >>(click a test gadget and go). The problem is this: Running the tests under >>AmigaOS 1.3 is just fine...results come up generally as expected. AmigaOS 2.0 >>is a completely different story. The tests are running _slower_! And not >>just a bit slower, but a great deal so. At first, I thought it was a strange > > There are a number of possibilities. First, background processes (such >as commodities like screenblankers) may be eating some CPU time. Second, some >calls might not be quite as fast due to relaxing certain limits on them (like >the longest line that can be drawn: was ~1024, now it's much larger). > > A good way to remove most extraneous variables is to turn off the >startup-sequence (hold both mouse buttons down, and disable it when the >BootMenu comes up). Then run your benchmark. > I ran dhrystones using Forbid(), so other tasks should be irrelevant. On my 3000 I got a difference of 120 dhrystones/sec between 1.3 and 2.0. >-- >Randell Jesup, Keeper of AmigaDos, Commodore Engineering. >{uunet|rutgers}!cbmvax!jesup, jesup@cbmvax.commodore.com BIX: rjesup >Thus spake the Master Ninjei: "If your application does not run correctly, >do not blame the operating system." (From "The Zen of Programming") ;-) -- Ethan Woody Allen on Los Angeles: "I mean, who would want to live in a place where the only cultural advantage is that you can turn right on a red light?"
jesup@cbmvax.commodore.com (Randell Jesup) (11/21/90)
In article <1990Nov20.233747.16565@cunixf.cc.columbia.edu> es1@cunixb.cc.columbia.edu (Ethan Solomita) writes: > I ran dhrystones using Forbid(), so other tasks should be >irrelevant. On my 3000 I got a difference of 120 dhrystones/sec >between 1.3 and 2.0. Not neccesarily. It depends on what interrupt services they register with (for example, vblank, or timers, etc.) Also, for such a small difference (relative to timing quantums), it could just be a more complex vblank handler (which is what handles hedley screens, etc - even if not active, it has to check if they are). Video setups might be different - the amount of overscan, screen position, etc, etc, etc. Timer device could be returning more accurate values under 2.0, or maybe it has a bit more overhead. If you want to really test JUST the processor, turn off interrupts, turn off sprites and the screen, and use a CIA timer directly (since timer device can affect you). -- Randell Jesup, Keeper of AmigaDos, Commodore Engineering. {uunet|rutgers}!cbmvax!jesup, jesup@cbmvax.commodore.com BIX: rjesup Thus spake the Master Ninjei: "If your application does not run correctly, do not blame the operating system." (From "The Zen of Programming") ;-)