On Tue, 25 Jan 2000, Philippe wrote: > At 16:33 16/01/00 -0600, you wrote: > >It's not quite as good as UMPS yet, but it's about three times faster > > Three time faster ? Really ? how did you make your measurement ? > > I would be interested to have some comparison table that include > all the test process ... > > Running a PIC without anything connected on UMPS can be as fast as real pic > at 4Mhz on a K6-400. BUT when you put breakpoint, when you connect > resource, when you open windows that need graphical drawing its > slow considerably the simulation. This is normal and Linux has nothing > to gain with its windows object system than MS-Windows itself. Okay, maybe it's 4 times faster :). For a basic simulation (with no stimuli), gpsim runs as fast as 20Mhz pic when simulated on a 450Mhz pentium II. But you're right, as soon as the gui has to be serviced, there's a noticable impact on the simulation speed. It's not a Linux/MS thing per se - and I didn't mean to imply that it was. > > >and infinitely cheaper! > > UMPS is not a freeware and is NOT dedicated to hobbyist, it is dedicated to > professionnal usage. I did not know people that work for free. I know > nobody that has bought a Ferrari for just $USD1000. You need something cheap: > take a look at freeware or shareware ... You need a $USD100 compiler, browse > the WEB you will find some, but I did not know any professional that will take > a risk to use unstable product. The works needed to setup > and solve potential problem on such product will cost much more than > buying a reasonable priced product with some support garantee. I didn't mean to imply UMPS is or should be freeware. I simply state the facts: gpsim is faster, but UMPS is easier to use. Furthermore, UMPS supports many more features than gpsim. But the implication that gpsim is unstable is false. As a professional PIC developer, I spend most of time away from the computer to just think. When I get an idea fully thought out, I code. Most of the stuff I've developed is part of a system and consequently is difficult to simulate. So I don't even bother. The things I tend to use the simulator for are repetitive algorithm development. For example, to develop the sine algorithm for the 18cxxx family, I used gpsim to rapidly sequence through all possible inputs. What would take minutes in mplab takes only milliseconds in gpsim and UMPS. I'm sure other professionals work similarly. gpsim is intended primarily for the hoobyist. It's a "grass roots" kind of program that starting to get a little momentum. I regularly receive patches that both fix bugs and enhance the features. Many of these patches are received from hobbyists, but some are received from professionals as well. That's one of the appeals of open-sourced software to me. So to re-iterate: UMPS is not an inferior product. It is better than gpsim in every way except for three: 1) It's slower, 2) It costs more, and 3) It only runs under one OS. For me, those just so happen to be three most important reasons to use gpsim. But that's just me. I provide a link on my web page to UMPS for those who think otherwise. Perhaps there should be a link on UMPS' web page to gpsim? Here:
UMPS is the best and fastest microcontroller simulator under MS windows. If you need a faster simulator that runs under Linux and is free then perhaps you should consider gpsim . However, it only simulates the PIC microcontroller family and from what I hear more difficult to use.
Scott