PASM question about declaring variables...

Chris_D

Hi guys,

I am still working on my first PASM program and am doing pretty good (had it actually working but needs much improvement).· The one thing that is causing me confusion (as I copied the method from examples) is how variables are declared within a PASM program.· THere are two ways it is done...

Variable_1····· LONG··· Preset value

or

Variable_2···· RES····· 1


I can see that the first method allows me to preset the value and method two does not.· However, I also seem to recall that there is something more "complex" going on which dictates when to use which method.· Can someone clarify this for me?

Thanks

Chris
·

BradC

As you correctly surmised, the former allows you to pre-allocate a value. This requires that variable actually consume a long of hub ram. The RES method allocates a COG variable, but it consumes no space in the cog. It's value is effectively undefined when the cog is loaded, so it needs to be properly initialised in the PASM program.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
lt's not particularly silly, is it?

jazzed

Also the RES variables should only be declared at the end of your COG code DAT section.

photomankc

Would that be because the loaded COG instructions would be laid over the variable space otherwise?

Chris_D

Brad,

Being that the first method consumes a variable of HUB memory, does that mean that HUB memory is accessed everytime that variable is accessed?· If so, that would me the time to access it would be much longer than an undefined variable.

Chris

BradC said...
As you correctly surmised, the former allows you to pre-allocate a value. This requires that variable actually consume a long of hub ram. The RES method allocates a COG variable, but it consumes no space in the cog. It's value is effectively undefined when the cog is loaded, so it needs to be properly initialised in the PASM program.

BradC

Chris_D said...
Brad,

Being that the first method consumes a variable of HUB memory, does that mean that HUB memory is accessed everytime that variable is accessed? If so, that would me the time to access it would be much longer than an undefined variable.

No. When you load the cog the contents are copied from HUB ram to the COG directly. From then on, every access to that variable from inside the cog is local, just the same as the RES variables.

That is unless you manually access the HUB ram incarnation of it using rdlong/wrlong.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
lt's not particularly silly, is it?

jazzed

photomankc said...
Would that be because the loaded COG instructions would be laid over the variable space otherwise?

Yes. It's almost like using long without an initialization value.

Chris_D

Thanks guys.

Chris