Memory use - how much do you use?

[ScaleRobotics]

I have some code that I have been working on for a PIC12F675. When I try to add another variable to it and then compile, I get - Error: Unable to fit variable y -. I am using the latest pbp, microcode studio, and mpasm v8.15. When I use Mackrackit's LST View program, it shows my successful compile (one less variable) as using about 70 percent of the code space, and 48 percent of the sram space. What would be stopping me from being able to use the other half of my sram? I looked in the 12F675.BAS file, and my memory map is BANK0 $0020, $005F, which is correct in the data sheet. 12F675 is selected in mcs.

Confused....

[mackrackit]

Can you post your code?

[ScaleRobotics]

My code isn't really this, but I get similar results using this example:

myvar1 var byte
myvar2 var byte
myvar3 var byte
myvar4 var byte
myvar5 var byte
myvar6 var byte
myvar7 var byte
myvar8 var byte
myvar9 var byte
myvar10 var byte
myvar11 var byte
myvar12 var byte
myvar13 var byte
myvar14 var byte
myvar15 var byte
myvar16 var byte
myvar17 var byte
myvar18 var byte
myvar19 var byte
myvar20 var byte
myvar21 var byte
myvar22 var byte
myvar23 var byte
myvar24 var byte
myvar25 var byte
myvar26 var byte
myvar27 var byte
myvar28 var byte
myvar29 var byte
myvar30 var byte
myvar31 var byte
myvar32 var byte
myvar33 var byte
myvar34 var byte
myvar35 var byte
myvar36 var byte
myvar37 var byte
myvar38 var byte
myvar39 var byte
myvar40 var byte
myvar41 var byte

I admit this is more than 48.44 percent of the 64 bytes of SRAM. I am not clear how to check SRAM use in any other way than using your program. Perhaps PBP uses the other bytes for itself. But then, your program should account for actual use, since it is using the lst file.

Let me know what you think.

Thanks,

Walter

[mackrackit]

I played with this some last night, did some reading and could not find a solution.
Maybe Darrel or Bruce will chime in...

[Darrel Taylor]

I played with this some last night, did some reading and could not find a solution.
Maybe Darrel or Bruce will chime in...

DING!

PBP's system vars use a minimum of 24 bytes.
Complex formulas can increase that, but it appears you don't have any of those yet.

So you create 40 additional byte variables and it's full. (24 + 40 = 64)

As for "LST View" ...

It apparently just looks for the last occurrence of "RAM_START +" and uses that address to calculate how much RAM was used.

That address is already referenced to RAM_START, but in the program, this line subtracts the start address again, which leaves the count too low.

ramU = ramF - ramS

Add 20h (32) to the 31 bytes reported and you get 63, the last address in RAM.

I think that line should be ...

    ramU = ramF + 1

However, RAM assignments are not always contiguous. You could have a single variable in BANK3 (chip other than 675) and it'll show almost all the RAM has been used, when there's only a few bytes being used. Or the last address may be the beginning of an array, which it won't count either.

It's a difficult task to count RAM bytes used.
<br>

[mackrackit]

Thanks Darrel.

I saw where LST View was giving some "funny" results last night also. Now I know why. Time to do a "bug" fix. If possible...

[Darrel Taylor]

Dave,

Just a little more info for updating LST View.
It calculates the Total RAM by taking (RAM_END - RAM_START + 1)
As an example this is from a 16F877 ...

  00000020            00012 RAM_START                       EQU     00020h
  000001EF            00013 RAM_END                         EQU     001EFh
  00000004            00014 RAM_BANKS                       EQU     00004h
  00000020            00015 BANK0_START                     EQU     00020h
  0000007F            00016 BANK0_END                       EQU     0007Fh
  000000A0            00017 BANK1_START                     EQU     000A0h
  000000EF            00018 BANK1_END                       EQU     000EFh
  00000110            00019 BANK2_START                     EQU     00110h
  0000016F            00020 BANK2_END                       EQU     0016Fh
  00000190            00021 BANK3_START                     EQU     00190h
  000001EF            00022 BANK3_END                       EQU     001EFh

(RAM_END - RAM_START + 1) = 1EFh - 20h + 1 = 464 bytes
But the 16F877 only has 368 bytes.

The extra count is from the SFR's that are at the top of each bank.

GP RAM in each bank doesn't start at the same place on every 16F, so I think you'll need to find the amount in each bank and add them together to get the total RAM.

(BANK0_END - BANK0_START + 1) + (BANK1_END - BANK1_START + 1) + (BANK2_END - BANK2_START + 1) + (BANK3_END - BANK3_START + 1)
(  0007Fh  -   00020h    + 1) + (  000EFh  -   000A0h    + 1) + (  0016Fh  -   00110h    + 1) + (  001EFh  -   00190h    + 1)
(            60h            ) + (            50h            ) + (            60h            ) + (            60h            )
= 170h = 368

Then it's the same for the Used RAM. You need to subtract the SFR space from the variables address, or somehow reference it to the BANKx_START of the bank that it's in to find out how much has been used.

hth,