main.s Error Code 2

[Jenna Liebe]

SFU ID: 301358537

Github username: jel-cmpt295

Github: Link to github repo (your assignment repos are private so these links are only accessible to TAs and Instructor): https://github.com/CMPT-295-SFU/assignment-3-jel-cmpt295

Line and file: main.s, "Load Matrices/Run Layers 1"

Expected behavior: Output result of m0 * input matmul

Observed behavior: Error code 2 (mismatched dimensions in matmul)

Question: I threw in a bunch of debug code to see what values my registers were storing before the jump to matmul, and I can't figure out what I'm doing wrong that I end up storing values that are all over the place - I have no idea where they're coming from. I made sure to reset the values of the a registers that I changed during my debug before going into matmul, but it still doesn't work. Where am I going wrong?

[ali sedaghati]

Put a screenshot of command and error here. And make sure that you're passing the correct values to a1,a2,a4,a5 when you're calling matmul.

[Arrvindh Shriraman]

Your read matrix is broken

1) the malloc of the row and column pointer have to happen in main not within read_matrix

Otherwise how does main know what the pointer is ? 

You seem to be mallocing the row and col ints within read_matrix which main does not know about.

[Jenna Liebe]

Okay, I managed to fix that problem - thank you! Unfortunately now I have a different error - although I can print out Layers 1 and 2 (matmul with m0 and input, and relu on the result), for some reason I am not able to load m1 using my read_matrix file. Input and m0 are read in and loaded up fine, and I made sure that the filepath I was passing into read_matrix for m1 was correct, but I keep getting the "Venus ran into a simulator error!

Attempting to access uninitialized memory between the stack and heap. Attempting to access '4' bytes at address..." error if I don't comment out anything to do with m1.  I attached a screenshot of my output when m1 read_matrix and load are commented out, and one when they aren't (if that helps).

I'm using read_matrix on m1 in line 128 of main.s, and allocating the space for its parameters in lines 135-139.

It's formatted almost exactly the same as input and m0, so I'm not sure why it isn't working.

[Arrvindh Shriraman]

That just means that the pointer to which you are trying to write is bad.

Check if you are reading the wrong location on the stack for m1's base address

- either you got the sp manipulation wrong

- OR you got the offset wrong

[Jenna Liebe]

I'm pretty sure I'm reading the right location with the right offset - when I print out argv[0], argv[1] and argv[2] it gives me 

./inputs/simple0/bin/inputs/input0.bin

./inputs/simple0/bin/m0.bin

./inputs/simple0/bin/m1.bin

That's right, isn't it? So does that mean my offset and sp manipulations are right? If that's the case, why wouldn't read_matrix work with just the last input (m1 in this case)? 

[Arrvindh Shriraman]

Your read matrix appears broken. 

What is this sequence doing:

lw s5, 0(s3)
	mul s5, s3, s4

s3 is a pointer? 

s4 is a pointer?

s5 is going to be a bad value. You need to lw the pointers before multiplying them