ECSE436 Tutorial Assembly and Linear Assembly Evgeny Kirshin, 05/10/2011 Outline • • • • • Introduction to Assembly Linear Assembly AMR Reading Demo: simple sine-wave generator E. Kirshin TMS320 Assemby Language E. Kirshin C6000 ISA E. Kirshin Instruction Packing E. Kirshin Sample Instructions E. Kirshin Sample Instruction E. Kirshin Instruction List E. Kirshin Instruction List E. Kirshin C program calling and ASM function E. Kirshin C program calling and ASM function -Values passed to the assembly functions using registers : A4,B4,A6,B6 and so on -Only registers A1,A2,B0,B1,B2 can be used as conditional registers -B3 contain the return address -A4 is the return value -Need to take into account the NOP E. Kirshin Linear Assembly • To effectively program a DSP using assembly language, you need to do the scheduling by hand! • Need to account for the number of clock cycles each functional unit takes, etc… • Difficult, so TI has linear assembly – you don’t have to schedule it, the compiler does it for you – can use CPU resources without worrying about scheduling, register allocation, etc… E. Kirshin What is Linear Assembly ? TI Linear Assembly Efficiency Ease of use C code Assembly E. Kirshin What is Linear Assembly ? It is a cross between C and Assembly Enables you to write assembly-like programs It lets you : • use symbolic names • forget pipeline issues • ignore putting NOPs, parallel bars, functional units, register names • more efficiently use CPU resources than C (e.g., hardware circular buffers) E. Kirshin File extensions File extension for c is .c File extension for linear assembly is .sa File extension for low-level assembly is .asm E. Kirshin Example Dot Product From Chassaing //DOTPclasm.c short dotp4clasmfunc(short *a, short *b, short ncount); #include <stdio.h> #define count 4 short x[count] = {0,1,2,3}; short y[count] = {100, -20, 30, -20}; volatile int result = 0; main() { } Function written in LASM result = dotp4clasmfunc(x,y,count); printf("result = %d decimal \n", result); E. Kirshin Example Dot Product From Chassaing The LASM code _dotp4clasmfunc: loop: [count] .def .cproc .reg zero ldh ldh mpy add sub b .return .endproc _dotp4clasmfunc ap, bp, count a, b, prod, sum sum *ap++,a *bp++,b a,b,prod prod,sum,sum count,1,count loop sum E. Kirshin Example Dot Product From Chassaing Define and ASM function called from C .def _dotp4clasmfunc Name of the function Start a section of Linear assembly _dotp4clasmfunc: .cproc ap, bp, count Arguments of the function Set up your variables (similar to C) .reg a, b, prod, sum E. Kirshin Example Dot Product From Chassaing Initialize your sum to zero zero sum Load the value from memory pointed by ap, copy it to register a and increment the pointer by 1 ldh *ap++,a ldh stands for load half word (a short in C) Do the same thing with b ldh *bp++,b E. Kirshin Example Dot Product From Chassaing Multiply a and b and write the result to prod mpy a,b,prod Add sum and prod and write the result to sum add prod,sum,sum Decrement count by 1 sub count,1,count E. Kirshin Example Dot Product From Chassaing If count is different than 0, branch to loop [count] b loop Return sum as the output of your function (exactly Like the return command in C) .return sum End linear assembly function .endproc E. Kirshin Circular Addressing E. Kirshin AMR Addressing mode register Read Chassaing at pages : 82-83 E. Kirshin AMR Addressing mode register E. Kirshin AMR Addressing mode register How to modify the AMR ? MVKL MVKH MVC .S2 .S2 .S2 0x0004,B2 0x0005,B2 B2,AMR MVC is the only function which can modify the content of a control register Be sure to reset the AMR at the end of your function Also, do the same inside your interrupt routine, check the book by Chassaing to know how to modify the AMR from C There is a good description on how AMR works in SPRA645, see further E. Kirshin References: TI documents • TMS320C6000 Optimizing Compiler User's Guide (SPRU187). Chapter 4 contains the description of the directives (.trip, .cprog, .reg, etc.) • TMS320C67x/C67x+ DSP CPU and Instruction Set Reference Guide (SPRU733). Contains the description of the instruction set (add, mpy, sub, mv, etc.) • TMS320C6000 Assembly Language Tools (SPRU186V). • Circular Buffering on TMS320C6000 (SPRA645). Nice description of hardware circular buffering and the AMR register E. Kirshin References: books • Chaissaing, pages : 87-88 and 112-115 • Kuo and Gan, pages : 243-245 Additional book: S.A.Tretter. Communication System Design Using DSP Algorithms. SpringerLink, 2008 Contains good description of the AIC, McBSP, interrupts and sine-wave generation examples. Available online in PDF (from McGill). E. Kirshin Questions ? E. Kirshin