Asm4GCN - GCN Assembler for AMD GPUs

an assembler/compiler for AMD’s GCN (Generation Core Next Architecture) Assembly Language

See the Article on CodeProject at http://www.codeproject.com/Articles/872477/Assembler-for-AMD-s-GCN-GPU for more details.

###Change History

• Feb 16 2015

  • Note: Initial Public Release

• Mar 1 2015

  • Added: ren command - A rename command has been added. This allows a variable to be renamed as its use changes.
  • Improved: enlarged the auto-complete box - it now fits the code a little better.
  • Improved: Cleaned up example code.
  • Improved: Syntax Highlighting - it now highlights, labels, registers, and defines. It also highlights matching words.
  • Improved: Updated CodeProject article
  • Fix: Variables would always use register 0
  • Fix: Removed single __asm4GCN block limitation - There can now be many __asm4GCN kernels in one program.
  • Change: Parameter names removed - Since the Parameter names are not used having them there could be confusing. Function headers are now in the form: __asm4GCN myAsmAddFunc (float*,float*){...}
  • Change: merged #ref command into normal variable declarations. Since the #ref command is almost identical to normal variable declarations except that it specifies a register it is best to combine these. It is cleaner and less confusing. Instead of the format for a ref being #ref s8u myVar s[2:3] is now just s8u myVar s[2:3].
  • Removed: Auto compile skip function. This function would skip a re-compile if there were no changes in the code windows. It was removed because it added complexity in the code and there was hardly any performance benefit since the compile process is so fast anyway.

• April 22 2015

  • Posted on GitHub

• July 10 2015

  • Fix: Fixed bug in SAPP encoding. Jumps were not working properly.
  • Fix: Auto-complete was not working on the GCN tab - it has been fixed.

• July 18 2015 - Switched Open CL Wrapper Library and More

  • Change: Switched the OpenCL wrapper to use NOpenCL by Tunnel Vision Laboratories. This is an awesome well-written wrapper.
  • Added: Indexing on variables. (i.e. myVar[1] would access the 2nd register in myVar)
  • Added: VINTRP encoded instructions
  • Updated: Updated article

• August 2 2015 - Lots of changes

• New: Variables automatically free themselves on the last line they were used.

• New: Jumps can now be used in front of any statement and not just by line.

• Refactor: Added variables class

• Refactor: Moved variable functionality into Variable class

• Refactor: reworked how jump functionality

• Refactor: everything used to be done in one-pass and now it is in two-passes

  1. Pass 1 - read in all statements and record positions of vars
  2. Process automatic variable freeing
  3. Process register assignments for variables
  4. Pass 2 - convert statements to bin

• Removed: Removed the 'ren' function as it resulted in ugly code. The same can be accomplished by declaring vars with specified registers.

• Broken: variable indexing is broken ( I am fixing this next)

• August 9 2015 - Lots of changes

  • New: multiple labels intermixed with statements can now be added to a single line
  • New: Freed Variable registers can now be re-used in the same instruction.
  • New: Inline variable decorations (e.g. v_mov_b32 v4u myNewVar, anyVar )
  • New: lines ending with '/' will append the following line. #defines and statements can be split across lines.
  • New: Cleaned up initial code and added a #define(...) to easily view any S or V variable.
  • New: Added Ctrl-Y as a redo operator for Visual Studio users.
  • New: Additional variable warning checking (like when it is never used or used only once)
  • New: When declaring variables an existing variable, with an option index, as the register you want
    to re-use.
  • Change: Spaces can no longer be used to separate operands. Only commas can be used. This does not apply to trailing instruction option parameters.
  • Change: Defines are now processed before labels.
  • Change: Defines are now processed in reverse order so defines replacements can contain previous #define replacements.
  • Fix: multi-register variables were not always aligned properly.
  • Fix: Fixed issue with variable indexes
  • Fix: syntax color highlighting issue on GCN tab where not all text was always highlight properly.

• August 16 2015 - Added Unit Tests / Fixed Encoder Issues

  • New: Added unit tests that double as examples.
  • New: Added example: A fast wavefront sum reduction using 18 instructions and no shared memory.
  • Fix: Did some minor fixes and adjustments in the instruction encoder.

• Sept 5 2015 - Performance Improvements / Fixed Unit Tests

  • OpenClWithGCN Performance: Updated RegEx in ExtractAsm4GCNBlocks. Comments are now stripped out first. In addition, the RegEx has been brought outside of the function so it does not have to be re-created each time.
  • OpenClWithGCN Performance: The last dummy source code is now cached so we can detect if there are changes. If there are no changes then we do not need to recompile it.
  • OpenClWithGCN Performance: Overall, re-running a modified a __asm4GCN is now about 10x to 50x faster! First time calls are roughly 10% faster.
  • OpenClWithGCN update: updated driver version warning for new driver versions
  • Tests Update: Corrected issue in Validate results in unit tests.
  • Tests Update: Added a new test to OpenClWithGCN reuse.

• September 15 2015

  • Improvement: (OpenClWithGCN Performance) Updated RegEx in ExtractAsm4GCNBlocks. Comments are now stripped out first. In addition, the RegEx has been brought outside of the function so it does not have to be re-created each time.
  • Improvement: (OpenClWithGCN Performance) The last dummy source code is now cached so we can detect if there are changes. If there are no changes then we do not need to recompile it.
  • Improvement: (OpenClWithGCN Performance) Overall, re-running a modified a __asm4GCN is now about 10x to 50x faster! First time calls are roughly 10% faster.
  • Updated: OpenClWithGCN update: updated driver version warning for new driver versions
  • Fix: Tests Update: Corrected issue in Validate results in unit tests.
  • Add: Tests Update: Added a new test to OpenClWithGCN reuse.

• September 21 2015

  • Fix: Fixed bug with dummy kernel. The dummy kernel was not expanding to the size it should so when large kernels were created they would not fit in the dummy kernels binary space.

• October 19 2015

  • Fix: Fixed an bug when there was an AMD and NVidia GPU in the same system and the NVidia GPU came up first

• October 24 2015

  • Refactored: Cleaned up GcnBlock.cs. Separated CleanupComments, processDefines, and processSingleStmt

• November 9 2015

  • New: Added a friendly syntax feature so that asm statements can be entered in a more readable format. Example: "v_add_i32 localSizeIdx, vcc, laneId, vLocalSize" can now be added like "localSizeIdx = laneId + vLocalSize". It is easier to read. It is only supported on +,-,*,>>,<<.
  • Refactor: Pulled out some of the Regular Expressions to their own file. This will be expanded on in the future.

• Jan - April 2016 

  • Update; Added prompt for kernel export.
  • Update: Added driver versions 15.300 and 16.150 to acceptable list; code cleanup.
  • Fix: Asm4GCN would fail if there was more than one AMD GCN GPU. It will now work but Asm4GCN will not work with both GPUs without adjustment
  • Update: Added a support for a newer driver version; add a "vcc" for the friendly converter for "+"
  • Fix: Corrected opcode for DS_WRITE_SRC2_B32=141 and DS_WRITE_SRC2_B64=205 (Thank you to Mateusz Szpakowski for finding that and correcting it.)

###To-Do / Broken

• cannot re-use the same variable name (maybe move these to a list under itself)
• cannot have multiple __asm4GCN kernels
• Performance: limit smart available register search to stop at 16 in size (instead of 128)
• DS type like "ds_swizzle_b32 temp, val, extra, extra offset1:123 offset0:456" only uses two params as well

###Future Wish List

• Correct the issue where the same variable name cannot be reused.
• GCN Generation 3 support
• Additional OpenCL 2.0 support
• Add compatibility for additional GPU driver versions.
• Inline Assembly – This was an original goal but generating a dummy kernel for it was problematic.
• Alternative friendly, easy to read, assembly statements. Instead of something like v_mul_i32_i24 varA, varB, varC maybe have something like varA = VarB * VarC. To do this we would need to know the types of the variable to select the correct instruction. This part is already in place.
• Use variable type information (I,U,F,B) to throw warnings when used with non-matching instructions. i.e. using v4f with v_mul_i32_i24