Changes between Version 10 and Version 11 of ClrxChangeLog


Ignore:
Timestamp:
Feb 21, 2018, 4:04:22 PM (7 months ago)
Author:
matszpk
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • ClrxChangeLog

    v10 v11  
    11= Change Log =
     2
     3CLRadeonExtender 0.1.7:
     4
     5* update AmdCL2ABI chapter
     6* fixed kernel arguments sizes in GalliumCompute binary format
     7* add new GPU devices gfx902-gfx905
     8* update device tables for Amd Crimson drivers
     9* small fixes in DynLibrary interface
     10* add relocations to GalliumCompute binary format (for scratch buffer symbols)
     11* make getXXXDisasmInputFromBinaryXX as public interface
     12* speeding up evaluation of simple expressions without symbols
     13* add '.for' and '.while' pseudo-ops ('for' and 'while' loops)
     14* fixed some grammar/typos in CLRX documentation
     15* add GPU device names from ROCm-OpenCL
     16* handle new ROCm binary format with YAML metadatas (assembler and disassembler)
     17* add few pseudo-ops to ROCm handling
     18* add new pseudo-ops to set parameters in ROCm YAML metadata
     19* fixes in GalliumCompute binary generator (for conformant with standards)
     20* add '.reqd_work_group_size' pseudo-op (equivalent of '.cws')
     21* add support for work_group_size_hint and vec_type hint in Amd OpenCL 2.0 binary format
     22* some small bug fixes in ROCm disassembler
     23* updates in README.md and INSTALL files
     24* small sanitizations in DisasmAmd, DisasmAmdCL2 (argument type checking)
     25* change behaviour of '.cws' (.reqd_work_group_size) while setting default values
     26* add calculation of section differences in an expressions (for ROCm handling)
     27* fixed invalid reads (potential segfault) after undefining symbol
     28* fixed old stupid bug: resolve symbol value by using new value (or just if undefined then
     29  do not resolve symbol) instead old unresolved symbol value later when expression
     30  has been evaluated
     31* Add GOT table handling in ROCm binary format
     32* add new option '--newROCmBinFormat'
     33* add untested support for ROCm in CLHelper and VectorAdd sample
     34* add support for multiple OpenCL platforms in CLHelper and samples
     35* allow te call_convetion to 0xffffffff in AMDHSA config
     36* handle special cases with relatives while evaluating binary/logical operators
     37* small fixes in CLRX documentation and Unix manuals
     38* developing unfinished AsmRegAlloc
     39* add a missing access qualifier to images 'read_write' for AMD OpenCL 2.0
     40
    241
    342CLRadeonExtender 0.1.6: