Lines Matching refs:ir3

1408 - ir3: Fix list corruption in legalize_block()
1409 - ir3: Reduce max const file indirect offset base to 9 bits
1410 - ir3, tu: Add compiler flag for robust UBO behavior
1415 - ir3/parser: Fix oob write with immediates array
1416 - ir3: Improve cat1 modifier disassembly
1417 - ir3: Assemble and disassemble swz/gat/sct
1418 - ir3: Prevent oob writes to inputs/outputs array
1420 - ir3, tu: Cleanup indirect i/o lowering
1425 - ir3/postsched: Fix dependencies for a0.x/p0.x
1426 - ir3/cp: Clone registers for compare-folding optimization
1427 - ir3/sched: Use correct src index
1428 - ir3/postsched: Use correct src index
1429 - ir3/delay: Remove special case for array deps
1430 - ir3/postsched: Fix ir3_postsched_node::delay calculation
1431 - ir3/cp_postsched: Fixup SSA use pointer for direct reads
1432 - ir3: Refactor nir->ir3 block handling
1433 - ir3: Make predecessors an array
1434 - ir3: Rework outputs
1435 - ir3: Don't assume regs[1] exists in ir3_fixup_src_type()
1437 - ir3/cf: Rewrite pass
1438 - ir3: Use round-to-nearest-even for fquantize2f16
1439 - ir3: Call nir_lower_wrmask() again after lowering scratch
1440 - ir3: Only use per-wave pvtmem layout for compute
1441 - ir3: Introduce phi and parallelcopy instructions
1442 - ir3: Add ir3_start_block()
1443 - ir3: Readd support for translating NIR phi nodes
1444 - ir3: Prepare for instructions with multiple destinations
1445 - ir3: Improve register printing for SSA
1446 - ir3: Add ir3_register::array.base
1447 - ir3/delay: Fix full->half and half->full delay
1448 - ir3: Add reg_elems(), reg_elem_size(), and reg_size()
1449 - ir3: Make branch conditions non-SSA
1450 - ir3: Rewrite delay calculation
1451 - ir3/delay: Delete pre-RA repeat handling
1452 - ir3/postsched: Don't use SSA source information
1453 - ir3: Remove unused check_src_cond()
1454 - ir3: Add dominance infrastructure
1455 - ir3: Add pass to lower arrays to SSA
1456 - ir3: Expose occupancy calculation functions
1457 - ir3: Rewrite register allocation
1458 - ir3/ra: Add a validation pass
1459 - ir3: Remove right and left copy prop restrictions
1460 - ir3/sched: Don't schedule collect early
1461 - ir3/sched: Make collects count against tex/sfu limits
1462 - ir3/sched: Consider unused destinations when computing live effect
1463 - ir3: Add simple CSE pass
1464 - ir3: Insert output collects in the main shader
1465 - ir3: Copy propagate immed/const to meta instructions
1466 - ir3: Improve printing of array parallelcopies/phis
1467 - ir3/ra: Fix array parallelcopy confusion
1468 - ir3: Make tied sources/destinations part of the IR
1469 - ir3: Split read-modify-write array dests in two
1470 - ir3: Update ir3_register::instr when cloning instructions
1471 - ir3: Validate that ir3_register::instr is correct
1472 - ir3: Add is_reg_special()
1473 - ir3: Make ir3_instruction::address a normal register
1474 - ir3: Split ir3_reg_create() into ir3_{src,dst}_create()
1475 - ir3: Add separate src/dst count in ir3_instr
1476 - ir3/legalize: Construct branch properly
1477 - ir3: Add srcs/dsts arrays to ir3_instruction
1480 - ir3/sched: Convert to srcs/dsts arrays
1481 - ir3/core: Switch to srcs/dsts arrays
1482 - ir3/ra: Switch to srcs/dsts arrays
1483 - ir3/parser: Switch to srcs/dsts arrays
1484 - ir3/array_to_ssa: Switch to srcs/dsts arrays
1485 - ir3/legalize: Switch to srcs/dsts arrays
1486 - ir3/print: Switch to srcs/dsts arrays
1487 - ir3/validate: Switch to srcs/dsts arrays
1488 - ir3/opts: Switch to srcs/dsts arrays
1489 - ir3/frontend: Switch to srcs/dsts arrays
1490 - ir3: Remove regs array
1491 - ir3: Remove IR3_REG_DEST
1492 - ir3/ra: Fix corner case in collect handling
1496 - ir3/sched: Speed up live_effect
1497 - ir3: Stop creating dummy dest registers
1498 - ir3: Prepare dest helpers for multi-dst instructions
1499 - ir3: Add foreach_dst/foreach_dst_n
1500 - ir3: Support multi-mov instructions
1501 - ir3/delay: Support multi-mov instructions
1502 - ir3/postsched: Support multi-mov instructions
1503 - ir3/legalize: Support multi-mov instructions
1504 - ir3: Use correct flags for movmsk & multi-mov
1505 - ir3/validate: Support multi-mov instructions
1506 - ir3: Print multi-mov instructions
1507 - ir3: Add min gen for multi-mov instructions
1508 - ir3/lower_parallelcopy: Use SWZ
1512 - tu, ir3: Plumb through support for CS subgroup size/id
1513 - ir3/nir: Call nir_lower_subgroups
1514 - ir3: Handle shared register liveness correctly
1515 - ir3: Handle unreachable blocks
1516 - ir3: Prevent propagating shared regs out of loops
1517 - ir3: Better valid flags for shared regs
1518 - ir3: Actually allow shared reg moves to be folded
1519 - ir3: Fix shared reg delay
1520 - ir3: Make MOVMSK use repeat
1521 - ir3: Fix infinite loop in scheduler when splitting
1522 - ir3/sched: Handle branch condition in split_pred()
1523 - ir3: Cleanup ir3_legalize jump optimization
1524 - ir3: Support any/all/getone branches
1525 - ir3: Add subgroup pseudoinstructions
1526 - ir3: Handle shared registers in lower_parallelcopy
1527 - ir3: Implement nir subgroup intrinsics
1528 - ir3: Fix convergence behavior for loops with continues
1529 - ir3/legalize: Fix loop convergence behavior
1531 - ir3/nir: Lower indirect references of compact variables
1532 - ir3: Add missing include to ir3_parser.y
1533 - ir3: Add ir3_collect() for fixed-size collects
1534 - ir3/lower_parallelcopy: Don't manually set wrmask
1535 - ir3: Update .editorconfig and .dir-locals.el
1536 - ir3: Manually reformat some places
1538 - ir3: Reformat source with clang-format
1539 - ir3/print: Manual formatting fixups
1540 - ir3: Preserve gl_ViewportIndex in the binning shader
1662 - ir3: treat 16b imul as mul.s24
1664 - ir3: do not double threadsize when exceeding branchstack limit
1665 - ir3: make possible to specify branchstack up to 64
1667 - ir3: do not move varying inputs that depend on unmovable instrs
1668 - ir3: do not fold cmps from different blocks with non-null address
1669 - ir3: memory_barrier also controls shared memory access order
1670 - ir3: update bar/fence bits in accordance to blob
1691 - ir3: add ldg.a,stg.a which allow complex in-place offset calculation
1693 - ir3: add newly found shlg.b16 instruction
1881 - freedreno/ir3: Add build id to the disassembler test
2077 - freedreno/ir3: Move the assert output to mesa_loge().
2079 - freedreno/ir3: Use mesa_log_stream() for ir3 disassembly.
2080 - freedreno/ir3: Move the native code output to mesa_log as well.
2082 - freedreno/ir3: Move NIR printing to mesa_log.
2102 - freedreno/ir3: Report RA failure with mesa_loge().
4499 - ir3: handle 16b op_i2b1
4555 - freedreno/ir3: Don't force RTNE if rounding mode is undefined
4579 - freedreno/ir3: Fix use after free
4620 - freedreno+ir3: Enable INT16
4625 - freedreno/ir3: Add float immed "FLUT" support
4630 - freedreno/ir3: Get tess_use_shared from fd_dev_info
4631 - freedreno/ir3: Get reg_size_vec4 from fd_dev_info