Changes between Version 22 and Version 23 of GcnInstrsVop1
- Timestamp:
- 05/29/16 10:00:30 (8 years ago)
Legend:
- Unmodified
- Added
- Removed
- Modified
-
GcnInstrsVop1
v22 v23 1274 1274 Description: Convert 32-bit floating point value from SRC0 to signed 32-bit integer, and 1275 1275 store result to VDST. Conversion adds 0.5 to value and rounds negative infinity (floor). 1276 If value is higher/lower than maximal/minimal integer then store MAX_INT32/MIN_INT32 to VDST.1277 If input value is NaN/-NaN then store MAX_INT32/MIN_INT32 to VDST.<br />1276 If value is higher/lower than maximal/minimal integer then store MAX_INT32/MIN_INT32 to 1277 VDST. If input value is NaN/-NaN then store MAX_INT32/MIN_INT32 to VDST.<br /> 1278 1278 Operation:<br /> 1279 1279 <code>FLOAT SF = ASFLOAT(SRC0) … … 1483 1483 Opcode VOP3A: 423 (0x1a7) for GCN 1.0/1.1; 353 (0x161) for GCN 1.2<br /> 1484 1484 Syntax: V_LOG_F32 VDST, SRC0<br /> 1485 Description: Approximate logarithm of base 2 from floating point value SRC0, and store result1486 to VDST. If SRC0 is negative then store -NaN to VDST.1485 Description: Approximate logarithm of base 2 from floating point value SRC0, and store 1486 result to VDST. If SRC0 is negative then store -NaN to VDST. 1487 1487 This instruction doesn't handle denormalized values regardless FLOAT MODE register setup.<br /> 1488 1488 Operation:<br /> … … 1498 1498 Opcode VOP3A: 453 (0x1c5) for GCN 1.1; 396 (0x18c) for GCN 1.2<br /> 1499 1499 Syntax: V_LOG_LEGACY_F32 VDST, SRC0<br /> 1500 Description: Approximate logarithm of base 2 from floating point value SRC0, and store result1501 to VDST. If SRC0 is negative then store -NaN to VDST.1500 Description: Approximate logarithm of base 2 from floating point value SRC0, and store 1501 result to VDST. If SRC0 is negative then store -NaN to VDST. 1502 1502 This instruction doesn't handle denormalized values regardless FLOAT MODE register setup. 1503 1503 This instruction returns slightly different results than V_LOG_F32.<br /> … … 1607 1607 Syntax: V_RCP_LEGACY_F32 VDST, SRC0<br /> 1608 1608 Description: Approximate reciprocal from floating point value SRC0 and store it to VDST. 1609 Guaranted error below 1ulp. If SRC0 or VDST is zero or infinity then store 0 with proper sign1610 to VDST.<br />1609 Guaranted error below 1ulp. If SRC0 or VDST is zero or infinity then store 0 with proper 1610 sign to VDST.<br /> 1611 1611 Operation:<br /> 1612 1612 <code>FLOAT SF = ASFLOAT(SRC0)