AMDGPU / GFX Instructions GPU Native ISA AMD

1960 instructions - showing 100 per page, page 4 of 20 - AMDGPU/GFX is AMD's native, low-level GPU instruction set family, assembled directly with no virtual intermediate layer. Supported instructions and their exact encodings vary by GFX compatibility target (e.g. gfx942, gfx1100). Scalar instructions execute once per wavefront on the Scalar ALU; vector instructions execute per-lane on the Vector ALU, gated by the EXEC mask.

By category → 100 instructions
Mnemonic ⇅ Syntax Format ⇅ GFX Targets Unit ⇅ Summary
ds_write_b16 ds_write_b16 DS gfx1100 Store 16 bits of data from a vector register into a data share.
ds_write_b16_d16_hi ds_write_b16_d16_hi DS gfx1100 Store 16 bits of data from the high bits of a vector register into a data share.
ds_write_b32 ds_write_b32 ADDR, DATA, offset DS gfx942, gfx1100 LDS Unit Write one 32-bit value per lane to the Local Data Share (LDS).
ds_write_b64 ds_write_b64 DS gfx1100 Store 64 bits of data from a vector input register into a data share.
ds_write_b8 ds_write_b8 DS gfx1100 Store 8 bits of data from a vector register into a data share.
ds_write_b8_d16_hi ds_write_b8_d16_hi DS gfx1100 Store 8 bits of data from the high bits of a vector register into a data share.
ds_write_b96 ds_write_b96 DS gfx1100 Store 96 bits of data from a vector input register into a data share.
ds_write_src2_b32 ds_write_src2_b32 DS AMDGPU DS vector instruction operating on b32 data. (Format and name extracted from LLVM's AMDGPU backend source - semantics not yet curated.)
ds_write_src2_b64 ds_write_src2_b64 DS AMDGPU DS vector instruction operating on b64 data. (Format and name extracted from LLVM's AMDGPU backend source - semantics not yet curated.)
ds_wrxchg2_rtn_b32 ds_wrxchg2_rtn_b32 DS gfx1100 Swap two unsigned 32-bit integer values in the data registers with two locations in a data share.
ds_wrxchg2_rtn_b64 ds_wrxchg2_rtn_b64 DS gfx1100 Swap two unsigned 64-bit integer values in the data registers with two locations in a data share.
ds_wrxchg2st64_rtn_b32 ds_wrxchg2st64_rtn_b32 DS gfx1100 Swap two unsigned 32-bit integer values in the data registers with two locations in a data share.
ds_wrxchg2st64_rtn_b64 ds_wrxchg2st64_rtn_b64 DS gfx1100 Swap two unsigned 64-bit integer values in the data registers with two locations in a data share.
ds_wrxchg_rtn_b32 ds_wrxchg_rtn_b32 DS gfx1100 Swap an unsigned 32-bit integer value in the data register with a location in a data share.
ds_wrxchg_rtn_b64 ds_wrxchg_rtn_b64 DS gfx1100 Swap an unsigned 64-bit integer value in the data register with a location in a data share.
ds_xor_b32 ds_xor_b32 DS gfx1100 Calculate bitwise XOR given two unsigned 32-bit integer values stored in the data register and a location in a data share.
ds_xor_b64 ds_xor_b64 DS gfx1100 Calculate bitwise XOR given two unsigned 64-bit integer values stored in the data register and a location in a data share.
ds_xor_rtn_b32 ds_xor_rtn_b32 DS gfx1100 Calculate bitwise XOR given two unsigned 32-bit integer values stored in the data register and a location in a data share.
ds_xor_rtn_b64 ds_xor_rtn_b64 DS gfx1100 Calculate bitwise XOR given two unsigned 64-bit integer values stored in the data register and a location in a data share.
ds_xor_src2_b32 ds_xor_src2_b32 DS AMDGPU DS vector instruction operating on b32 data. (Format and name extracted from LLVM's AMDGPU backend source - semantics not yet curated.)
ds_xor_src2_b64 ds_xor_src2_b64 DS AMDGPU DS vector instruction operating on b64 data. (Format and name extracted from LLVM's AMDGPU backend source - semantics not yet curated.)
flat_atomic_add flat_atomic_add FLAT gfx1100 Add two unsigned 32-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_add_f32 flat_atomic_add_f32 FLAT gfx1100 Add a single-precision float value in the data register to a location in the flat aperture.
flat_atomic_add_f64 flat_atomic_add_f64 FLAT gfx942 Add a double-precision float value in the data register to a location in the flat aperture.
flat_atomic_add_u32 flat_atomic_add_u32 FLAT gfx1100 Add two unsigned 32-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_add_u64 flat_atomic_add_u64 FLAT gfx1100 Add two unsigned 64-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_add_x2 flat_atomic_add_x2 FLAT gfx1100 Add two unsigned 64-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_and flat_atomic_and FLAT gfx1100 Calculate bitwise AND given two unsigned 32-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_and_b32 flat_atomic_and_b32 FLAT gfx1100 Calculate bitwise AND given two unsigned 32-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_and_b64 flat_atomic_and_b64 FLAT gfx1100 Calculate bitwise AND given two unsigned 64-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_and_x2 flat_atomic_and_x2 FLAT gfx1100 Calculate bitwise AND given two unsigned 64-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_cmpswap flat_atomic_cmpswap FLAT gfx1100 Compare two unsigned 32-bit integer values stored in the data comparison register and a location in the flat aperture.
flat_atomic_cmpswap_b32 flat_atomic_cmpswap_b32 FLAT gfx1100 Compare two unsigned 32-bit integer values stored in the data comparison register and a location in the flat aperture.
flat_atomic_cmpswap_b64 flat_atomic_cmpswap_b64 FLAT gfx1100 Compare two unsigned 64-bit integer values stored in the data comparison register and a location in the flat aperture.
flat_atomic_cmpswap_f32 flat_atomic_cmpswap_f32 FLAT gfx1100 Compare two single-precision float values stored in the data comparison register and a location in the flat aperture.
flat_atomic_cmpswap_x2 flat_atomic_cmpswap_x2 FLAT gfx1100 Compare two unsigned 64-bit integer values stored in the data comparison register and a location in the flat aperture.
flat_atomic_cond_sub_u32 flat_atomic_cond_sub_u32 FLAT Subtract an unsigned 32-bit integer value in the data register from a location in the flat aperture only if the memory value is greater than or equal…
flat_atomic_csub_u32 flat_atomic_csub_u32 FLAT AMDGPU FLAT vector instruction operating on u32 data. (Format and name extracted from LLVM's AMDGPU backend source - semantics not yet curated.)
flat_atomic_dec flat_atomic_dec FLAT gfx1100 Decrement an unsigned 32-bit integer value from a location in the flat aperture with wraparound to a value in the data register if the decrement…
flat_atomic_dec_u32 flat_atomic_dec_u32 FLAT gfx1100 Decrement an unsigned 32-bit integer value from a location in the flat aperture with wraparound to a value in the data register if the decrement…
flat_atomic_dec_u64 flat_atomic_dec_u64 FLAT gfx1100 Decrement an unsigned 64-bit integer value from a location in the flat aperture with wraparound to a value in the data register if the decrement…
flat_atomic_dec_x2 flat_atomic_dec_x2 FLAT gfx1100 Decrement an unsigned 64-bit integer value from a location in the flat aperture with wraparound to a value in the data register if the decrement…
flat_atomic_fcmpswap flat_atomic_fcmpswap FLAT gfx1100 Compare two single-precision float values stored in the data comparison register and a location in the flat aperture.
flat_atomic_fcmpswap_x2 flat_atomic_fcmpswap_x2 FLAT Compare two double-precision float values stored in the data comparison register and a location in the flat aperture.
flat_atomic_fmax flat_atomic_fmax FLAT gfx1100 Select the maximum of two single-precision float inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_fmax_x2 flat_atomic_fmax_x2 FLAT Select the maximum of two double-precision float inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_fmin flat_atomic_fmin FLAT gfx1100 Select the minimum of two single-precision float inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_fmin_x2 flat_atomic_fmin_x2 FLAT Select the minimum of two double-precision float inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_inc flat_atomic_inc FLAT gfx1100 Increment an unsigned 32-bit integer value from a location in the flat aperture with wraparound to 0 if the value exceeds a value in the data…
flat_atomic_inc_u32 flat_atomic_inc_u32 FLAT gfx1100 Increment an unsigned 32-bit integer value from a location in the flat aperture with wraparound to 0 if the value exceeds a value in the data…
flat_atomic_inc_u64 flat_atomic_inc_u64 FLAT gfx1100 Increment an unsigned 64-bit integer value from a location in the flat aperture with wraparound to 0 if the value exceeds a value in the data…
flat_atomic_inc_x2 flat_atomic_inc_x2 FLAT gfx1100 Increment an unsigned 64-bit integer value from a location in the flat aperture with wraparound to 0 if the value exceeds a value in the data…
flat_atomic_max_f32 flat_atomic_max_f32 FLAT gfx1100 Select the maximum of two single-precision float inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_max_f64 flat_atomic_max_f64 FLAT gfx942 Select the maximum of two double-precision float inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_max_i32 flat_atomic_max_i32 FLAT gfx1100 Select the maximum of two signed 32-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_max_i64 flat_atomic_max_i64 FLAT gfx1100 Select the maximum of two signed 64-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_max_u32 flat_atomic_max_u32 FLAT gfx1100 Select the maximum of two unsigned 32-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_max_u64 flat_atomic_max_u64 FLAT gfx1100 Select the maximum of two unsigned 64-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_min_f32 flat_atomic_min_f32 FLAT gfx1100 Select the minimum of two single-precision float inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_min_f64 flat_atomic_min_f64 FLAT gfx942 Select the minimum of two double-precision float inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_min_i32 flat_atomic_min_i32 FLAT gfx1100 Select the minimum of two signed 32-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_min_i64 flat_atomic_min_i64 FLAT gfx1100 Select the minimum of two signed 64-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_min_u32 flat_atomic_min_u32 FLAT gfx1100 Select the minimum of two unsigned 32-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_min_u64 flat_atomic_min_u64 FLAT gfx1100 Select the minimum of two unsigned 64-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_or flat_atomic_or FLAT gfx1100 Calculate bitwise OR given two unsigned 32-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_or_b32 flat_atomic_or_b32 FLAT gfx1100 Calculate bitwise OR given two unsigned 32-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_or_b64 flat_atomic_or_b64 FLAT gfx1100 Calculate bitwise OR given two unsigned 64-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_or_x2 flat_atomic_or_x2 FLAT gfx1100 Calculate bitwise OR given two unsigned 64-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_pk_add_bf16 flat_atomic_pk_add_bf16 FLAT Add a packed 2-component BF16 float value in the data register to a location in the flat aperture.
flat_atomic_pk_add_f16 flat_atomic_pk_add_f16 FLAT Add a packed 2-component half-precision float value in the data register to a location in the flat aperture.
flat_atomic_smax flat_atomic_smax FLAT gfx1100 Select the maximum of two signed 32-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_smax_x2 flat_atomic_smax_x2 FLAT gfx1100 Select the maximum of two signed 64-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_smin flat_atomic_smin FLAT gfx1100 Select the minimum of two signed 32-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_smin_x2 flat_atomic_smin_x2 FLAT gfx1100 Select the minimum of two signed 64-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_sub flat_atomic_sub FLAT gfx1100 Subtract an unsigned 32-bit integer value stored in the data register from a value stored in a location in the flat aperture.
flat_atomic_sub_u32 flat_atomic_sub_u32 FLAT gfx1100 Subtract an unsigned 32-bit integer value stored in the data register from a value stored in a location in the flat aperture.
flat_atomic_sub_u64 flat_atomic_sub_u64 FLAT gfx1100 Subtract an unsigned 64-bit integer value stored in the data register from a value stored in a location in the flat aperture.
flat_atomic_sub_x2 flat_atomic_sub_x2 FLAT gfx1100 Subtract an unsigned 64-bit integer value stored in the data register from a value stored in a location in the flat aperture.
flat_atomic_swap flat_atomic_swap FLAT gfx1100 Swap an unsigned 32-bit integer value in the data register with a location in the flat aperture.
flat_atomic_swap_b32 flat_atomic_swap_b32 FLAT gfx1100 Swap an unsigned 32-bit integer value in the data register with a location in the flat aperture.
flat_atomic_swap_b64 flat_atomic_swap_b64 FLAT gfx1100 Swap an unsigned 64-bit integer value in the data register with a location in the flat aperture.
flat_atomic_swap_x2 flat_atomic_swap_x2 FLAT gfx1100 Swap an unsigned 64-bit integer value in the data register with a location in the flat aperture.
flat_atomic_umax flat_atomic_umax FLAT gfx1100 Select the maximum of two unsigned 32-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_umax_x2 flat_atomic_umax_x2 FLAT gfx1100 Select the maximum of two unsigned 64-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_umin flat_atomic_umin FLAT gfx1100 Select the minimum of two unsigned 32-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_umin_x2 flat_atomic_umin_x2 FLAT gfx1100 Select the minimum of two unsigned 64-bit integer inputs, given two values stored in the data register and a location in the flat aperture.
flat_atomic_xor flat_atomic_xor FLAT gfx1100 Calculate bitwise XOR given two unsigned 32-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_xor_b32 flat_atomic_xor_b32 FLAT gfx1100 Calculate bitwise XOR given two unsigned 32-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_xor_b64 flat_atomic_xor_b64 FLAT gfx1100 Calculate bitwise XOR given two unsigned 64-bit integer values stored in the data register and a location in the flat aperture.
flat_atomic_xor_x2 flat_atomic_xor_x2 FLAT gfx1100 Calculate bitwise XOR given two unsigned 64-bit integer values stored in the data register and a location in the flat aperture.
flat_load_b128 flat_load_b128 FLAT gfx1100 Load 128 bits of data from the flat aperture into a vector register.
flat_load_b32 flat_load_b32 FLAT gfx1100 Load 32 bits of data from the flat aperture into a vector register.
flat_load_b64 flat_load_b64 FLAT gfx1100 Load 64 bits of data from the flat aperture into a vector register.
flat_load_b96 flat_load_b96 FLAT gfx1100 Load 96 bits of data from the flat aperture into a vector register.
flat_load_d16_b16 flat_load_d16_b16 FLAT gfx1100 Load 16 bits of unsigned data from the flat aperture and store the result into the low 16 bits of a 32-bit vector register.
flat_load_d16_hi_b16 flat_load_d16_hi_b16 FLAT gfx1100 Load 16 bits of unsigned data from the flat aperture and store the result into the high 16 bits of a 32-bit vector register.
flat_load_d16_hi_i8 flat_load_d16_hi_i8 FLAT gfx1100 Load 8 bits of signed data from the flat aperture, sign extend to 16 bits and store the result into the high 16 bits of a 32-bit vector register.
flat_load_d16_hi_u8 flat_load_d16_hi_u8 FLAT gfx1100 Load 8 bits of unsigned data from the flat aperture, zero extend to 16 bits and store the result into the high 16 bits of a 32-bit vector register.
flat_load_d16_i8 flat_load_d16_i8 FLAT gfx1100 Load 8 bits of signed data from the flat aperture, sign extend to 16 bits and store the result into the low 16 bits of a 32-bit vector register.
flat_load_d16_u8 flat_load_d16_u8 FLAT gfx1100 Load 8 bits of unsigned data from the flat aperture, zero extend to 16 bits and store the result into the low 16 bits of a 32-bit vector register.

Source

Normalized from AMD's official ROCm documentation, with the LLVM AMDGPU backend documentation as supplementary compiler-target information. ROCm documentation ↗