From 96e4a422b795a13b5a2e917114c0c477f8c55b80 Mon Sep 17 00:00:00 2001 From: Xan Date: Wed, 25 Apr 2018 12:42:41 +0100 Subject: [PATCH] --- ...RVP_vs_Andes_Packed_SIMD_ISA_proposal.mdwn | 54 +++++++++---------- 1 file changed, 27 insertions(+), 27 deletions(-) diff --git a/Comparative_analysis_Harmonised_RVP_vs_Andes_Packed_SIMD_ISA_proposal.mdwn b/Comparative_analysis_Harmonised_RVP_vs_Andes_Packed_SIMD_ISA_proposal.mdwn index b76f22a70..a3671bb82 100644 --- a/Comparative_analysis_Harmonised_RVP_vs_Andes_Packed_SIMD_ISA_proposal.mdwn +++ b/Comparative_analysis_Harmonised_RVP_vs_Andes_Packed_SIMD_ISA_proposal.mdwn @@ -1,32 +1,5 @@ # Comparative analysis of Andes Packed ISA proposal vs RVP Harmonised (with RV Vector spec) -## Proposed Harmonised RVP vector op instruction encoding - -Register x 2 -> register operations: - -| 31 30 29 28 27 26 | 25 | 24 23 22 21 20 | 19 18 17 16 15 | 14 | 13 12 | 11 10 9 8 7 | 6 5 4 3 2 1 0 | -| ----------------- | -- | -------------- | -------------- | -- | ----- | ----------- | ------------- | -| func_6 | 0 | rs2 | rs1 | 0 | mm | rd1 | VOP opcode | - -Immediate + register -> register operations: - -| 31 30 29 | 28 27 26 | 25 | 24 23 22 21 20 | 19 18 17 16 15 | 14 | 13 12 | 11 10 9 8 7 | 6 5 4 3 2 1 0 | -| -------- | -------- | -- | -------------- | -------------- | -- | ----- | ----------- | ------------- | -| func_3 | imm[7:5] | 1 | imm[4:0] | rs1 | 0 | mm | rd1 | VOP opcode | - -Register x 3 -> register operations: - -| 31 30 29 28 27 | 26 25 | 24 23 22 21 20 | 19 18 17 16 15 | 14 | 13 12 | 11 10 9 8 7 | 6 5 4 3 2 1 0 | -| ----------------------- | -------------- | -------------- | -- | ----- | ----------- | ------------- | -| rs3 | func_2 | rs2 | rs1 | 1 | mm | rd1 | VOP opcode | - -Values for mm field (bits 12:13 above): - -* mm = 00 -> no predicate mask, and use current global saturation / rounding settings -* mm = 00 -> no predicate mask, and force saturation or rounding for this instruction only -* mm = 10 -> use v1 as predicate mask, and use global saturation / rounding settings -* mm = 11 -> use ~v1 as predicate mask, and use global saturation / rounding settings - ## Register file comparison The default Harmonised RVP GPR register file is divided into a lower bank of Vector[INT8] and an upper bank of Vector[INT16]. @@ -71,6 +44,33 @@ In contrast, the Andes Packed SIMD ISA permits any GPR to be used for either INT | v31 | 32bit GPR or Vector[4xINT8 or 2xINT16] | 32bit GPR or Vector[1xSINT32] | +## Proposed Harmonised RVP vector op instruction encoding + +Register x 2 -> register operations: + +| 31 30 29 28 27 26 | 25 | 24 23 22 21 20 | 19 18 17 16 15 | 14 | 13 12 | 11 10 9 8 7 | 6 5 4 3 2 1 0 | +| ----------------- | -- | -------------- | -------------- | -- | ----- | ----------- | ------------- | +| func_6 | 0 | rs2 | rs1 | 0 | mm | rd1 | VOP opcode | + +Immediate + register -> register operations: + +| 31 30 29 | 28 27 26 | 25 | 24 23 22 21 20 | 19 18 17 16 15 | 14 | 13 12 | 11 10 9 8 7 | 6 5 4 3 2 1 0 | +| -------- | -------- | -- | -------------- | -------------- | -- | ----- | ----------- | ------------- | +| func_3 | imm[7:5] | 1 | imm[4:0] | rs1 | 0 | mm | rd1 | VOP opcode | + +Register x 3 -> register operations: + +| 31 30 29 28 27 | 26 25 | 24 23 22 21 20 | 19 18 17 16 15 | 14 | 13 12 | 11 10 9 8 7 | 6 5 4 3 2 1 0 | +| ----------------------- | -------------- | -------------- | -- | ----- | ----------- | ------------- | +| rs3 | func_2 | rs2 | rs1 | 1 | mm | rd1 | VOP opcode | + +Values for mm field (bits 12:13 above): + +* mm = 00 -> no predicate mask, and use current global saturation / rounding settings +* mm = 00 -> no predicate mask, and force saturation or rounding for this instruction only +* mm = 10 -> use v1 as predicate mask, and use global saturation / rounding settings +* mm = 11 -> use ~v1 as predicate mask, and use global saturation / rounding settings + ## 16-bit Arithmetic | Andes Mnemonic | 16-bit Instruction | Harmonised RVP Equivalent | -- 2.30.2