Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[HW] Add support for single lane configuration #194

Open
wants to merge 16 commits into
base: main
Choose a base branch
from
Open

[HW] Add support for single lane configuration #194

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
16 commits
Select commit Hold shift + click to select a range
d960af8
[hardware] Changes for single lane configuration support
sharafat-10xEngineers Dec 9, 2022
cc09801
[hardware] hardware change for single lane.
sharafat-10xEngineers Dec 12, 2022
ef1a1e4
[hardware] Update hardware for single lane.
sharafat-10xEngineers Dec 13, 2022
d43e617
[hardware] updated hardware for vfirst single lane
sharafat-10xEngineers Dec 13, 2022
c042f1b
[hardware] Handle single lane case in addrgen module
sharafat-10xEngineers Dec 13, 2022
75c1b97
[hardware] :bug: fix issue with vfirst hardware
M-Ijaz-10x Dec 13, 2022
5502484
[hardware] Fixed vluxei, vsuxei
sharafat-10xEngineers Dec 15, 2022
074f214
[hardware] vid fixed
sharafat-10xEngineers Dec 19, 2022
e0c7f01
[config] Add the 1_lane configuration
sharafat-10xEngineers Nov 2, 2022
70dfdd9
[config] Update constraint on VLEN(8*64*NrLanes) for all configration
sharafat-10xEngineers Nov 2, 2022
0c18343
[ci] Add the 1_lane configuration to the CI
sharafat-10xEngineers Dec 23, 2022
588578e
[riscv-tests] Updated tests for single lane
sharafat-10xEngineers Dec 23, 2022
5f571d1
[makefile] Update makefile to run the seperate tests for single lane
sharafat-10xEngineers Dec 23, 2022
1734e04
:droplet: Fixed clang-format
sharafat-10xEngineers Dec 28, 2022
9338326
[makefile] Update Makefile
sharafat-10xEngineers Dec 28, 2022
c1c96c1
[config] update vlen to 128 for 1_lane.mk
sharafat-10xEngineers Dec 28, 2022
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 7 additions & 7 deletions .github/workflows/ci.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -167,7 +167,7 @@ jobs:
runs-on: ubuntu-20.04
strategy:
matrix:
ara_config: [2_lanes, 4_lanes, 8_lanes, 16_lanes]
ara_config: [1_lane, 2_lanes, 4_lanes, 8_lanes, 16_lanes]
needs: tc-llvm
steps:
- uses: actions/checkout@v3
Expand Down Expand Up @@ -198,7 +198,7 @@ jobs:
runs-on: ubuntu-20.04
strategy:
matrix:
ara_config: [2_lanes, 4_lanes, 8_lanes, 16_lanes]
ara_config: [1_lane, 2_lanes, 4_lanes, 8_lanes, 16_lanes]
needs: ["tc-llvm", "tc-gcc", "tc-isa-sim"]
steps:
- uses: actions/checkout@v3
Expand Down Expand Up @@ -236,7 +236,7 @@ jobs:
runs-on: ubuntu-20.04
strategy:
matrix:
ara_config: [2_lanes, 4_lanes, 8_lanes, 16_lanes]
ara_config: [1_lane, 2_lanes, 4_lanes, 8_lanes, 16_lanes]
needs: ["tc-verilator", "tc-isa-sim"]
steps:
- uses: actions/checkout@v3
Expand Down Expand Up @@ -286,7 +286,7 @@ jobs:
strategy:
matrix:
app: [hello_world, imatmul, fmatmul, iconv2d, fconv2d, fconv3d, jacobi2d, dropout, fft, dwt, exp, softmax, dotproduct, fdotproduct, pathfinder, roi_align]
ara_config: [2_lanes, 4_lanes, 8_lanes, 16_lanes]
ara_config: [1_lane, 2_lanes, 4_lanes, 8_lanes, 16_lanes]
needs: ["compile-ara", "compile-apps"]
steps:
- uses: actions/checkout@v3
Expand Down Expand Up @@ -318,7 +318,7 @@ jobs:
runs-on: ubuntu-20.04
strategy:
matrix:
ara_config: [2_lanes, 4_lanes, 8_lanes, 16_lanes]
ara_config: [1_lane, 2_lanes, 4_lanes, 8_lanes, 16_lanes]
needs: ["compile-ara", "compile-riscv-tests"]
steps:
- uses: actions/checkout@v3
Expand Down Expand Up @@ -450,7 +450,7 @@ jobs:
runs-on: ubuntu-20.04
strategy:
matrix:
ara_config: [2_lanes, 4_lanes, 8_lanes, 16_lanes]
ara_config: [1_lane, 2_lanes, 4_lanes, 8_lanes, 16_lanes]
needs: ["compile-ara", "compile-apps"]
steps:
- uses: actions/checkout@v3
Expand Down Expand Up @@ -668,7 +668,7 @@ jobs:
runs-on: ubuntu-20.04
strategy:
matrix:
ara_config: [2_lanes, 4_lanes, 8_lanes, 16_lanes]
ara_config: [1_lane, 2_lanes, 4_lanes, 8_lanes, 16_lanes]
if: always()
needs: ["simulate", "riscv-tests-spike", "riscv-tests-simv"]
steps:
Expand Down
18 changes: 17 additions & 1 deletion apps/Makefile
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -109,7 +109,15 @@ riscv_tests: $(CVA6_BINARIES) $(ARA_BINARIES)
define rvtest_compile_template
TESTS_$(1) := $(addprefix bin/, $($(addsuffix _ara_tests, $1)))

ifeq ($(nr_lanes), 1)
ifeq ($(1), rv64uv)
bin/$(1)-ara-%: $(TESTS_DIR)/$(1)/1_lane_tests/%.$(2) $(RUNTIME_GCC) linker_script
else
bin/$(1)-ara-%: $(TESTS_DIR)/$(1)/%.$(2) $(RUNTIME_GCC) linker_script
endif
else
bin/$(1)-ara-%: $(TESTS_DIR)/$(1)/%.$(2) $(RUNTIME_GCC) linker_script
endif
mkdir -p bin/
$$(RISCV_CC_GCC) -Iinclude -I$$(TESTS_DIR)/macros/scalar -I$$(TESTS_DIR)/macros/vector $$(RISCV_CCFLAGS_GCC) $$(RISCV_LDFLAGS_GCC) -o $$@ $$< $(RUNTIME_GCC) -T$$(CURDIR)/common/link.ld
$$(RISCV_OBJDUMP) $$(RISCV_OBJDUMP_FLAGS) -D $$@ > $$@.dump
Expand All @@ -119,7 +127,15 @@ endef
define rvtest_compile_template_c
TESTS_$(1) := $(addprefix bin/, $($(addsuffix _ara_tests, $1)))

ifeq ($(nr_lanes), 1)
ifeq ($(1), rv64uv)
bin/$(1)-ara-%: $(TESTS_DIR)/$(1)/1_lane_tests/%.$(2) $(RUNTIME_LLVM) linker_script
else
bin/$(1)-ara-%: $(TESTS_DIR)/$(1)/%.$(2) $(RUNTIME_LLVM) linker_script
endif
else
bin/$(1)-ara-%: $(TESTS_DIR)/$(1)/%.$(2) $(RUNTIME_LLVM) linker_script
endif
mkdir -p bin/
$$(RISCV_CC) -Iinclude -I$$(TESTS_DIR)/macros/scalar -I$$(TESTS_DIR)/macros/vector $$(RISCV_CCFLAGS) $$(RISCV_LDFLAGS) -o $$@ $$< $(RUNTIME_LLVM) -T$$(CURDIR)/common/link.ld
$$(RISCV_OBJDUMP) $$(RISCV_OBJDUMP_FLAGS) -D $$@ > $$@.dump
Expand Down Expand Up @@ -181,4 +197,4 @@ clean: riscv_tests_spike_clean benchmarks_clean
rm -vf $(RUNTIME_SPIKE)
for app in $(APPS); do cd $(APPS_DIR)/$${app} && rm -f $$(find . -name "*.c.o*" -o -name "*.S.o*") && cd ..; done

.INTERMEDIATE: $(addsuffix /main.c.o,$(APPS))
.INTERMEDIATE: $(addsuffix /main.c.o,$(APPS))
6 changes: 5 additions & 1 deletion apps/common/riscv_tests.mk
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,7 +8,11 @@ include $(TESTS_DIR)/rv64um/Makefrag
include $(TESTS_DIR)/rv64ua/Makefrag
include $(TESTS_DIR)/rv64uf/Makefrag
include $(TESTS_DIR)/rv64ud/Makefrag
include $(TESTS_DIR)/rv64uv/Makefrag
ifeq ($(nr_lanes), 1)
include $(TESTS_DIR)/rv64uv/1_lane_tests/Makefrag
else
include $(TESTS_DIR)/rv64uv/Makefrag
endif
include $(TESTS_DIR)/rv64si/Makefrag

rv64ui_ara_tests := $(addprefix rv64ui-ara-, $(rv64ui_sc_tests))
Expand Down
178 changes: 178 additions & 0 deletions apps/riscv-tests/isa/rv64uv/1_lane_tests/Makefrag
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,178 @@
# Copyright 2021 ETH Zurich and University of Bologna.
# Solderpad Hardware License, Version 0.51, see LICENSE for details.
# SPDX-License-Identifier: SHL-0.51
#
# Author: Matheus Cavalcante <matheusd@iis.ee.ethz.ch>
# Basile Bougenot <bbougenot@student.ethz.ch>

rv64uv_sc_tests = vaadd \
vaaddu\
vsadd \
vsaddu \
vsmul \
vssra \
vssrl \
vnclip \
vnclipu \
vadd \
vsub \
vrsub \
vwaddu \
vwsubu \
vwadd \
vwsub \
vsext \
vzext \
vadc \
vmadc \
vsbc \
vmsbc \
vand \
vor \
vxor \
vsll \
vsrl \
vsra \
vnsrl \
vnsra \
vmseq \
vmsne \
vmsltu \
vmslt \
vmsleu \
vmsle \
vmsgtu \
vmsgt \
vminu \
vmin \
vmaxu \
vmax \
vmul \
vmulh \
vmulhu \
vmulhsu \
vdivu \
vdiv \
vremu \
vrem \
vwmul \
vwmulu \
vwmulsu \
vmacc \
vnmsac \
vmadd \
vnmsub \
vwmaccu \
vwmacc \
vwmaccsu \
vwmaccus \
vmerge \
vmv \
vmvxs \
vmvsx \
vfmvfs \
vfmvsf \
vmvnrr \
vredsum \
vredmaxu \
vredmax \
vredminu \
vredmin \
vredand \
vredor \
vredxor \
vwredsumu \
vwredsum \
vfadd \
vfsub \
vfrsub \
vfwadd \
vfwsub \
vfmul \
vfdiv \
vfrdiv \
vfwmul \
vfmacc \
vfnmacc \
vfmsac \
vfnmsac \
vfmadd \
vfnmadd \
vfmsub \
vfnmsub \
vfwmacc \
vfwnmacc \
vfwmsac \
vfwnmsac \
vfsqrt \
vfmin \
vfmax \
vfredusum \
vfredosum \
vfredmin \
vfredmax \
vfwredusum \
vfwredosum \
vfclass \
vfsgnj \
vfsgnjn \
vfsgnjx \
vfmerge \
vfmv \
vmfeq \
vmfne \
vmflt \
vmfle \
vmfgt \
vmfge \
vfcvt \
vfwcvt \
vfncvt \
vmand \
vmnand \
vmandnot \
vmor \
vmnor \
vmornot \
vmxor \
vmxnor \
vslideup \
vslidedown \
vslide1up \
vfslide1up \
vslide1down \
vfslide1down \
vl \
vl1r \
vle1 \
vls \
vluxei \
vs \
vs1r \
vse1 \
vss \
vsuxei \
vsetivli\
vsetvli\
vsetvl\
vmsbf \
vmsof \
vmsif \
viota \
vid \
vcpop \
vfirst \
vle8 \
vse8 \
vle16 \
vse16 \
vle32 \
vse32 \
vle64 \
vse64

#rv64uv_sc_tests = vaadd vaaddu vadc vasub vasubu vcompress vfirst vid viota vl vlff vl_nocheck vlx vmsbf vmsif vmsof vpopc_m vrgather vsadd vsaddu vsetvl vsetivli vsetvli vsmul vssra vssrl vssub vssubu vsux vsx

rv64uv_p_tests = $(addprefix rv64uv-p-, $(rv64uv_sc_tests))

spike_ctests += $(rv64uv_p_tests)
59 changes: 59 additions & 0 deletions apps/riscv-tests/isa/rv64uv/1_lane_tests/vaadd.c
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,59 @@
// Copyright 2021 ETH Zurich and University of Bologna.
// Solderpad Hardware License, Version 0.51, see LICENSE for details.
// SPDX-License-Identifier: SHL-0.51
//
// Author: Matheus Cavalcante <matheusd@iis.ee.ethz.ch>
// Basile Bougenot <bbougenot@student.ethz.ch>

#include "vector_macros.h"

void TEST_CASE1(void) {
set_vxrm(0); // setting vxrm to rnu rounding mode
VSET(4, e8, m1);
VLOAD_8(v1, 1, -2, -3, 4);
VLOAD_8(v2, 1, 2, -3, 3);
__asm__ volatile("vaadd.vv v3, v1, v2" ::);
VCMP_U8(1, v3, 1, 0, -3, 4);
}

void TEST_CASE2(void) {
set_vxrm(1); // setting vxrm to rne rounding mode
VSET(4, e8, m1);
VLOAD_8(v1, 1, -2, -3, 4);
VLOAD_8(v2, 1, 9, -3, 5);
VLOAD_8(v0, 0xA, 0x0, 0x0, 0x0);
VCLEAR(v3);
__asm__ volatile("vaadd.vv v3, v1, v2, v0.t" ::);
VCMP_U8(2, v3, 0, 4, 0, 4);
}

void TEST_CASE3(void) {
set_vxrm(2); // setting vxrm to rdn rounding mode
VSET(4, e32, m1);
VLOAD_32(v1, 1, -2, 3, -4);
const uint32_t scalar = 5;
__asm__ volatile("vaadd.vx v3, v1, %[A]" ::[A] "r"(scalar));
VCMP_U32(3, v3, 3, 1, 4, 0);
}

// Dont use VCLEAR here, it results in a glitch where are values are off by 1
void TEST_CASE4(void) {
set_vxrm(3); // setting vxrm to rod rounding mode
VSET(4, e32, m1);
VLOAD_32(v1, 1, 2, 3, 4);
const uint32_t scalar = 5;
VLOAD_32(v0, 0xA, 0x0, 0x0, 0x0);
VCLEAR(v3);
__asm__ volatile("vaadd.vx v3, v1, %[A], v0.t" ::[A] "r"(scalar));
VCMP_U32(4, v3, 0, 3, 0, 5);
}

int main(void) {
INIT_CHECK();
enable_vec();
TEST_CASE1();
TEST_CASE2();
TEST_CASE3();
TEST_CASE4();
EXIT_CHECK();
}
Loading