Merge pull request #2190 from hansungk/graphics

Add GPU project chipyard changes

Author: richardyrh

Diff for: .github/scripts/check-commit.sh

@@ -46,7 +46,7 @@ search () {
 }
 
 
-submodules=("cva6" "boom" "ibex" "gemmini" "icenet" "nvdla" "rocket-chip" "rocket-chip-blocks" "rocket-chip-inclusive-cache" "testchipip" "riscv-sodor" "mempress" "bar-fetchers" "shuttle" "constellation" "fft-generator" "hardfloat" "caliptra-aes-acc" "rocc-acc-utils" "diplomacy" "rerocc" "compress-acc" "saturn" "ara" "vexiiriscv" "tacit")
+submodules=("cva6" "boom" "ibex" "gemmini" "icenet" "nvdla" "rocket-chip" "rocket-chip-blocks" "rocket-chip-inclusive-cache" "testchipip" "riscv-sodor" "mempress" "bar-fetchers" "shuttle" "constellation" "fft-generator" "hardfloat" "caliptra-aes-acc" "rocc-acc-utils" "diplomacy" "rerocc" "compress-acc" "saturn" "ara" "vexiiriscv" "tacit" "radiance")
 dir="generators"
 branches=("master" "main" "dev")
 search

Diff for: .github/scripts/defaults.sh

@@ -25,7 +25,7 @@ REMOTE_COURSIER_CACHE=$REMOTE_WORK_DIR/.coursier-cache
 
 # key value store to get the build groups
 declare -A grouping
-grouping["group-cores"]="chipyard-cva6 chipyard-ibex chipyard-rocket chipyard-hetero chipyard-boomv3 chipyard-boomv4 chipyard-sodor chipyard-digitaltop chipyard-multiclock-rocket chipyard-nomem-scratchpad chipyard-spike chipyard-clone chipyard-prefetchers chipyard-shuttle chipyard-shuttle3 chipyard-vexiiriscv chipyard-tacit-rocket"
+grouping["group-cores"]="chipyard-cva6 chipyard-ibex chipyard-rocket chipyard-hetero chipyard-boomv3 chipyard-boomv4 chipyard-sodor chipyard-digitaltop chipyard-multiclock-rocket chipyard-nomem-scratchpad chipyard-spike chipyard-clone chipyard-prefetchers chipyard-shuttle chipyard-shuttle3 chipyard-vexiiriscv chipyard-tacit-rocket chipyard-radiance"
 grouping["group-peripherals"]="chipyard-dmirocket chipyard-dmiboomv3 chipyard-dmiboomv4 chipyard-spiflashwrite chipyard-mmios chipyard-nocores chipyard-manyperipherals chipyard-chiplike chipyard-tethered chipyard-symmetric chipyard-llcchiplet"
 grouping["group-accels"]="chipyard-compressacc chipyard-mempress chipyard-gemmini chipyard-manymmioaccels chipyard-nvdla chipyard-aes256ecb chipyard-rerocc chipyard-rocketvector chipyard-shuttlevector chipyard-hlsacc" # chipyard-shuttleara - Add when Ara works again
 grouping["group-constellation"]="chipyard-constellation"
@@ -80,6 +80,7 @@ mapping["chipyard-rerocc"]=" CONFIG=ReRoCCTestConfig"
 mapping["chipyard-rocketvector"]=" CONFIG=MINV128D64RocketConfig"
 mapping["chipyard-shuttlevector"]=" CONFIG=GENV256D128ShuttleConfig"
 mapping["chipyard-shuttleara"]=" CONFIG=V4096Ara2LaneShuttleConfig USE_ARA=1 verilog"
+mapping["chipyard-radiance"]=" CONFIG=RadianceFP16ClusterConfig verilog"
 
 mapping["constellation"]=" SUB_PROJECT=constellation"
 mapping["icenet"]="SUB_PROJECT=icenet"

Diff for: .github/scripts/run-tests.sh

@@ -173,6 +173,9 @@ case $1 in
     chipyard-tacit-rocket)
         run_binary LOADMEM=1 BINARY=$RISCV/riscv64-unknown-elf/share/riscv-tests/benchmarks/dhrystone.riscv
         ;;
+    chipyard-radiance)
+        # Verilator fails to build sim binary, just generate verilog
+        ;;
     icenet)
         run_binary BINARY=none
         ;;

Diff for: .github/workflows/chipyard-run-tests.yml

@@ -535,6 +535,29 @@ jobs:
           group-key: "group-cores"
           project-key: "chipyard-shuttle3"
 
+  chipyard-radiance-run-tests:
+    name: chipyard-radiance-run-tests
+    needs: prepare-chipyard-cores
+    runs-on: as4
+    steps:
+      - name: Delete old checkout
+        run: |
+            ls -alh .
+            rm -rf ${{ github.workspace }}/* || true
+            rm -rf ${{ github.workspace }}/.* || true
+            ls -alh .
+      - name: Checkout
+        uses: actions/checkout@v4
+      - name: Git workaround
+        uses: ./.github/actions/git-workaround
+      - name: Create conda env
+        uses: ./.github/actions/create-conda-env
+      - name: Run tests
+        uses: ./.github/actions/run-tests
+        with:
+          group-key: "group-cores"
+          project-key: "chipyard-radiance"
+
   chipyard-tacit-rocket-run-tests:
     name: chipyard-tacit-rocket-run-tests
     needs: prepare-chipyard-cores

Diff for: .gitmodules

@@ -157,3 +157,6 @@
 [submodule "generators/tacit"]
 	path = generators/tacit
 	url = https://github.com/ucb-bar/tacit.git
+[submodule "generators/radiance"]
+	path = generators/radiance
+	url = https://github.com/ucb-bar/radiance.git

Diff for: build.sbt

@@ -156,7 +156,7 @@ lazy val testchipip = (project in file("generators/testchipip"))
 lazy val chipyard = (project in file("generators/chipyard"))
   .dependsOn(testchipip, rocketchip, boom, rocketchip_blocks, rocketchip_inclusive_cache,
     dsptools, rocket_dsp_utils,
-    gemmini, icenet, tracegen, cva6, nvdla, sodor, ibex, fft_generator,
+    radiance, gemmini, icenet, tracegen, cva6, nvdla, sodor, ibex, fft_generator,
     constellation, mempress, barf, shuttle, caliptra_aes, rerocc,
     compressacc, saturn, ara, firrtl2_bridge, vexiiriscv, tacit)
   .settings(libraryDependencies ++= rocketLibDeps.value)
@@ -243,6 +243,17 @@ lazy val sodor = (project in file("generators/riscv-sodor"))
   .settings(libraryDependencies ++= rocketLibDeps.value)
   .settings(commonSettings)
 
+lazy val radiance = (project in file("generators/radiance"))
+  .dependsOn(rocketchip, gemmini)
+  .settings(libraryDependencies ++= rocketLibDeps.value)
+  .settings(libraryDependencies ++= Seq(
+      "edu.berkeley.cs" %% "chiseltest" % chiselTestVersion,
+      "org.scalatest" %% "scalatest" % "3.2.+" % "test",
+      "junit" % "junit" % "4.13" % "test",
+      "org.scalacheck" %% "scalacheck" % "1.14.3" % "test",
+  ))
+  .settings(commonSettings)
+
 lazy val gemmini = freshProject("gemmini", file("generators/gemmini"))
   .dependsOn(rocketchip)
   .settings(libraryDependencies ++= rocketLibDeps.value)

Diff for: common.mk

@@ -16,6 +16,8 @@ HELP_COMPILATION_VARIABLES += \
 "   EXTRA_SIM_LDFLAGS         = additional LDFLAGS for building simulators" \
 "   EXTRA_SIM_SOURCES         = additional simulation sources needed for simulator" \
 "   EXTRA_SIM_REQS            = additional make requirements to build the simulator" \
+"   EXTRA_SIM_OUT_NAME        = additional suffix appended to the simulation .out log filename" \
+"   EXTRA_SIM_PREPROC_DEFINES = additional Verilog preprocessor defines passed to the simulator" \
 "   ENABLE_YOSYS_FLOW         = if set, add compilation flags to enable the vlsi flow for yosys(tutorial flow)" \
 "   EXTRA_CHISEL_OPTIONS      = additional options to pass to the Chisel compiler" \
 "   MFC_BASE_LOWERING_OPTIONS = override lowering options to pass to the MLIR FIRRTL compiler" \
@@ -26,6 +28,7 @@ EXTRA_SIM_CXXFLAGS   ?=
 EXTRA_SIM_LDFLAGS    ?=
 EXTRA_SIM_SOURCES    ?=
 EXTRA_SIM_REQS       ?=
+EXTRA_SIM_OUT_NAME   ?=
 
 ifneq ($(ASPECTS), )
 	comma = ,
@@ -67,6 +70,7 @@ include $(base_dir)/generators/ibex/ibex.mk
 include $(base_dir)/generators/ara/ara.mk
 include $(base_dir)/generators/tracegen/tracegen.mk
 include $(base_dir)/generators/nvdla/nvdla.mk
+include $(base_dir)/generators/radiance/radiance.mk
 include $(base_dir)/tools/torture.mk
 
 #########################################################################################
@@ -233,6 +237,8 @@ $(TOP_SMEMS_CONF) $(MODEL_SMEMS_CONF) &:  $(MFC_SMEMS_CONF) $(MFC_MODEL_HRCHY_JS
 		--model-module-name $(MODEL) \
 		--out-dut-smems-conf $(TOP_SMEMS_CONF) \
 		--out-model-smems-conf $(MODEL_SMEMS_CONF)
+#	for blackboxed SRAMs: add custom.mems.conf as blackbox and use generated module name in blackbox verilog source
+	-[ -f $(GEN_COLLATERAL_DIR)/custom.mems.conf ] && cat $(GEN_COLLATERAL_DIR)/custom.mems.conf >> $(TOP_SMEMS_CONF)
 
 # This file is for simulation only. VLSI flows should replace this file with one containing hard SRAMs
 TOP_MACROCOMPILER_MODE ?= --mode synflops
@@ -256,7 +262,7 @@ ifneq (,$(EXT_FILELISTS))
 else
 	rm -f $@
 endif
-	sort -u $(sim_files) $(ALL_MODS_FILELIST) | grep -v '.*\.\(svh\|h\)$$' >> $@
+	sort -u $(sim_files) $(ALL_MODS_FILELIST) | grep -v '.*\.\(svh\|h\|conf\)$$' >> $@
 	echo "$(TOP_SMEMS_FILE)" >> $@
 	echo "$(MODEL_SMEMS_FILE)" >> $@
 
@@ -305,15 +311,15 @@ get_loadarch_flag = +loadarch=$(subst mem.elf,loadarch,$(1))
 endif
 
 # get the output path base name for simulation outputs, First arg is the binary
-get_sim_out_name = $(output_dir)/$(call get_out_name,$(1))
+get_sim_out_name = $(output_dir)/$(call get_out_name,$(1))$(if $(EXTRA_SIM_OUT_NAME),.$(EXTRA_SIM_OUT_NAME),)
 # sim flags that are common to run-binary/run-binary-fast/run-binary-debug
 get_common_sim_flags = $(SIM_FLAGS) $(EXTRA_SIM_FLAGS) $(SEED_FLAG) $(call get_loadmem_flag,$(1)) $(call get_loadarch_flag,$(1))
 
 .PHONY: %.run %.run.debug %.run.fast
 
 # run normal binary with hardware-logged insn dissassembly
 run-binary: check-binary $(BINARY).run
-run-binaries: check-binaries $(addsuffix .run,$(BINARIES))
+run-binaries: check-binaries $(addsuffix .run,$(wildcard $(BINARIES)))
 
 %.run: %.check-exists $(SIM_PREREQ) | $(output_dir)
 	(set -o pipefail && $(NUMA_PREFIX) $(sim) \
@@ -327,7 +333,7 @@ run-binaries: check-binaries $(addsuffix .run,$(BINARIES))
 
 # run simulator as fast as possible (no insn disassembly)
 run-binary-fast: check-binary $(BINARY).run.fast
-run-binaries-fast: check-binaries $(addsuffix .run.fast,$(BINARIES))
+run-binaries-fast: check-binaries $(addsuffix .run.fast,$(wildcard $(BINARIES)))
 
 %.run.fast: %.check-exists $(SIM_PREREQ) | $(output_dir)
 	(set -o pipefail && $(NUMA_PREFIX) $(sim) \
@@ -340,7 +346,9 @@ run-binaries-fast: check-binaries $(addsuffix .run.fast,$(BINARIES))
 
 # run simulator with as much debug info as possible
 run-binary-debug: check-binary $(BINARY).run.debug
-run-binaries-debug: check-binaries $(addsuffix .run.debug,$(BINARIES))
+run-binary-debug-bg: check-binary $(BINARY).run.debug.bg
+run-binaries-debug: check-binaries $(addsuffix .run.debug,$(wildcard $(BINARIES)))
+run-binaries-debug-bg: check-binaries $(addsuffix .run.debug.bg,$(wildcard $(BINARIES)))
 
 %.run.debug: %.check-exists $(SIM_DEBUG_PREREQ) | $(output_dir)
 ifeq (1,$(DUMP_BINARY))
@@ -356,6 +364,19 @@ endif
 		$(BINARY_ARGS) \
 		</dev/null 2> >(spike-dasm > $(call get_sim_out_name,$*).out) | tee $(call get_sim_out_name,$*).log)
 
+%.run.debug.bg: %.check-exists $(SIM_DEBUG_PREREQ) | $(output_dir)
+	if [ "$*" != "none" ]; then riscv64-unknown-elf-objdump -D -S $* > $(call get_sim_out_name,$*).dump ; fi
+	(set -o pipefail && $(NUMA_PREFIX) $(sim_debug) \
+		$(PERMISSIVE_ON) \
+		$(call get_common_sim_flags,$*) \
+		$(VERBOSE_FLAGS) \
+		$(call get_waveform_flag,$(call get_sim_out_name,$*)) \
+		$(PERMISSIVE_OFF) \
+		$* \
+		$(BINARY_ARGS) \
+		</dev/null 2> >(spike-dasm > $(call get_sim_out_name,$*).out) >$(call get_sim_out_name,$*).log \
+		& echo "PID=$$!")
+
 run-fast: run-asm-tests-fast run-bmark-tests-fast
 
 #########################################################################################

Diff for: generators/chipyard/src/main/scala/DigitalTop.scala

@@ -32,6 +32,8 @@ class DigitalTop(implicit p: Parameters) extends ChipyardSystem
   with sifive.blocks.devices.gpio.HasPeripheryGPIO // Enables optionally adding the sifive GPIOs
   with sifive.blocks.devices.spi.HasPeripherySPIFlash // Enables optionally adding the sifive SPI flash controller
   with sifive.blocks.devices.spi.HasPeripherySPI // Enables optionally adding the sifive SPI port
+  with radiance.memory.CanHaveMemtraceCore // Enables memtrace core
+  with radiance.memory.CanHaveRadianceROMs // Enables radiance argument ROMs
   with icenet.CanHavePeripheryIceNIC // Enables optionally adding the IceNIC for FireSim
   with chipyard.example.CanHavePeripheryInitZero // Enables optionally adding the initzero example widget
   with chipyard.example.CanHavePeripheryGCD // Enables optionally adding the GCD example widget

Diff for: generators/chipyard/src/main/scala/config/CoalescerConfigs.scala

@@ -0,0 +1,20 @@
+package chipyard
+
+import org.chipsalliance.cde.config.{Config}
+import freechips.rocketchip.prci.AsynchronousCrossing
+
+class MemtraceCoreConfig extends Config(
+  // Memtrace
+  new radiance.subsystem.WithMemtraceCore("vecadd.core1.thread4.trace",
+    traceHasSource = false) ++
+  // new radiance.subsystem.WithMemtraceCore("nvbit.vecadd.n100000.filter_sm0.lane32.trace",
+  //   traceHasSource = false) ++
+  new radiance.subsystem.WithCoalescer(nNewSrcIds = 2) ++
+  new radiance.subsystem.WithSimtConfig(nMemLanes = 4, nSrcIds = 8) ++
+  // L2
+  new freechips.rocketchip.subsystem.WithInclusiveCache(nWays=8, capacityKB=512) ++
+  new freechips.rocketchip.subsystem.WithNBanks(4) ++
+  new chipyard.config.WithSystemBusWidth(16 * 8) ++
+  new chipyard.NoCoresConfig
+  )
+