[VTA][TSIM] Enable TSIM CI Testing (#4407)

* Update task_python_vta.sh

* install sbt=1.1.1 with apt-get

* update verilator_opt

* install verilator with major version 4.0

* disable multi-threading for now

* bug fix for correcting uop fetch address in LoadUop module

* bug fix for correcting uop fetch address in LoadUop module

* adjustment to read from dram_offset

* enable USE_THREADS with verilator 4.x

* DEBUG: try avoid core dump with verilator 4.x

* bug fix in LoadUop module

* log mega cycles in tsim

* download cat.png to avoid fetching in each run

* bug fix in LoadUop module

* solve dram_even/sram_even issue

* bug fix

* introduce scalalint in ci

* speedup tsim in ci

* bug fix

* lint scala code before building

* disable multi-threading

* split fsim/tsim script

* update Jenkins settings

* duplicate task_python_vta_fsim.sh as task_python_vta.sh for now

Co-authored-by: Thierry Moreau <tmoreau@octoml.ai>

diff --git a/Makefile b/Makefile
index d34fbe4..ad24baa 100644 (file)
--- a/Makefile
+++ b/Makefile
@@ -17,7 +17,7 @@
 
 ROOTDIR = $(CURDIR)
 
-.PHONY: clean all test doc pylint cpplint lint\
+.PHONY: clean all test doc pylint cpplint scalalint lint\
         cython cython2 cython3 web runtime vta
 
 ifndef DMLC_CORE_PATH
@@ -80,7 +80,10 @@ pylint:
 jnilint:
        python3 3rdparty/dmlc-core/scripts/lint.py tvm4j-jni cpp jvm/native/src
 
-lint: cpplint pylint jnilint
+scalalint:
+       make -C vta/hardware/chisel lint
+
+lint: cpplint pylint jnilint scalalint
 
 doc:
        doxygen docs/Doxyfile

diff --git a/tests/scripts/task_python_vta.sh b/tests/scripts/task_python_vta.sh
index 1e5055e..a6bfb1c 100755 (executable)
--- a/tests/scripts/task_python_vta.sh
+++ b/tests/scripts/task_python_vta.sh
@@ -29,6 +29,9 @@ rm -rf ~/.tvm
 # Rebuild cython
 make cython3
 
+# Reset default fsim simulation
+cp vta/config/fsim_sample.json vta/config/vta_config.json
+
 # Run unit tests in functional/fast simulator
 echo "Running unittest in fsim..."
 python3 -m pytest -v vta/tests/python/unittest
@@ -36,20 +39,3 @@ python3 -m pytest -v vta/tests/python/unittest
 # Run unit tests in functional/fast simulator
 echo "Running integration test in fsim..."
 python3 -m pytest -v vta/tests/python/integration
-
-# # Build VTA chisel design and verilator simulator
-# (make -C vta/hardware/chisel/)
-
-# # Set default VTA config to use TSIM cycle accurate sim
-# cp vta/config/tsim_sample.json vta/config/vta_config.json
-
-# # Run unit tests in cycle accurate simulator
-# echo "Running unittest in tsim..."
-# python3 -m pytest -v vta/tests/python/unittest
-
-# # Run unit tests in cycle accurate simulator
-# echo "Running integration test in tsim..."
-# python3 -m pytest -v vta/tests/python/integration
-
-# # Reset default fsim simulation
-# cp vta/config/fsim_sample.json vta/config/vta_config.json

diff --git a/tests/scripts/task_python_vta_fsim.sh b/tests/scripts/task_python_vta_fsim.sh
index 1e5055e..a6bfb1c 100755 (executable)
--- a/tests/scripts/task_python_vta_fsim.sh
+++ b/tests/scripts/task_python_vta_fsim.sh
@@ -29,6 +29,9 @@ rm -rf ~/.tvm
 # Rebuild cython
 make cython3
 
+# Reset default fsim simulation
+cp vta/config/fsim_sample.json vta/config/vta_config.json
+
 # Run unit tests in functional/fast simulator
 echo "Running unittest in fsim..."
 python3 -m pytest -v vta/tests/python/unittest
@@ -36,20 +39,3 @@ python3 -m pytest -v vta/tests/python/unittest
 # Run unit tests in functional/fast simulator
 echo "Running integration test in fsim..."
 python3 -m pytest -v vta/tests/python/integration
-
-# # Build VTA chisel design and verilator simulator
-# (make -C vta/hardware/chisel/)
-
-# # Set default VTA config to use TSIM cycle accurate sim
-# cp vta/config/tsim_sample.json vta/config/vta_config.json
-
-# # Run unit tests in cycle accurate simulator
-# echo "Running unittest in tsim..."
-# python3 -m pytest -v vta/tests/python/unittest
-
-# # Run unit tests in cycle accurate simulator
-# echo "Running integration test in tsim..."
-# python3 -m pytest -v vta/tests/python/integration
-
-# # Reset default fsim simulation
-# cp vta/config/fsim_sample.json vta/config/vta_config.json

diff --git a/tests/scripts/task_python_vta_tsim.sh b/tests/scripts/task_python_vta_tsim.sh
index 1e5055e..eba62e5 100755 (executable)
--- a/tests/scripts/task_python_vta_tsim.sh
+++ b/tests/scripts/task_python_vta_tsim.sh
@@ -29,27 +29,25 @@ rm -rf ~/.tvm
 # Rebuild cython
 make cython3
 
-# Run unit tests in functional/fast simulator
-echo "Running unittest in fsim..."
-python3 -m pytest -v vta/tests/python/unittest
-
-# Run unit tests in functional/fast simulator
-echo "Running integration test in fsim..."
-python3 -m pytest -v vta/tests/python/integration
+# Set default VTA config to use TSIM cycle accurate sim
+cp vta/config/tsim_sample.json vta/config/vta_config.json
 
-# # Build VTA chisel design and verilator simulator
-# (make -C vta/hardware/chisel/)
+# Check style of scala code
+echo "Check style of scala code..."
+make -C vta/hardware/chisel lint
 
-# # Set default VTA config to use TSIM cycle accurate sim
-# cp vta/config/tsim_sample.json vta/config/vta_config.json
+# Build VTA chisel design and verilator simulator
+echo "Building VTA chisel design..."
+make -C vta/hardware/chisel cleanall
+make -C vta/hardware/chisel USE_THREADS=0 lib
 
-# # Run unit tests in cycle accurate simulator
-# echo "Running unittest in tsim..."
-# python3 -m pytest -v vta/tests/python/unittest
+# Run unit tests in cycle accurate simulator
+echo "Running unittest in tsim..."
+python3 -m pytest -v vta/tests/python/unittest
 
-# # Run unit tests in cycle accurate simulator
-# echo "Running integration test in tsim..."
-# python3 -m pytest -v vta/tests/python/integration
+# Run unit tests in cycle accurate simulator
+echo "Running integration test in tsim..."
+python3 -m pytest -v vta/tests/python/integration
 
-# # Reset default fsim simulation
-# cp vta/config/fsim_sample.json vta/config/vta_config.json
+# Reset default fsim simulation
+cp vta/config/fsim_sample.json vta/config/vta_config.json

diff --git a/vta/hardware/chisel/Makefile b/vta/hardware/chisel/Makefile
index 743f6c6..a98c52c 100644 (file)
--- a/vta/hardware/chisel/Makefile
+++ b/vta/hardware/chisel/Makefile
@@ -94,7 +94,7 @@ else
 endif
 
 ifneq ($(USE_THREADS), 0)
-  verilator_opt += -threads $(USE_THREADS)
+  verilator_opt += --threads $(USE_THREADS)
   cxx_flags += -DVL_THREADED
   cxx_objs += $(verilator_build_dir)/verilated_threads.o
 endif

diff --git a/vta/hardware/chisel/src/main/scala/core/LoadUop.scala b/vta/hardware/chisel/src/main/scala/core/LoadUop.scala
index 3432ec6..c7d0f44 100644 (file)
--- a/vta/hardware/chisel/src/main/scala/core/LoadUop.scala
+++ b/vta/hardware/chisel/src/main/scala/core/LoadUop.scala
@@ -81,7 +81,8 @@ class LoadUop(debug: Boolean = false)(implicit p: Parameters) extends Module {
   val xmax = (1 << mp.lenBits).U
   val xmax_bytes = ((1 << mp.lenBits) * mp.dataBits / 8).U
 
-  val offsetIsEven = (dec.sram_offset % 2.U) === 0.U
+  val dram_even = (dec.dram_offset % 2.U) === 0.U
+  val sram_even = (dec.sram_offset % 2.U) === 0.U
   val sizeIsEven = (dec.xsize % 2.U) === 0.U
 
   val sIdle :: sReadCmd :: sReadData :: Nil = Enum(3)
@@ -129,11 +130,10 @@ class LoadUop(debug: Boolean = false)(implicit p: Parameters) extends Module {
   // read-from-dram
   val maskOffset = VecInit(Seq.fill(M_DRAM_OFFSET_BITS)(true.B)).asUInt
   when(state === sIdle) {
-    when(offsetIsEven) {
+    when(dram_even) {
       raddr := io.baddr | (maskOffset & (dec.dram_offset << log2Ceil(uopBytes)))
     }.otherwise {
-      raddr := (io.baddr | (maskOffset & (dec.dram_offset << log2Ceil(
-        uopBytes)))) - uopBytes.U
+      raddr := (io.baddr | (maskOffset & (dec.dram_offset << log2Ceil(uopBytes)))) - uopBytes.U
     }
   }.elsewhen(state === sReadData && xcnt === xlen && xrem =/= 0.U) {
     raddr := raddr + xmax_bytes
@@ -162,7 +162,7 @@ class LoadUop(debug: Boolean = false)(implicit p: Parameters) extends Module {
   val mem = SyncReadMem(uopDepth, chiselTypeOf(wdata))
   val wmask = Reg(Vec(numUop, Bool()))
 
-  when(offsetIsEven) {
+  when(sram_even) {
     when(sizeIsEven) {
       wmask := "b_11".U.asTypeOf(wmask)
     }.elsewhen(io.vme_rd.cmd.fire()) {
@@ -192,6 +192,12 @@ class LoadUop(debug: Boolean = false)(implicit p: Parameters) extends Module {
   }
 
   wdata := io.vme_rd.data.bits.asTypeOf(wdata)
+  when(dram_even === false.B && sram_even) {
+    wdata(0) := io.vme_rd.data.bits.asTypeOf(wdata)(1)
+  }.elsewhen(sram_even === false.B && dram_even) {
+    wdata(1) := io.vme_rd.data.bits.asTypeOf(wdata)(0)
+  }
+
   when(io.vme_rd.data.fire()) {
     mem.write(waddr, wdata, wmask)
   }

diff --git a/vta/hardware/dpi/tsim_device.cc b/vta/hardware/dpi/tsim_device.cc
index 1dae273..d197fbd 100644 (file)
--- a/vta/hardware/dpi/tsim_device.cc
+++ b/vta/hardware/dpi/tsim_device.cc
@@ -141,6 +141,8 @@ int VTADPISim() {
       tfp->dump(static_cast<vluint64_t>(trace_count * 2 + 1));
 #endif
     trace_count++;
+    if ((trace_count % 1000000) == 1)
+      fprintf(stderr, "[traced %dM cycles]\n", trace_count / 1000000);
     while (top->sim_wait) {
       top->clock = 0;
       std::this_thread::sleep_for(std::chrono::milliseconds(100));

diff --git a/vta/tutorials/frontend/deploy_vision_on_vta.py b/vta/tutorials/frontend/deploy_vision_on_vta.py
index 47773bd..154eb85 100644 (file)
--- a/vta/tutorials/frontend/deploy_vision_on_vta.py
+++ b/vta/tutorials/frontend/deploy_vision_on_vta.py
@@ -225,10 +225,11 @@ synset = eval(open(categ_fn).read())
 
 # Download test image
 image_url = 'https://homes.cs.washington.edu/~moreau/media/vta/cat.jpg'
-response = requests.get(image_url)
+image_fn = 'cat.png'
+download.download(image_url, image_fn)
 
 # Prepare test image for inference
-image = Image.open(BytesIO(response.content)).resize((224, 224))
+image = Image.open(image_fn).resize((224, 224))
 plt.imshow(image)
 plt.show()
 image = np.array(image) - np.array([123., 117., 104.])