调试器#
TVM 调试器是调试 TVM 计算图执行的接口。它有助于在 TVM 运行时提供对图结构和张量值的访问。
调试交换格式#
1. 计算图#
通过 relay 以 json 序列化格式构建的优化计算图被转储。它包含了关于图的全部信息。UX 可以直接使用这个图，也可以将这个图转换成 UX 可以理解的格式。
下面将解释 Graph JSON 格式：
1. nodes
Nodes are either placeholders or computational nodes in json. The nodes are stored
as a list. A node contains the below information
op- 运算类型，null表示它是占位符/变量/输入节点，而tvm_op表示这个节点可以被执行。name- 节点名称inputs- 此运算的输入位置，输入是元组列表，每个元组包含 (nodeid, index, version)。(可选)attrs- 节点的属性，包含以下信息：
flatten_data- 执行前是否需要展平此数据
func_name- 融合后的函数名称，对应于由 relay 编译过程生成的库中的符号。
num_inputs- 此节点的输入数量
num_outputs- 此节点产生的输出数量
2. arg_nodes
arg_nodes is a list of indices of nodes which is placeholder/variable/input or constant/param to the graph.
3. heads
heads is a list of entries as the output of the graph.
4. node_row_ptr
node_row_ptr stores the history of forward path, so you can skip constructing the entire graph in inference tasks.
5. attrs
attrs can contain version numbers or similar helpful information.
storage_id- 存储布局中每个节点的内存插槽ID。dtype- 每个节点的数据类型（枚举值）。dltype- 按顺序的每个节点的数据类型。shape- 每个节点 k 阶的形状。device_index- 图中每个条目的设备分配。
转储图的示例：
{
"nodes": [ # List of nodes
{
"op": "null", # operation type = null, this is a placeholder/variable/input or constant/param node
"name": "x", # Name of the argument node
"inputs": [] # inputs for this node, its none since this is an argument node
},
{
"op": "tvm_op", # operation type = tvm_op, this node can be executed
"name": "relu0", # Name of the node
"attrs": { # Attributes of the node
"flatten_data": "0", # Whether this data need to be flattened
"func_name": "fuse_l2_normalize_relu", # Fused function name, corresponds to the symbol in the lib generated by compilation process
"num_inputs": "1", # Number of inputs for this node
"num_outputs": "1" # Number of outputs this node produces
},
"inputs": [[0, 0, 0]] # Position of the inputs for this operation
}
],
"arg_nodes": [0], # Which all nodes in this are argument nodes
"node_row_ptr": [0, 1, 2], # Row indices for faster depth first search
"heads": [[1, 0, 0]], # Position of the output nodes for this operation
"attrs": { # Attributes for the graph
"storage_id": ["list_int", [1, 0]], # memory slot id for each node in the storage layout
"dtype": ["list_int", [0, 0]], # Datatype of each node (enum value)
"dltype": ["list_str", [ # Datatype of each node in order
"float32",
"float32"]],
"shape": ["list_shape", [ # Shape of each node k order
[1, 3, 20, 20],
[1, 3, 20, 20]]],
"device_index": ["list_int", [1, 1]], # Device assignment for each node in order
}
}
2. Tensor dumping#
执行后收到的张量在 tvm.ndarray 类型中所有的张量将以二进制字节的序列化格式保存。结果二进制字节可以通过 API “load_params” 加载。
- 加载参数的示例
- ::
- with open(path_params, “rb”) as fi:
loaded_params = bytearray(fi.read())
module.load_params(loaded_params)
如果使用 Debugger？#
在
config.cmake中设置USE_PROFILER为ON# Whether enable additional graph debug functions set(USE_PROFILER ON)
执行 make tvm，这样它就会生成
libtvm_runtime.so在前端脚本中替换
from tvm.contrib import graph_executor导入为from tvm.contrib.debugger.debug_executor import GraphModuleDebug
from tvm.contrib.debugger.debug_executor import GraphModuleDebug
m = GraphModuleDebug(
lib["debug_create"]("default", dev),
[dev],
lib.graph_json,
dump_root="/tmp/tvmdbg",
)
# set inputs
m.set_input('data', tvm.nd.array(data.astype(dtype)))
m.set_input(**params)
# execute
m.run()
tvm_out = m.get_output(0, tvm.nd.empty(out_shape, dtype)).numpy()
- 如果网络之前已经使用
lib.export_library("network.so")导出到外部库，那么可以使用以下代码导入该库： 与共享对象文件/动态链接库类似，初始化调试运行时会有一些不同。
- 如果网络之前已经使用
lib = tvm.runtime.load_module("network.so")
m = graph_executor.create(lib["get_graph_json"](), lib, dev, dump_root="/tmp/tvmdbg")
# set inputs
m.set_input('data', tvm.nd.array(data.astype(dtype)))
m.set_input(**params)
# execute
m.run()
tvm_out = m.get_output(0, tvm.nd.empty(out_shape, dtype)).numpy()
输出会转储到 /tmp 文件夹中的临时文件夹，或者在创建运行时时指定的文件夹。
输出示例#
下面是调试器的输出示例：
Node Name Ops Time(us) Time(%) Start Time End Time Shape Inputs Outputs
--------- --- -------- ------- ---------- -------- ----- ------ -------
1_NCHW1c fuse___layout_transform___4 56.52 0.02 15:24:44.177475 15:24:44.177534 (1, 1, 224, 224) 1 1
_contrib_conv2d_nchwc0 fuse__contrib_conv2d_NCHWc 12436.11 3.4 15:24:44.177549 15:24:44.189993 (1, 1, 224, 224, 1) 2 1
relu0_NCHW8c fuse___layout_transform___broadcast_add_relu___layout_transform__ 4375.43 1.2 15:24:44.190027 15:24:44.194410 (8, 1, 5, 5, 1, 8) 2 1
_contrib_conv2d_nchwc1 fuse__contrib_conv2d_NCHWc_1 213108.6 58.28 15:24:44.194440 15:24:44.407558 (1, 8, 224, 224, 8) 2 1
relu1_NCHW8c fuse___layout_transform___broadcast_add_relu___layout_transform__ 2265.57 0.62 15:24:44.407600 15:24:44.409874 (64, 1, 1) 2 1
_contrib_conv2d_nchwc2 fuse__contrib_conv2d_NCHWc_2 104623.15 28.61 15:24:44.409905 15:24:44.514535 (1, 8, 224, 224, 8) 2 1
relu2_NCHW2c fuse___layout_transform___broadcast_add_relu___layout_transform___1 2004.77 0.55 15:24:44.514567 15:24:44.516582 (8, 8, 3, 3, 8, 8) 2 1
_contrib_conv2d_nchwc3 fuse__contrib_conv2d_NCHWc_3 25218.4 6.9 15:24:44.516628 15:24:44.541856 (1, 8, 224, 224, 8) 2 1
reshape1 fuse___layout_transform___broadcast_add_reshape_transpose_reshape 1554.25 0.43 15:24:44.541893 15:24:44.543452 (64, 1, 1) 2 1