hummingbird
/
graphengine
Not watched
2
0
Fork 0
Code
Releases 13 Wiki Activity
Issues 0 Pull Requests 0 Datasets Model Cloudbrain
You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.
2600 Commits
28 Branches
21 MB
0 Download
 
 
 
 
 
 
Tree: 04f27bc560
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from '04f27bc560'
${ noResults }
graphengine/ge
zhangxiaokun 04f27bc560 Fix loop nesting 3 years ago
..
analyzer ge log optimize 3 years ago
client !1676 GE支持将混合精度黑白灰名单配置参数modify_mixlist传递给FE 3 years ago
common !1711 Add subscribe for step_info. 3 years ago
engine_manager fix sc warning of graphengine 3 years ago
executor support memcpy in singleop 3 years ago
ge_local_engine fixed compile issue 3 years ago
ge_runtime modified: CMakeLists.txt 3 years ago
generator bugfix for infer format for single op 3 years ago
graph Fix loop nesting 3 years ago
host_cpu_engine fixed compile issue 3 years ago
host_kernels modified: ge/host_kernels/gather_v2_kernel.cc 3 years ago
hybrid Fix loop nesting 3 years ago
inc update code from yellow zone 20201013 3 years ago
init Support session scope memory 3 years ago
ir_build !1676 GE支持将混合精度黑白灰名单配置参数modify_mixlist传递给FE 3 years ago
model run_graph_with_stream. 3 years ago
offline fixed compile issue 3 years ago
opskernel_manager report errormsg 3 years ago
plugin/engine fixed compile issue 3 years ago
proto change op_mapping and dump_task namespace. 3 years ago
session add modify_mixlist parameter. 3 years ago
single_op support memcpy in singleop 3 years ago
stub python3,7 modify python3 3 years ago
CMakeLists.txt support memcpy in singleop 3 years ago
README.md update cmake 4 years ago
ge_inference.mk !1564 Bugfix: Delete abandoned optimize pass 3 years ago
ge_runner.mk !1564 Bugfix: Delete abandoned optimize pass 3 years ago
module.mk update ge folder 4 years ago

README.md

Delete task

Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.

cancel
confirm

图引擎模块(GE)是MindSpore的一个子模块，其代码由C++实现，位于前端模块ME和底层硬件之间，起到承接作用。图引擎模块以ME下发的图作为输入，然后进行一系列的深度图优化操作，最后输出一张可以在底层硬件上高效运行的图。GE针对昇腾AI处理器的硬件结构特点，做了特定的优化工作，以此来充分发挥出昇腾AI处理器的强大算力。在进行模型训练/推理时，GE会被自动调用而用户并不感知。GE主要由GE API和GE Core两部分组成，详细的架构图如下所示

C++ Text Makefile Shell CMake other