readme: first commit

.gitmodules

@@ -1,4 +1,4 @@
 [submodule "3rdparty/ps-lite"]
 	path = 3rdparty/ps-lite
-	url = git@code.byted.org:lab/ps-lite.git
+	url = https://github.com/bytedance/ps-lite
 	branch = byteps

README.md

@@ -1 +1,52 @@
 # BytePS
+
+BytePS is a high performance and general distributed training framework that supports TensorFlow, Keras, PyTorch, and MXNet. 
+BytePS achieves higher performance than existing Parameter Server (PS) or Collective-based frameworks (such as MXNet-KVstore and Horovod). BytePS supports both TCP and RDMA. 
+
+
+BytePS achieves high performance because we leverage a hybrid communication strategy: combines PS and Collective primitives together, and incorporates acceleration techniques such as pipelining, tensor partitioning, NUMA-aware local communication and priority-based scheduling, etc. 
+
+* [Documentations](docs)
+* [Dockerfiles & Installations](docker)
+* [Examples](example)
+
+# Quick Start
+
+Before using BytePS, we assume you have already installed one or many of the following frameworks: TensorFlow / Keras / PyTorch / MXNet.
+
+First you need to clone and install [ps-lite](https://github.com/bytedance/ps-lite) as third party dependency:
+
+```
+git clone --recurse-submodules https://github.com/bytedance/byteps
+cd byteps/3rdparty/ps-lite && make -j4
+```
+
+Then install BytePS:
+
+```
+python byteps/setup.py install
+```
+
+Now you can try our examples. Let's say you are using MXNet and want to try a Resnet50 training benchmark:
+
+```
+export NVIDIA_VISIBLE_DEVICES=0,1 DMLC_NUM_WORKER=1 DMLC_NUM_SERVER=1 DMLC_WORKER_ID=0 DMLC_ROLE=worker DMLC_PS_ROOT_URI=10.0.0.1 DMLC_PS_ROOT_PORT=1234 DMLC_INTERFACE=eth0 
+python byteps/launcher/launch.py  byteps/example/mxnet/train_imagenet_horovod.py --benchmark 1 --batch-size=32 
+```
+
+# Performance
+
+Below shows BytePS's performance on a PCIe machine (8 GPUs). BytePS beats Horovod by 20% on VGG16 (communication-intensive) and 10% on Resnet50 (computation-intensive). 
+
+<img src="/images/perf_8gpu.png" width="360" height="240">
+
+
+For distributed training with RDMA (two workers, each with 8 GPUs), BytePS outperforms Horovod by 10-50%.
+
+<img src="/images/perf_16gpu_rdma.png" width="360" height="240">
+
+For distributed training with TCP (two workers, each with 8 GPUs), since the networking bandwidth is the bottleneck, the performance gain of BytePS is even higher. For example, BytePS outperforms Horovod by 100% on Alexnet training.
+
+<img src="/images/perf_16gpu_tcp.png" width="360" height="240">
+
+