There were two countries with completely different cultures, the Parallel Empire and the Serial Empire.
In this picture book, these two countries interact with each other using the latest technology.
パラレル帝国とシリアル皇国(the Parallel Empire and the Serial Empire.)Imaoka Micihihiro
パラレル帝国とシリアル皇国という全く文化の異なる二つの国がありました。その二つの国が最新のテクノロジーを用いて交流するという絵本です。
There were two countries with completely different cultures, the Parallel Empire and the Serial Empire.
In this picture book, these two countries interact with each other using the latest technology.
El gráfico muestra la temperatura del nodo final conectado a Internet con un teléfono inteligente.Con este mecanismo, puede ver el valor de cualquier sensor en el mundo conectado a Internet con el teléfono inteligente. Presentando el modelo de IoT más simple en la demostración.
Chart display the temperature of the end node connected to the Internet with a smartphone.With this mechanism you can see the value of any sensor in the world connected to the Internet with the smartphone.Presenting the simplest IoT model in the demo.
Controlamos el LED conectado al nodo final en Internet desde el teléfono inteligente.Presentando el modelo de IoT más simple en la demostración.El modelo te hace imaginar que todos los equipos conectados pueden controlar eléctricamente incluso si está en el borde del mundo.
How to control remote LED at the easiest and cheapest with AzureImaoka Micihihiro
We control the LED connected to the end node on the Internet from the smart phone.
Presenting the simplest IoT model in the demo.
The model make you image that all connected equipment can electric control even if it is on edge of the world.
6. 各社 動向
• Microsoft : Catapult,BrainWave
• Amazon : AWS F1
• Google : TPU
• IBM : Power-PC + GPU
7. Silicon alternatives for DNNs
CPU GPU
FPGA
(Soft DPU)
Hard DPU ASICs
Flexibility Efficency
Cerebras
Google TPU
Graphcore
Groq
Intel
Nervana
Movidius
Wave
Computing
BrainWave
Baidu SDA
Deephi Tech
ESE
Teradeep
Etc.
DSP
出典
Microsoft Accelerating
Persistent Neural Networks at
Datacenter Scale
8. 8
FPGAs ideal for adapting to rapidly evolving ML
CNNs, LSTMs, MLPs, reinforcement learning, feature
extraction, decision trees, etc.
Inference-optimized numerical precision
Exploit sparsity, deep compression for larger, faster models
Excellent inference performance at low batch sizes
Ultra-low latency serving on modern DNNs
>10X lower than CPUs and GPUs
Scale to many FPGAs in single DNN service
The power of Deep Learning on FPGA
Performance
Flexibility
Scale
Microsoft has the world’s largest cloud investment in FPGAs
Multiple Exa-Ops of aggregate AI capacity
BrainWave runs on Microsoft’s scale infrastructure
出典
Microsoft Accelerating Persistent Neural Networks at Datacenter Scale