Retro Studio

Introducing Retro Studio our web IDE running an emulator of the Retro Watch.

Retro Studio will allow users to develop and test their own watch faces, games and applications before publishing them to the Retro Watch.

The back-end is written using Blazor Wasm and WebGL.

Stay tuned for more progress reports on this application!


Running tests on the DA14706 DevKit which has the Renesas ARM Cortex-M33F SoC we're using in the RetroWatch. Take a look at femto8 which is a PICO-8 fantasy console emulator as well as a port of DOOM both playable using a Bluetooth LE connected controller. So far performance is impressive and we're excited about bringing more games and emulators to the platform!

GPU Workload on EKS Cluster

Today I deployed my first GPU accelerated workload using the Karpenter autoscaler and the NVIDIA gpu-operator running on Amazon Web Services (AWS)'s Elastic Kubernetes Service 🙂

$ kubectl logs -l app=pytorch-mnist
Train Epoch: 2 [55680/60000 (93%)]  Loss: 0.161822
Train Epoch: 2 [56320/60000 (94%)]  Loss: 0.008159
Train Epoch: 2 [56960/60000 (95%)]  Loss: 0.217744
Train Epoch: 2 [57600/60000 (96%)]  Loss: 0.042390
Train Epoch: 2 [58240/60000 (97%)]  Loss: 0.055086
Train Epoch: 2 [58880/60000 (98%)]  Loss: 0.090941
Train Epoch: 2 [59520/60000 (99%)]  Loss: 0.156907

Test set: Average loss: 0.0345, Accuracy: 9890/10000 (99%)