Hard to predict what kind of impact this will have. Not a huge problem, but your local development may start to diverge what may be in your production environment in subtle ways. For example, your M1 mac will download the ARM based version of an Nginx image vs the intel version. That being said, even with that improvement, the M1 is about 1/4 the speed of an Intel mac when running Intel based docker containers.Īlso another less obvious discovery is that the docker engine will default download your machine's arch type if there is a choice in images. My unit tests ran faster when I assigned 4 instead of 8 cores. Spreading the workload to all 8 cores will split the work between the small and large cores. The machine I was testing on has 4 large cores and 4 small cores. I have seen a significant improvement in using LESS cores on the ARM chips. WARNING: The requested image's platform (linux/amd64) does not match the detected host platform (linux/arm64/v8) and no specific platform was requestedįASTCGI_PASS not set using default of 'unix:/var/run/php5-fpm.sock'.Įxecuting 'nginx -c /etc/nginx/nf'.ģ 14:17:09 9#9: io_setup() failed (38: Function not implemented)ģ 14:17:09 11#11: io_setup() failed (38: Function not implemented)įor what it is worth, I have been experimenting with M1 macs and our Docker environments. For example: $ docker run -it -rm mbentley/nginx As a matter of fact, Docker for Mac will give you a notice when you're running a non-native image as the first line of log output. If you're running a native image, there is no QEMU in play. sharing your entire home directory will be total crap).ĭocker for Mac has a setting in the UI where you can toggle between virtualization framework (the new one that came in Big Sur) and hypervisor framework (traditional method).

If you're not and just running a container and bind mounting volumes, volumes are sketchy performance wise, especially if you are bind mounting a large folder (e.g.

If you're doing a docker build where everything is in the context, that is one thing. It would be great if you could explain exactly what you're doing, how you're running the image and what your build method is.

If you're relying on sharing a large number of files into the container for your compile, you're probably going to have a bad time.

Close docker on mac