Windows build is broken.

[vinjn]

go build .
# github.com/jmorganca/ollama/llm

llm\llm.go:83:17: undefined: gpu.GetGPUInfo
llm\llm.go:89:9: undefined: nativeInit
llm\llm.go:92:109: undefined: extServer
llm\llm.go:94:15: undefined: newDynamicShimExtServer
llm\llm.go:101:9: undefined: newDefaultExtServer
llm\llama.go:211:24: undefined: libEmbed
llm\llama.go:218:19: undefined: libEmbed

[dhiltgen]

I forgot to update the docs (I'll post a PR later today). Make sure to enable CGO. Assuming powershell - $env:CGO_ENABLED="1"

You'll also need to install MinGW so it can compile the C/C++ code with the GCC toolchain.

Note: things are still a bit rough around the edges for native windows. I'm working on some improvements to the windows build so we should wind up with a single binary that can run without any special PATH and operate either on the CPU or CUDA card natively.

[vinjn]

Thanks Daniel. Will rebuild with your instructions.

[dhiltgen]

Keep an eye on https://github.com/jmorganca/ollama/pull/1680 which is where I'm planning to post the windows native build improvements once I get them sorted out.

[vinjn]

@dhiltgen great that #1680 is merged, does that mean I can also close this ticket?