mirror of
https://github.com/ollama/ollama.git
synced 2025-09-14 20:02:34 +02:00
Centralize GPU configuration vars
This should aid in troubleshooting by capturing and reporting the GPU settings at startup in the logs along with all the other server settings.
This commit is contained in:
@@ -57,6 +57,17 @@ var (
|
||||
SchedSpread bool
|
||||
// Set via OLLAMA_TMPDIR in the environment
|
||||
TmpDir string
|
||||
|
||||
// Set via CUDA_VISIBLE_DEVICES in the environment
|
||||
CudaVisibleDevices string
|
||||
// Set via HIP_VISIBLE_DEVICES in the environment
|
||||
HipVisibleDevices string
|
||||
// Set via ROCR_VISIBLE_DEVICES in the environment
|
||||
RocrVisibleDevices string
|
||||
// Set via GPU_DEVICE_ORDINAL in the environment
|
||||
GpuDeviceOrdinal string
|
||||
// Set via HSA_OVERRIDE_GFX_VERSION in the environment
|
||||
HsaOverrideGfxVersion string
|
||||
)
|
||||
|
||||
type EnvVar struct {
|
||||
@@ -66,7 +77,7 @@ type EnvVar struct {
|
||||
}
|
||||
|
||||
func AsMap() map[string]EnvVar {
|
||||
return map[string]EnvVar{
|
||||
ret := map[string]EnvVar{
|
||||
"OLLAMA_DEBUG": {"OLLAMA_DEBUG", Debug, "Show additional debug information (e.g. OLLAMA_DEBUG=1)"},
|
||||
"OLLAMA_FLASH_ATTENTION": {"OLLAMA_FLASH_ATTENTION", FlashAttention, "Enabled flash attention"},
|
||||
"OLLAMA_HOST": {"OLLAMA_HOST", Host, "IP Address for the ollama server (default 127.0.0.1:11434)"},
|
||||
@@ -84,6 +95,14 @@ func AsMap() map[string]EnvVar {
|
||||
"OLLAMA_SCHED_SPREAD": {"OLLAMA_SCHED_SPREAD", SchedSpread, "Always schedule model across all GPUs"},
|
||||
"OLLAMA_TMPDIR": {"OLLAMA_TMPDIR", TmpDir, "Location for temporary files"},
|
||||
}
|
||||
if runtime.GOOS != "darwin" {
|
||||
ret["CUDA_VISIBLE_DEVICES"] = EnvVar{"CUDA_VISIBLE_DEVICES", CudaVisibleDevices, "Set which NVIDIA devices are visible"}
|
||||
ret["HIP_VISIBLE_DEVICES"] = EnvVar{"HIP_VISIBLE_DEVICES", HipVisibleDevices, "Set which AMD devices are visible"}
|
||||
ret["ROCR_VISIBLE_DEVICES"] = EnvVar{"ROCR_VISIBLE_DEVICES", RocrVisibleDevices, "Set which AMD devices are visible"}
|
||||
ret["GPU_DEVICE_ORDINAL"] = EnvVar{"GPU_DEVICE_ORDINAL", GpuDeviceOrdinal, "Set which AMD devices are visible"}
|
||||
ret["HSA_OVERRIDE_GFX_VERSION"] = EnvVar{"HSA_OVERRIDE_GFX_VERSION", HsaOverrideGfxVersion, "Override the gfx used for all detected AMD GPUs"}
|
||||
}
|
||||
return ret
|
||||
}
|
||||
|
||||
func Values() map[string]string {
|
||||
@@ -256,6 +275,12 @@ func LoadConfig() {
|
||||
if err != nil {
|
||||
slog.Error("invalid setting", "OLLAMA_HOST", Host, "error", err, "using default port", Host.Port)
|
||||
}
|
||||
|
||||
CudaVisibleDevices = clean("CUDA_VISIBLE_DEVICES")
|
||||
HipVisibleDevices = clean("HIP_VISIBLE_DEVICES")
|
||||
RocrVisibleDevices = clean("ROCR_VISIBLE_DEVICES")
|
||||
GpuDeviceOrdinal = clean("GPU_DEVICE_ORDINAL")
|
||||
HsaOverrideGfxVersion = clean("HSA_OVERRIDE_GFX_VERSION")
|
||||
}
|
||||
|
||||
func getModelsDir() (string, error) {
|
||||
|
Reference in New Issue
Block a user