mirrors/ollama
1
0
Fork 0
mirror of https://github.com/ollama/ollama.git synced 2025-05-11 10:26:53 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

Support forced spreading for multi GPU

Browse source 
Our default behavior today is to try to fit into a single GPU if possible.
Some users would prefer the old behavior of always spreading across
multiple GPUs even if the model can fit into one.  This exposes that
tunable behavior.
This commit is contained in:
Daniel Hiltgen 2024-05-08 14:32:42 -07:00
parent 6fd04ca922
commit 5e8ff556cb
2 changed files with 18 additions and 4 deletions
Download patch file Download diff file Expand all files Collapse all files

10
server/sched.go
View file

@ -558,10 +558,12 @@ func pickBestFitGPUs(req *LlmRequest, ggml *llm.GGML, gpus gpu.GpuInfoList) gpu.
sort.Sort(sort.Reverse(gpu.ByFreeMemory(sgl)))
// First attempt to fit the model into a single GPU
for _, g := range sgl {
if ok, estimatedVRAM = llm.PredictServerFit([]gpu.GpuInfo{g}, ggml, req.model.AdapterPaths, req.model.ProjectorPaths, req.opts); ok {
slog.Debug("new model will fit in available VRAM in single GPU, loading", "model", req.model.ModelPath, "gpu", g.ID, "available", g.FreeMemory, "required", format.HumanBytes2(estimatedVRAM))
return []gpu.GpuInfo{g}
if !envconfig.SchedSpread {
for _, g := range sgl {
if ok, estimatedVRAM = llm.PredictServerFit([]gpu.GpuInfo{g}, ggml, req.model.AdapterPaths, req.model.ProjectorPaths, req.opts); ok {
slog.Debug("new model will fit in available VRAM in single GPU, loading", "model", req.model.ModelPath, "gpu", g.ID, "available", g.FreeMemory, "required", format.HumanBytes2(estimatedVRAM))
return []gpu.GpuInfo{g}
}
}
}