mirror of
https://github.com/ollama/ollama.git
synced 2025-07-30 05:13:53 +02:00
35fda7b4af
We don't get valid UUIDs for AMD GPUs on Windows, so the best option is to use the ordinal IDs. This brings us in line with what we currently do on the Ollama server - the only exception is AMD GPUs on Linux, which falls back to using ordinal IDs. The GGML implementation has no fallback but it doesn't appear to occur for any of the GPUs that we support. It's also possible that there are collisions between ordinal IDs for different libraries - however the only places where we use them are AMD on Windows and Metal on Mac, which can never occur on the same system.
109 lines
5.1 KiB
Diff
109 lines
5.1 KiB
Diff
From 0000000000000000000000000000000000000000 Mon Sep 17 00:00:00 2001
|
|
From: Jesse Gross <jesse@ollama.com>
|
|
Date: Thu, 24 Apr 2025 14:48:51 -0700
|
|
Subject: [PATCH] ggml: Export GPU UUIDs
|
|
|
|
This enables matching up devices and information reported by the backend
|
|
with tools (e.g. nvidia-smi) and system management libraries (e.g. nvml).
|
|
---
|
|
ggml/include/ggml-backend.h | 1 +
|
|
ggml/src/ggml-cuda/ggml-cuda.cu | 39 ++++++++++++++++++++++++++++++++
|
|
ggml/src/ggml-metal/ggml-metal.m | 1 +
|
|
3 files changed, 41 insertions(+)
|
|
|
|
diff --git a/ggml/include/ggml-backend.h b/ggml/include/ggml-backend.h
|
|
index 74e46716..48839339 100644
|
|
--- a/ggml/include/ggml-backend.h
|
|
+++ b/ggml/include/ggml-backend.h
|
|
@@ -152,6 +152,7 @@ extern "C" {
|
|
struct ggml_backend_dev_props {
|
|
const char * name;
|
|
const char * description;
|
|
+ const char * id;
|
|
size_t memory_free;
|
|
size_t memory_total;
|
|
enum ggml_backend_dev_type type;
|
|
diff --git a/ggml/src/ggml-cuda/ggml-cuda.cu b/ggml/src/ggml-cuda/ggml-cuda.cu
|
|
index cb0d8528..d6960174 100644
|
|
--- a/ggml/src/ggml-cuda/ggml-cuda.cu
|
|
+++ b/ggml/src/ggml-cuda/ggml-cuda.cu
|
|
@@ -2884,6 +2884,7 @@ struct ggml_backend_cuda_device_context {
|
|
int device;
|
|
std::string name;
|
|
std::string description;
|
|
+ std::string id;
|
|
};
|
|
|
|
static const char * ggml_backend_cuda_device_get_name(ggml_backend_dev_t dev) {
|
|
@@ -2896,6 +2897,11 @@ static const char * ggml_backend_cuda_device_get_description(ggml_backend_dev_t
|
|
return ctx->description.c_str();
|
|
}
|
|
|
|
+static const char * ggml_backend_cuda_device_get_id(ggml_backend_dev_t dev) {
|
|
+ ggml_backend_cuda_device_context * ctx = (ggml_backend_cuda_device_context *)dev->context;
|
|
+ return ctx->id.c_str();
|
|
+}
|
|
+
|
|
static void ggml_backend_cuda_device_get_memory(ggml_backend_dev_t dev, size_t * free, size_t * total) {
|
|
ggml_backend_cuda_device_context * ctx = (ggml_backend_cuda_device_context *)dev->context;
|
|
ggml_cuda_set_device(ctx->device);
|
|
@@ -2910,6 +2916,7 @@ static enum ggml_backend_dev_type ggml_backend_cuda_device_get_type(ggml_backend
|
|
static void ggml_backend_cuda_device_get_props(ggml_backend_dev_t dev, ggml_backend_dev_props * props) {
|
|
props->name = ggml_backend_cuda_device_get_name(dev);
|
|
props->description = ggml_backend_cuda_device_get_description(dev);
|
|
+ props->id = ggml_backend_cuda_device_get_id(dev);
|
|
props->type = ggml_backend_cuda_device_get_type(dev);
|
|
ggml_backend_cuda_device_get_memory(dev, &props->memory_free, &props->memory_total);
|
|
|
|
@@ -3458,6 +3465,38 @@ ggml_backend_reg_t ggml_backend_cuda_reg() {
|
|
CUDA_CHECK(cudaGetDeviceProperties(&prop, i));
|
|
dev_ctx->description = prop.name;
|
|
|
|
+ #if !defined(GGML_USE_HIP)
|
|
+ char id[64];
|
|
+ snprintf(id, sizeof(id),
|
|
+ "GPU-%02x%02x%02x%02x-%02x%02x-%02x%02x-%02x%02x-%02x%02x%02x%02x%02x%02x",
|
|
+ (unsigned char)prop.uuid.bytes[0],
|
|
+ (unsigned char)prop.uuid.bytes[1],
|
|
+ (unsigned char)prop.uuid.bytes[2],
|
|
+ (unsigned char)prop.uuid.bytes[3],
|
|
+ (unsigned char)prop.uuid.bytes[4],
|
|
+ (unsigned char)prop.uuid.bytes[5],
|
|
+ (unsigned char)prop.uuid.bytes[6],
|
|
+ (unsigned char)prop.uuid.bytes[7],
|
|
+ (unsigned char)prop.uuid.bytes[8],
|
|
+ (unsigned char)prop.uuid.bytes[9],
|
|
+ (unsigned char)prop.uuid.bytes[10],
|
|
+ (unsigned char)prop.uuid.bytes[11],
|
|
+ (unsigned char)prop.uuid.bytes[12],
|
|
+ (unsigned char)prop.uuid.bytes[13],
|
|
+ (unsigned char)prop.uuid.bytes[14],
|
|
+ (unsigned char)prop.uuid.bytes[15]
|
|
+ );
|
|
+ dev_ctx->id = id;
|
|
+ #else
|
|
+ #ifdef _WIN32
|
|
+ char id[16];
|
|
+ snprintf(id, sizeof(id), "%d", i);
|
|
+ dev_ctx->id = id;
|
|
+ #else
|
|
+ dev_ctx->id = "GPU-" + std::string(prop.uuid.bytes, 16);
|
|
+ #endif
|
|
+ #endif
|
|
+
|
|
ggml_backend_dev_t dev = new ggml_backend_device {
|
|
/* .iface = */ ggml_backend_cuda_device_interface,
|
|
/* .reg = */ ®,
|
|
diff --git a/ggml/src/ggml-metal/ggml-metal.m b/ggml/src/ggml-metal/ggml-metal.m
|
|
index 1b56f858..a9eeebc6 100644
|
|
--- a/ggml/src/ggml-metal/ggml-metal.m
|
|
+++ b/ggml/src/ggml-metal/ggml-metal.m
|
|
@@ -5703,6 +5703,7 @@ static enum ggml_backend_dev_type ggml_backend_metal_device_get_type(ggml_backen
|
|
static void ggml_backend_metal_device_get_props(ggml_backend_dev_t dev, struct ggml_backend_dev_props * props) {
|
|
props->name = ggml_backend_metal_device_get_name(dev);
|
|
props->description = ggml_backend_metal_device_get_description(dev);
|
|
+ props->id = "0";
|
|
props->type = ggml_backend_metal_device_get_type(dev);
|
|
ggml_backend_metal_device_get_memory(dev, &props->memory_free, &props->memory_total);
|
|
props->caps = (struct ggml_backend_dev_caps) {
|