github/gpt4all
1
0
Fork 0
mirror of https://github.com/nomic-ai/gpt4all.git synced 2025-09-07 11:30:05 +00:00
Code Issues Packages Projects Releases Wiki Activity

python: do not print GPU name with verbose=False, expose this info via properties (#2222)

Browse Source 
* llamamodel: only print device used in verbose mode

Signed-off-by: Jared Van Bortel <jared@nomic.ai>

* python: expose backend and device via GPT4All properties

Signed-off-by: Jared Van Bortel <jared@nomic.ai>

* backend: const correctness fixes

Signed-off-by: Jared Van Bortel <jared@nomic.ai>

* python: bump version

Signed-off-by: Jared Van Bortel <jared@nomic.ai>

* python: typing fixups

Signed-off-by: Jared Van Bortel <jared@nomic.ai>

* python: fix segfault with closed GPT4All

Signed-off-by: Jared Van Bortel <jared@nomic.ai>

---------

Signed-off-by: Jared Van Bortel <jared@nomic.ai>
This commit is contained in:
Jared Van Bortel
2024-04-18 14:52:02 -04:00
committed by GitHub
parent 271d752701
commit ba53ab5da0
8 changed files with 91 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
gpt4all-backend/llamamodel.cpp
View File

@@ -364,8 +364,10 @@ bool LLamaModel::loadModel(const std::string &modelPath, int n_ctx, int ngl)
d_ptr->end_tokens = {llama_token_eos(d_ptr->model)};
#ifdef GGML_USE_KOMPUTE
if (usingGPUDevice() && ggml_vk_has_device()) {
std::cerr << "llama.cpp: using Vulkan on " << ggml_vk_current_device().name << std::endl;
if (usingGPUDevice()) {
if (llama_verbose()) {
std::cerr << "llama.cpp: using Vulkan on " << ggml_vk_current_device().name << std::endl;
}
d_ptr->backend_name = "kompute";
}
#endif
@@ -558,7 +560,7 @@ bool LLamaModel::initializeGPUDevice(int device, std::string *unavail_reason) co
#endif
}
bool LLamaModel::hasGPUDevice()
bool LLamaModel::hasGPUDevice() const
{
#if defined(GGML_USE_KOMPUTE)
return d_ptr->device != -1;
@@ -567,10 +569,12 @@ bool LLamaModel::hasGPUDevice()
#endif
}
bool LLamaModel::usingGPUDevice()
bool LLamaModel::usingGPUDevice() const
{
#if defined(GGML_USE_KOMPUTE)
return hasGPUDevice() && d_ptr->model_params.n_gpu_layers > 0;
bool hasDevice = hasGPUDevice() && d_ptr->model_params.n_gpu_layers > 0;
assert(!hasDevice || ggml_vk_has_device());
return hasDevice;
#elif defined(GGML_USE_METAL)
return true;
#else
@@ -578,6 +582,19 @@ bool LLamaModel::usingGPUDevice()
#endif
}
const char *LLamaModel::backendName() const {
return d_ptr->backend_name;
}
const char *LLamaModel::gpuDeviceName() const {
#if defined(GGML_USE_KOMPUTE)
if (usingGPUDevice()) {
return ggml_vk_current_device().name;
}
#endif
return nullptr;
}
void llama_batch_add(
struct llama_batch & batch,
llama_token id,