mirror of
https://github.com/dogkeeper886/ollama37.git
synced 2025-12-17 19:27:00 +00:00
Sync with upstream ollama/ollama and restore Tesla K80 (compute 3.7) support
This commit represents a complete rework after pulling the latest changes from official ollama/ollama repository and re-applying Tesla K80 compatibility patches. ## Key Changes ### CUDA Compute Capability 3.7 Support (Tesla K80) - Added sm_37 (compute 3.7) to CMAKE_CUDA_ARCHITECTURES in CMakeLists.txt - Updated CMakePresets.json to include compute 3.7 in "CUDA 11" preset - Using 37-virtual (PTX with JIT compilation) for maximum compatibility ### Legacy Toolchain Compatibility - **NVIDIA Driver**: 470.256.02 (last version supporting Kepler/K80) - **CUDA Version**: 11.4.4 (last CUDA 11.x supporting compute 3.7) - **GCC Version**: 10.5.0 (required by CUDA 11.4 host_config.h) ### CPU Architecture Trade-offs Due to GCC 10.5 limitation, sacrificed newer CPU optimizations: - Alderlake CPU variant enabled WITHOUT AVX_VNNI (requires GCC 11+) - Still supports: SSE4.2, AVX, F16C, AVX2, BMI2, FMA - Performance impact: ~3-7% on newer CPUs (acceptable for K80 compatibility) ### Build System Updates - Modified ml/backend/ggml/ggml/src/ggml-cuda/CMakeLists.txt for compute 3.7 - Added -Wno-deprecated-gpu-targets flag to suppress warnings - Updated ml/backend/ggml/ggml/src/CMakeLists.txt for Alderlake without AVX_VNNI ### Upstream Sync Merged latest llama.cpp changes including: - Enhanced KV cache management with ISWA and hybrid memory support - Improved multi-modal support (mtmd framework) - New model architectures (Gemma3, Llama4, Qwen3, etc.) - GPU backend improvements for CUDA, Metal, and ROCm - Updated quantization support and GGUF format handling ### Documentation - Updated CLAUDE.md with comprehensive build instructions - Documented toolchain constraints and CPU architecture trade-offs - Removed outdated CI/CD workflows (tesla-k80-*.yml) - Cleaned up temporary development artifacts ## Rationale This fork maintains Tesla K80 GPU support (compute 3.7) which was dropped in official Ollama due to legacy driver/CUDA requirements. The toolchain constraint creates a deadlock: - K80 → Driver 470 → CUDA 11.4 → GCC 10 → No AVX_VNNI We accept the loss of cutting-edge CPU optimizations to enable running modern LLMs on legacy but still capable Tesla K80 hardware (12GB VRAM per GPU). 🤖 Generated with [Claude Code](https://claude.com/claude-code) Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
@@ -125,7 +125,7 @@ func (p *Parser) parseToolCall() *api.ToolCall {
|
||||
}
|
||||
|
||||
var args map[string]any
|
||||
if found, i := findArguments(p.buffer); found == nil {
|
||||
if found, i := findArguments(tool, p.buffer); found == nil {
|
||||
return nil
|
||||
} else {
|
||||
args = found
|
||||
@@ -219,42 +219,80 @@ func findTool(tools []api.Tool, buf []byte) (*api.Tool, int) {
|
||||
// objects for functions that have all-optional parameters
|
||||
// e.g. `{"name": "get_conditions", "arguments": {}}` will work but
|
||||
// `{"name": "get_conditions"}` will not currently work
|
||||
func findArguments(buffer []byte) (map[string]any, int) {
|
||||
func findArguments(tool *api.Tool, buffer []byte) (map[string]any, int) {
|
||||
if len(buffer) == 0 {
|
||||
return nil, 0
|
||||
}
|
||||
|
||||
start := -1
|
||||
var braces int
|
||||
var start int = -1
|
||||
var inString, escaped bool
|
||||
|
||||
for i := range buffer {
|
||||
c := buffer[i]
|
||||
|
||||
if escaped {
|
||||
escaped = false
|
||||
continue
|
||||
}
|
||||
|
||||
if c == '\\' {
|
||||
escaped = true
|
||||
continue
|
||||
}
|
||||
|
||||
if c == '"' {
|
||||
inString = !inString
|
||||
continue
|
||||
}
|
||||
|
||||
if inString {
|
||||
continue
|
||||
}
|
||||
|
||||
for i, c := range buffer {
|
||||
if c == '{' {
|
||||
if braces == 0 {
|
||||
start = i
|
||||
}
|
||||
braces++
|
||||
} else if c == '}' && braces > 0 {
|
||||
} else if c == '}' {
|
||||
braces--
|
||||
if braces == 0 && start != -1 {
|
||||
object := buffer[start : i+1]
|
||||
|
||||
var data map[string]any
|
||||
if err := json.Unmarshal(object, &data); err != nil {
|
||||
// not a valid object, keep looking
|
||||
start = -1
|
||||
continue
|
||||
}
|
||||
|
||||
var findObject func(obj map[string]any) (map[string]any, bool)
|
||||
findObject = func(obj map[string]any) (map[string]any, bool) {
|
||||
if _, hasName := obj["name"]; hasName {
|
||||
if args, ok := obj["arguments"].(map[string]any); ok {
|
||||
findMap := func(name string, obj map[string]any) (map[string]any, bool) {
|
||||
if args, ok := obj[name].(map[string]any); ok {
|
||||
return args, true
|
||||
}
|
||||
if args, ok := obj["parameters"].(map[string]any); ok {
|
||||
if argsStr, ok := obj[name].(string); ok {
|
||||
var argsData map[string]interface{}
|
||||
if err := json.Unmarshal([]byte(argsStr), &argsData); err == nil {
|
||||
return argsData, ok
|
||||
}
|
||||
}
|
||||
return nil, false
|
||||
}
|
||||
if _, hasName := obj["name"]; hasName {
|
||||
if args, ok := findMap("arguments", obj); ok {
|
||||
return args, true
|
||||
}
|
||||
if args, ok := findMap("parameters", obj); ok {
|
||||
return args, true
|
||||
}
|
||||
return nil, true
|
||||
}
|
||||
if args, ok := findMap(tool.Function.Name, obj); ok {
|
||||
return args, true
|
||||
}
|
||||
|
||||
for _, v := range obj {
|
||||
switch child := v.(type) {
|
||||
@@ -282,6 +320,10 @@ func findArguments(buffer []byte) (map[string]any, int) {
|
||||
|
||||
return data, i
|
||||
}
|
||||
|
||||
if braces < 0 {
|
||||
braces = 0
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
|
||||
Reference in New Issue
Block a user