补充算子,支持llama (#23)

* module:linear简化 fix

* llama:RMSNorm ok,验证rpoe

* llama:rpoe todo

* llama:rpoe todo

* invert,rpowscalar:cuda&cpu

* equal,equalscalar,less,lessscalar,greater,greaterscalar,switch

Author: miaobyte

doc/excuter/op-mem-cuda/list.md

@@ -10,8 +10,10 @@
 | transpose | miaobyte | transpose(tensor<any> A, vector<int32> dim_order)->(tensor<any> C) | T2 = T1.transpose(dimorder=[1,0]) | transpose(tensor<any> A, vector<int32> dim_order)->(tensor<any> C) |
 | reshape | miaobyte | reshape(tensor<any> A, vector<int32> shape)->(tensor<any> B) | T1.reshape(shape)->T2 | reshape(tensor<any> A, vector<int32> shape)->(tensor<any> B) |
 | matmul | cublas | matmul(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=T1 @ T2 | matmul(tensor<any> A, tensor<any> B)->(tensor<any> C) |
-| comparescalar | miaobyte | comparescalar(tensor<any> A, var<any> scalar)->(tensor<int8> mask) | mask=compare(T1, scalar) | comparescalar(tensor<any> A, var<any> scalar)->(tensor<int8> mask) |
-| compare | miaobyte | compare(tensor<any> A, tensor<any> B)->(tensor<int8> mask) | mask=compare(T1, T2) | compare(tensor<any> A, tensor<any> B)->(tensor<int8> mask) |
+| equalscalar | miaobyte | equalscalar(tensor<any> A, var<any> scalar, var<float64> epsilon)->(tensor<bool> mask) | mask=compare(T1, scalar) | equalscalar(tensor<any> A, var<any> scalar, var<float64> epsilon)->(tensor<bool> mask) |
+| prod | miaobyte | prod(tensor<any> A, vector<int32> dims, var<bool> keepdims)->(tensor<any> B) | B = prod(A, axis=[1 2], keepdims=false) | prod(tensor<any> A, vector<int32> dims, var<bool> keepdims)->(tensor<any> B) |
+| min | miaobyte | min(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=min(T1, T2) | min(tensor<any> A, tensor<any> B)->(tensor<any> C) |
+| maxscalar | miaobyte | maxscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) | T3=max(T1, scalar) | maxscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) |
 | uniform | miaobyte | uniform(tensor<any> t, var<any> low, var<any> high, var<int32> seed)->() | uniform(T1,low,high,seed) | uniform(tensor<any> t, var<any> low, var<any> high, var<int32> seed)->() |
 | addscalar | miaobyte | addscalar(tensor<any> A, var<any> b)->(tensor<any> C) | T3=T1+scalar | addscalar(tensor<any> A, var<any> b)->(tensor<any> C) |
 | log | miaobyte | log(tensor<float64|float32|float16|bfloat16> A)->(tensor<float64|float32|float16|bfloat16> C) | T3=log(T1) | log(tensor<float64|float32|float16|bfloat16> A)->(tensor<float64|float32|float16|bfloat16> C) |
@@ -22,28 +24,28 @@
 | add | cublas | add(tensor<any> a, tensor<any> b)->(tensor<any> c) | T3=T1+T2 | add(tensor<any> a, tensor<any> b)->(tensor<any> c) |
 | add | miaobyte | add(tensor<any> a, tensor<any> b)->(tensor<any> c) | T3=T1+T2 | add(tensor<any> a, tensor<any> b)->(tensor<any> c) |
 | copytensor |  none  | copytensor(tensor<any> src, tensor<any> dst)->() | T2.data = T1.data | copytensor(tensor<any> src, tensor<any> dst)->() |
-| prod | miaobyte | prod(tensor<any> A, vector<int32> dims, var<bool> keepdims)->(tensor<any> B) | B = prod(A, axis=[1 2], keepdims=false) | prod(tensor<any> A, vector<int32> dims, var<bool> keepdims)->(tensor<any> B) |
-| min | miaobyte | min(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=min(T1, T2) | min(tensor<any> A, tensor<any> B)->(tensor<any> C) |
 | print | miaobyte | print(tensor<any> )->() | print(T1) | print(tensor<any> )->() |
 | print | miaobyte | print(tensor<any> , var<string> )->() | print(T1) | print(tensor<any> , var<string> )->() |
 | newtensor |  none  | newtensor(vector<int32> shape)->(tensor<any> tensor1) | T1 = zeros(shape) | newtensor(vector<int32> shape)->(tensor<any> tensor1) |
 | newtensor |  none  | newtensor(var<string> shape)->(tensor<any> tensor1) | T1 = zeros(shape) | newtensor(var<string> shape)->(tensor<any> tensor1) |
-| minscalar | miaobyte | minscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) | T3=min(T1, scalar) | minscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) |
-| rdivscalar | miaobyte | rdivscalar(var<any> scalar, tensor<any> A)->(tensor<any> C) | T3=scalar/T1 | rdivscalar(var<any> scalar, tensor<any> A)->(tensor<any> C) |
 | constant | miaobyte | constant(tensor<any> t, var<any> value)->() | constant(T1) | constant(tensor<any> t, var<any> value)->() |
-| powscalar | miaobyte | powscalar(tensor<float64|float32> A, var<float64|float32> scalar)->(tensor<float64|float32> C) | T3=pow(T1, scalar) | powscalar(tensor<float64|float32> A, var<float64|float32> scalar)->(tensor<float64|float32> C) |
+| powscalar | miaobyte | powscalar(tensor<float64|float32> A, var<float64|int32> scalar)->(tensor<float64|float32> C) | T3=pow(T1, scalar) | powscalar(tensor<float64|float32> A, var<float64|int32> scalar)->(tensor<float64|float32> C) |
 | vecset |  none  | vecset(vector<any> value)->(vector<any> name) | shape = [3  4  5] | vecset(vector<any> value)->(vector<any> name) |
 | reducemin | miaobyte | reducemin(tensor<any> A, vector<int32> dims, var<bool> keepdims)->(tensor<any> B) | B = reducemin(A, axis=[1 2], keepdims=false) | reducemin(tensor<any> A, vector<int32> dims, var<bool> keepdims)->(tensor<any> B) |
 | subscalar | miaobyte | subscalar(tensor<any> A, var<any> b)->(tensor<any> C) | T3=T1-scalar | subscalar(tensor<any> A, var<any> b)->(tensor<any> C) |
 | sqrt | miaobyte | sqrt(tensor<float64|float32|float16|bfloat16> A)->(tensor<float64|float32|float16|bfloat16> C) | T3=sqrt(T1) | sqrt(tensor<float64|float32|float16|bfloat16> A)->(tensor<float64|float32|float16|bfloat16> C) |
+| minscalar | miaobyte | minscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) | T3=min(T1, scalar) | minscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) |
+| rdivscalar | miaobyte | rdivscalar(var<any> scalar, tensor<any> A)->(tensor<any> C) | T3=scalar/T1 | rdivscalar(var<any> scalar, tensor<any> A)->(tensor<any> C) |
+| rpowscalar | miaobyte | rpowscalar(var<float64|int32> scalar, tensor<float64|float32> A)->(tensor<float64|float32> C) | T3=pow(scalar, T1) | rpowscalar(var<float64|int32> scalar, tensor<float64|float32> A)->(tensor<float64|float32> C) |
 | sub | miaobyte | sub(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=T1-T2 | sub(tensor<any> A, tensor<any> B)->(tensor<any> C) |
 | sum | miaobyte | sum(tensor<any> A, vector<int32> dims, var<bool> keepdims)->(tensor<any> B) | B = sum(A, axis=[1 2], keepdims=false) | sum(tensor<any> A, vector<int32> dims, var<bool> keepdims)->(tensor<any> B) |
 | argset |  none  | argset(var<any> value)->(var<any> name) | var argname = argvalue | argset(var<any> value)->(var<any> name) |
+| equal | miaobyte | equal(tensor<any> A, tensor<any> B, var<float64> epsilon)->(tensor<bool> mask) | mask=compare(T1, T2) | equal(tensor<any> A, tensor<any> B, var<float64> epsilon)->(tensor<bool> mask) |
 | mulscalar | miaobyte | mulscalar(tensor<any> A, var<any> b)->(tensor<any> C) | T3=T1*scalar | mulscalar(tensor<any> A, var<any> b)->(tensor<any> C) |
 | div | miaobyte | div(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=T1/T2 | div(tensor<any> A, tensor<any> B)->(tensor<any> C) |
+| invert | miaobyte | invert(tensor<int64|int32|int16|int8> A)->(tensor<int64|int32|int16|int8> C) | T3=~T1 | invert(tensor<int64|int32|int16|int8> A)->(tensor<int64|int32|int16|int8> C) |
 | max | miaobyte | max(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=max(T1, T2) | max(tensor<any> A, tensor<any> B)->(tensor<any> C) |
 | pow | miaobyte | pow(tensor<float64|float32> A, tensor<float64|float32> B)->(tensor<float64|float32> C) | T3=pow(T1, T2) | pow(tensor<float64|float32> A, tensor<float64|float32> B)->(tensor<float64|float32> C) |
-| maxscalar | miaobyte | maxscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) | T3=max(T1, scalar) | maxscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) |
 | mul | miaobyte | mul(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=T1*T2 | mul(tensor<any> A, tensor<any> B)->(tensor<any> C) |
 | exp | miaobyte | exp(tensor<float64|float32|float16|bfloat16> A)->(tensor<float64|float32|float16|bfloat16> C) | T3=exp(T1) | exp(tensor<float64|float32|float16|bfloat16> A)->(tensor<float64|float32|float16|bfloat16> C) |
 | deltensor |  none  | deltensor(tensor<any> t)->() | del T1 | deltensor(tensor<any> t)->() |

doc/excuter/op-mem-ompsimd/list.md

@@ -12,6 +12,7 @@
 | matmul | cblas | matmul(tensor<float64|float32> A, tensor<float64|float32> B)->(tensor<float64|float32> C) | T3=T1 @ T2 | matmul(tensor<float64|float32> A, tensor<float64|float32> B)->(tensor<float64|float32> C) |
 | matmul | miaobyte | matmul(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=T1 @ T2 | matmul(tensor<any> A, tensor<any> B)->(tensor<any> C) |
 | comparescalar | miaobyte | comparescalar(tensor<any> A, var<any> scalar)->(tensor<float32> mask) | mask=compare(T1,scalar) | comparescalar(tensor<any> A, var<any> scalar)->(tensor<float32> mask) |
+| compare | miaobyte | compare(tensor<any> A, tensor<any> B)->(tensor<float32> mask) | mask=compare(T1,T2) | compare(tensor<any> A, tensor<any> B)->(tensor<float32> mask) |
 | uniform | miaobyte | uniform(tensor<any> t, var<any> low, var<any> high, var<int32> seed)->() | uniform(T1,low,high,seed) | uniform(tensor<any> t, var<any> low, var<any> high, var<int32> seed)->() |
 | addscalar | miaobyte | addscalar(tensor<any> a, var<any> scalar)->(tensor<any> c) | T3=T1+scalar | addscalar(tensor<any> a, var<any> scalar)->(tensor<any> c) |
 | log | miaobyte | log(tensor<any> A)->(tensor<any> C) | T3=log(T1) | log(tensor<any> A)->(tensor<any> C) |
@@ -38,11 +39,12 @@
 | argset |  none  | argset(var<any> value)->(var<any> name) | var argname = argvalue | argset(var<any> value)->(var<any> name) |
 | mulscalar | miaobyte | mulscalar(tensor<any> A, var<any> b)->(tensor<any> C) | T3=T1*scalar | mulscalar(tensor<any> A, var<any> b)->(tensor<any> C) |
 | div | miaobyte | div(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=T1/T2 | div(tensor<any> A, tensor<any> B)->(tensor<any> C) |
+| invert | miaobyte | invert(tensor<int64|int32|int16|int8> A)->(tensor<int64|int32|int16|int8> C) | T3=~T1 | invert(tensor<int64|int32|int16|int8> A)->(tensor<int64|int32|int16|int8> C) |
 | max | miaobyte | max(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=max(T1,T2) | max(tensor<any> A, tensor<any> B)->(tensor<any> C) |
 | pow | miaobyte | pow(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=T1^T2 | pow(tensor<any> A, tensor<any> B)->(tensor<any> C) |
 | maxscalar | miaobyte | maxscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) | T3=max(T1,scalar) | maxscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) |
 | mul | miaobyte | mul(tensor<any> A, tensor<any> B)->(tensor<any> C) | T3=T1*T2 | mul(tensor<any> A, tensor<any> B)->(tensor<any> C) |
 | exp | miaobyte | exp(tensor<any> A)->(tensor<any> C) | T3=exp(T1) | exp(tensor<any> A)->(tensor<any> C) |
 | rdivscalar | miaobyte | rdivscalar(var<any> scalar, tensor<any> A)->(tensor<any> C) | T3=scalar/T1 | rdivscalar(var<any> scalar, tensor<any> A)->(tensor<any> C) |
+| rpowscalar | miaobyte | rpowscalar(var<any> scalar, tensor<any> A)->(tensor<any> C) | T3=scalar^T1 | rpowscalar(var<any> scalar, tensor<any> A)->(tensor<any> C) |
 | minscalar | miaobyte | minscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) | T3=min(T1,scalar) | minscalar(tensor<any> A, var<any> scalar)->(tensor<any> C) |
-| compare | miaobyte | compare(tensor<any> A, tensor<any> B)->(tensor<float32> mask) | mask=compare(T1,T2) | compare(tensor<any> A, tensor<any> B)->(tensor<float32> mask) |

excuter/cpp-common/src/deepx/tensor.hpp

@@ -4,6 +4,7 @@
 #include <vector>
 #include <string>
 #include <fstream>
+#include <memory>
 
 #include "deepx/shape.hpp"
 #include "deepx/dtype.hpp"
@@ -12,7 +13,7 @@
 namespace deepx
 {
     using namespace std;
-    
+
     template <typename T>
     struct Tensor : public TensorBase
     {
@@ -28,11 +29,11 @@ namespace deepx
         CopyFn copyer; // 拷贝内存
 
         Tensor() = default;
-        Tensor(const vector<int> &s)  
+        Tensor(const vector<int> &s)
         {
             shape = Shape(s);
         }
-        Tensor(const Shape &s)  
+        Tensor(const Shape &s)
         {
             shape = s;
         }
@@ -140,28 +141,6 @@ namespace deepx
             tensor.newer = nullptr;
             return *this;
         }
-    };
-
-    // template <typename T>
-    // struct TensorSlice {
-    //     Slice  slice;
-    //     Tensor<T> tensor;
-    // };
-
-    // 添加一个新的类用于类型擦除
-    struct TensorVoid : public TensorBase {
-        void* data;
-        void (*deleter)(void*);
-        void (*copyer)(void*, void*, int);
-        void* (*newer)(int);
-        
-        TensorVoid() = default;
-        ~TensorVoid() {
-            if (data && deleter) {
-                deleter(data);
-                data = nullptr;
-            }
-        }
-    };
+    }; 
 }
 #endif