SYCL: SOFTMAX F16 mask support and other fixes

qnixsynapse · qnixsynapse · commit 9af1835df5e0 · 2025-01-16T12:57:19.000+05:30
diff --git a/ggml/src/ggml-sycl/common.cpp b/ggml/src/ggml-sycl/common.cpp
@@ -79,7 +79,7 @@ void ggml_sycl_op_flatten(ggml_backend_sycl_context & ctx, const ggml_tensor *sr
     float * src0_ddf = (float *) src0->data;
     float * src1_ddf = use_src1 ? (float *) src1->data : nullptr;
     float *  dst_ddf = (float *) dst->data;
-
+ /* These are never used */
     ggml_sycl_pool_alloc<float> src0_f(ctx.pool());
     ggml_sycl_pool_alloc<float> src1_f(ctx.pool());
     ggml_sycl_pool_alloc<float>  dst_f(ctx.pool());
diff --git a/ggml/src/ggml-sycl/dmmv.cpp b/ggml/src/ggml-sycl/dmmv.cpp
@@ -2,6 +2,7 @@
 #include "dmmv.hpp"
 #include "dequantize.hpp"
 #include "presets.hpp"
+#include "ggml-impl.h"
 
 
 static void convert_f16(const void * vx, const int64_t ib, const int iqs, dfloat2 & v){
@@ -973,6 +974,7 @@ void ggml_sycl_op_dequantize_mul_mat_vec(
     }
 #else
     const dfloat * src1_dfloat = (const dfloat *) src1_ddf_i; // dfloat == float, no conversion
+    GGML_UNUSED(ctx);
 #endif // GGML_SYCL_F16
 
     switch (src0->type) {
@@ -1010,7 +1012,7 @@ void ggml_sycl_op_dequantize_mul_mat_vec(
             convert_mul_mat_vec_f16_sycl(src0_dd_i, src1_dfloat, dst_dd_i, ne00, row_diff, stream);
             break;
         default:
-            printf("ggml_sycl_op_dequantize_mul_mat_vec unsupported GGML_TYPE %d\n", src0->type);
+            GGML_LOG_ERROR("ggml_sycl_op_dequantize_mul_mat_vec unsupported GGML_TYPE %d\n", src0->type);
             GGML_ABORT("fatal error");
             break;
     }
diff --git a/ggml/src/ggml-sycl/ggml-sycl.cpp b/ggml/src/ggml-sycl/ggml-sycl.cpp
@@ -3802,10 +3802,6 @@ static void ggml_sycl_diag_mask_inf(ggml_backend_sycl_context & ctx, ggml_tensor
     ggml_sycl_op_flatten(ctx, dst->src[0], dst->src[1], dst, ggml_sycl_op_diag_mask_inf);
 }
 
-static void ggml_sycl_soft_max(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
-    ggml_sycl_op_flatten(ctx, dst->src[0], dst->src[1], dst, ggml_sycl_op_soft_max);
-}
-
 static void ggml_sycl_rope(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
     GGML_ASSERT(ggml_is_contiguous(dst->src[0])); // TODO: this restriction is temporary until non-cont support is implemented
     ggml_sycl_op_flatten(ctx, dst->src[0], dst->src[1], dst, ggml_sycl_op_rope);
@@ -4014,7 +4010,7 @@ bool ggml_sycl_compute_forward(ggml_backend_sycl_context & ctx, struct ggml_tens
             ggml_sycl_diag_mask_inf(ctx, dst);
             break;
         case GGML_OP_SOFT_MAX:
-            ggml_sycl_soft_max(ctx, dst);
+            ggml_sycl_op_soft_max(ctx, dst);
             break;
         case GGML_OP_ROPE:
             ggml_sycl_rope(ctx, dst);
diff --git a/ggml/src/ggml-sycl/softmax.cpp b/ggml/src/ggml-sycl/softmax.cpp
@@ -1,7 +1,15 @@
-#include "norm.hpp"
+#include "softmax.hpp"
 
-template <bool vals_smem, int ncols_template, int block_size_template>
-static void soft_max_f32(const float * x, const float * mask, float * dst, const int ncols_par,
+template <typename T> static inline float t2f32(T val) {
+    return static_cast<float>(val);
+}
+
+template <> inline float t2f32<sycl::half>(sycl::half val) {
+    return static_cast<float>(val);
+}
+
+template <bool vals_smem, int ncols_template, int block_size_template, typename T>
+static void soft_max_f32(const float * x, const T * mask, float * dst, const int ncols_par,
                          const int nrows_y, const float scale, const float max_bias, const float m0,
                          const float m1, uint32_t n_head_log2, const sycl::nd_item<3> &item_ct1, float *buf) {
     const int ncols = ncols_template == 0 ? ncols_par : ncols_template;
@@ -29,9 +37,10 @@ static void soft_max_f32(const float * x, const float * mask, float * dst, const
         slope = sycl::pow(base, float(exp));
     }
 
-    float *vals = vals_smem ? buf + std::max(nwarps, WARP_SIZE) : dst + rowx * ncols;
+    float *vals = vals_smem ? buf + sycl::max(nwarps, WARP_SIZE) : dst + rowx * ncols;
     float max_val = -INFINITY;
 
+#pragma unroll
     for (int col0 = 0; col0 < ncols; col0 += block_size) {
         const int col = col0 + tid;
 
@@ -42,7 +51,7 @@ static void soft_max_f32(const float * x, const float * mask, float * dst, const
         const int ix = rowx*ncols + col;
         const int iy = rowy*ncols + col;
 
-        const float val = x[ix]*scale + (mask ? slope*mask[iy] : 0.0f);
+        const float val = x[ix]*scale + (mask ? slope*t2f32(mask[iy]) : 0.0f);
 
         vals[col] = val;
         max_val = sycl::max(max_val, val);
@@ -65,7 +74,7 @@ static void soft_max_f32(const float * x, const float * mask, float * dst, const
         item_ct1.barrier(sycl::access::fence_space::local_space);
         max_val = buf[lane_id];
         for (size_t i = 1; i < nreduce; i += 1) {
-            max_val = std::max(max_val, buf[lane_id + i * WARP_SIZE]);
+            max_val = sycl::max(max_val, buf[lane_id + i * WARP_SIZE]);
         }
         max_val = warp_reduce_max(max_val, item_ct1);
     }
@@ -122,8 +131,8 @@ static void soft_max_f32(const float * x, const float * mask, float * dst, const
     }
 }
 
-template <bool vals_smem, int ncols_template, int block_size_template>
-static void soft_max_f32_submitter(const float * x, const float * mask, float * dst, const int ncols_par,
+template <bool vals_smem, int ncols_template, int block_size_template, typename T>
+static void soft_max_f32_submitter(const float * x, const T * mask, float * dst, const int ncols_par,
                                    const int nrows_y, const float scale, const float max_bias, const float m0,
                                    const float m1, uint32_t n_head_log2, sycl::range<3> block_nums, sycl::range<3> block_dims,
                                    const size_t n_local_scratch, queue_ptr stream) {
@@ -133,15 +142,16 @@ static void soft_max_f32_submitter(const float * x, const float * mask, float *
         cgh.parallel_for(
             sycl::nd_range<3>(block_nums * block_dims, block_dims),
             [=](sycl::nd_item<3> item_ct1) [[intel::reqd_sub_group_size(WARP_SIZE)]] {
-                soft_max_f32<vals_smem, ncols_template, block_size_template>(x, mask, dst, ncols_par,
+                soft_max_f32<vals_smem, ncols_template, block_size_template, T>(x, mask, dst, ncols_par,
                                                                              nrows_y, scale, max_bias, m0,
                                                                              m1, n_head_log2, item_ct1,
                                                                              get_pointer(local_buf_acc));
             });
     });
 }
 
-static void soft_max_f32_sycl(const float * x, const float * mask,
+template<typename T>
+static void soft_max_f32_sycl(const float * x, const T * mask,
                               float * dst, const int ncols_x, const int nrows_x,
                               const int nrows_y, const float scale, const float max_bias,
                               queue_ptr stream, int device) {
@@ -164,88 +174,99 @@ static void soft_max_f32_sycl(const float * x, const float * mask,
     const size_t local_mem_size = stream->get_device().get_info<sycl::info::device::local_mem_size>();
     if (n_local_scratch*sizeof(float) < local_mem_size) {
         if (ncols_x > max_block_size) {
-            soft_max_f32_submitter<true, 0, 0>(x, mask, dst, ncols_x, nrows_y, scale,
+            soft_max_f32_submitter<true, 0, 0, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                max_bias, m0, m1, n_head_log2, block_nums,
                                                block_dims, n_local_scratch, stream);
             return;
         }
         switch (ncols_x) {
             case 32:
-                soft_max_f32_submitter<true, 32, 32>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 32, 32, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                      max_bias, m0, m1, n_head_log2, block_nums,
                                                      block_dims, n_local_scratch, stream);
                 break;
             case 64:
-                soft_max_f32_submitter<true, 64, 64>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 64, 64, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                      max_bias, m0, m1, n_head_log2, block_nums,
                                                      block_dims, n_local_scratch, stream);
                 break;
             case 128:
-                soft_max_f32_submitter<true, 128, 128>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 128, 128, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                        max_bias, m0, m1, n_head_log2, block_nums,
                                                        block_dims, n_local_scratch, stream);
                 break;
             case 256:
-                soft_max_f32_submitter<true, 256, 256>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 256, 256, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                        max_bias, m0, m1, n_head_log2, block_nums,
                                                        block_dims, n_local_scratch, stream);
                 break;
             case 512:
-                soft_max_f32_submitter<true, 512, 512>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 512, 512, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                        max_bias, m0, m1, n_head_log2, block_nums,
                                                        block_dims, n_local_scratch, stream);
                 break;
             case 1024:
-                soft_max_f32_submitter<true, 1024, 1024>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 1024, 1024, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                          max_bias, m0, m1, n_head_log2, block_nums,
                                                          block_dims, n_local_scratch, stream);
                 break;
             case 2048:
-                soft_max_f32_submitter<true, 2048, 1024>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 2048, 1024, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                          max_bias, m0, m1, n_head_log2, block_nums,
                                                          block_dims, n_local_scratch, stream);
                 break;
             case 4096:
-                soft_max_f32_submitter<true, 4096, 1024>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 4096, 1024, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                          max_bias, m0, m1, n_head_log2, block_nums,
                                                          block_dims, n_local_scratch, stream);
                 break;
             default:
-                soft_max_f32_submitter<true, 0, 0>(x, mask, dst, ncols_x, nrows_y, scale,
+                soft_max_f32_submitter<true, 0, 0, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                                    max_bias, m0, m1, n_head_log2, block_nums,
                                                    block_dims, n_local_scratch, stream);
                 break;
         }
     } else {
-        soft_max_f32_submitter<false, 0, 0>(x, mask, dst, ncols_x, nrows_y, scale,
+        soft_max_f32_submitter<false, 0, 0, T>(x, mask, dst, ncols_x, nrows_y, scale,
                                             max_bias, m0, m1, n_head_log2, block_nums,
                                             block_dims, WARP_SIZE, stream);
     }
 }
 
-void ggml_sycl_op_soft_max(ggml_backend_sycl_context & ctx, const ggml_tensor *src0,
-                                  const ggml_tensor *src1, ggml_tensor *dst,
-                                  const float *src0_dd, const float *src1_dd,
-                                  float *dst_dd,
-                                  const queue_ptr &main_stream) {
+void ggml_sycl_op_soft_max(ggml_backend_sycl_context & ctx, ggml_tensor * dst) {
 
-    GGML_ASSERT(src0->type == GGML_TYPE_F32);
+    GGML_ASSERT(dst->src[0]->type == GGML_TYPE_F32);
     GGML_ASSERT( dst->type == GGML_TYPE_F32);
 
-#pragma message("TODO: add ggml_sycl_op_soft_max() F16 src1 support")
-#pragma message("ref:  https://github.com/ggerganov/llama.cpp/pull/5021")
-    GGML_ASSERT(!src1 || src1->type == GGML_TYPE_F32); // src1 contains mask and it is optional
+    GGML_ASSERT(!dst->src[1] || dst->src[1]->type == GGML_TYPE_F16 || dst->src[1]->type == GGML_TYPE_F32); // src1 contains mask and it is optional
 
-    const int64_t ne00 = src0->ne[0];
-    const int64_t nrows_x = ggml_nrows(src0);
-    const int64_t nrows_y = src0->ne[1];
+    const int64_t ne00 = dst->src[0]->ne[0];
+    const int64_t nrows_x = ggml_nrows(dst->src[0]);
+    const int64_t nrows_y = dst->src[0]->ne[1];
 
     float scale = 1.0f;
     float max_bias = 0.0f;
 
     memcpy(&scale, dst->op_params + 0, sizeof(float));
     memcpy(&max_bias, dst->op_params + 1, sizeof(float));
 
-    soft_max_f32_sycl(src0_dd, src1 ? src1_dd : nullptr, dst_dd, ne00,
-        nrows_x, nrows_y, scale, max_bias, main_stream, ctx.device);
+    const float * src0_dd = static_cast<const float *>(dst->src[0]->data);
+    float * dst_dd = static_cast<float *>(dst->data);
+
+    ggml_sycl_set_device(ctx.device);
+    dpct::queue_ptr main_stream = ctx.stream();
+
+    if (dst->src[1] && dst->src[1]->type == GGML_TYPE_F16) {
+        //printf("%s: fp16 mask\n", __func__);
+        const sycl::half * src1_dd = static_cast<sycl::half *>(dst->src[1]->data);
+        soft_max_f32_sycl<sycl::half>(src0_dd, src1_dd, dst_dd, ne00, nrows_x, nrows_y, scale, max_bias,
+                          main_stream, ctx.device);
+    } else if (dst->src[1] && dst->src[1]->type == GGML_TYPE_F32) {
+        //printf("%s: fp32 mask\n", __func__);
+        const float * src1_dd = static_cast<const float *>(dst->src[1]->data);
+        soft_max_f32_sycl<float>(src0_dd, src1_dd, dst_dd, ne00, nrows_x, nrows_y, scale, max_bias, main_stream, ctx.device);
+    } else {
+        /* mask unavailable */
+        soft_max_f32_sycl<float>(src0_dd, nullptr, dst_dd, ne00, nrows_x, nrows_y, scale, max_bias, main_stream, ctx.device);
+    }
 }
diff --git a/ggml/src/ggml-sycl/softmax.hpp b/ggml/src/ggml-sycl/softmax.hpp
@@ -15,10 +15,6 @@
 
 #include "common.hpp"
 
-void ggml_sycl_op_soft_max(ggml_backend_sycl_context &ctx, const ggml_tensor *src0,
-    const ggml_tensor *src1, ggml_tensor *dst,
-    const float *src0_dd, const float *src1_dd,
-    float *dst_dd,
-    const queue_ptr &main_stream);
+void ggml_sycl_op_soft_max(ggml_backend_sycl_context &ctx, ggml_tensor *dst);
 
 #endif // GGML_SYCL_SOFTMAX_HPP
diff --git a/ggml/src/ggml-sycl/wkv6.cpp b/ggml/src/ggml-sycl/wkv6.cpp
@@ -97,9 +97,6 @@ static void rwkv_wkv_f32_kernel(
 
 void ggml_sycl_op_rwkv_wkv6(ggml_backend_sycl_context& ctx, ggml_tensor* dst) {
 
-    const ggml_tensor *src0 = dst->src[0];
-    const ggml_tensor *src1 = dst->src[1];
-
     const float* k_d = (const float*)dst->src[0]->data;
     const float* v_d = (const float*)dst->src[1]->data;
     const float* r_d = (const float*)dst->src[2]->data;
@@ -137,7 +134,4 @@ void ggml_sycl_op_rwkv_wkv6(ggml_backend_sycl_context& ctx, ggml_tensor* dst) {
                 );
             });
     });
-
-    GGML_UNUSED(src0);
-    GGML_UNUSED(src1);
 }
