metal : handle F16 inf values, fix FA partial offload (ggerganov#7434)

ggml-ci
teleprint-me · May 21, 2024 · 6369bf0 · 6369bf0
1 parent e402de3
commit 6369bf0
Showing 1 changed file with 12 additions and 15 deletions.
diff --git a/ggml-metal.metal b/ggml-metal.metal
@@ -2204,11 +2204,7 @@ kernel void kernel_flash_attn_ext_f16(
         // pointer to the mask
         device const half * mp = (device const half *) (mask + iq1*nb31);
 
-        // prepare diagonal scale matrix
-        simdgroup_float8x8 mscale(scale);
-
-        // prepare diagonal slope matrix
-        simdgroup_float8x8 mslope(1.0f);
+        float slope = 1.0f;
 
         // ALiBi
         if (max_bias > 0.0f) {
@@ -2217,7 +2213,7 @@ kernel void kernel_flash_attn_ext_f16(
             const float base = h < n_head_log2 ? m0 : m1;
             const int   exph = h < n_head_log2 ? h + 1 : 2*(h - n_head_log2) + 1;
 
-            mslope = simdgroup_float8x8(pow(base, exph));
+            slope = pow(base, exph);
         }
 
         // loop over the KV cache
@@ -2242,18 +2238,20 @@ kernel void kernel_flash_attn_ext_f16(
                         simdgroup_multiply_accumulate(mqk, mq[i], mk, mqk);
                     }
 
+                    simdgroup_store(mqk, ss + 8*cc, TF, 0, false);
+
+                    const short tx = tiisg%4;
+                    const short ty = tiisg/4;
+
                     if (mask != q) {
                         // mqk = mqk*scale + mask*slope
-                        simdgroup_half8x8 mm;
-                        simdgroup_load(mm, mp + ic + 8*cc, nb31/sizeof(half), 0, false);
-                        simdgroup_multiply(mm, mslope, mm);
-                        simdgroup_multiply_accumulate(mqk, mqk, mscale, mm);
+                        ss[8*cc + ty*TF + 2*tx + 0] = scale*ss[8*cc + ty*TF + 2*tx + 0] + slope*mp[ic + 8*cc + ty*nb31/sizeof(half) + 2*tx + 0];
+                        ss[8*cc + ty*TF + 2*tx + 1] = scale*ss[8*cc + ty*TF + 2*tx + 1] + slope*mp[ic + 8*cc + ty*nb31/sizeof(half) + 2*tx + 1];
                     } else {
                         // mqk = mqk*scale
-                        simdgroup_multiply(mqk, mscale, mqk);
+                        ss[8*cc + ty*TF + 2*tx + 0] *= scale;
+                        ss[8*cc + ty*TF + 2*tx + 1] *= scale;
                     }
-
-                    simdgroup_store(mqk, ss + 8*cc, TF, 0, false);
                 }
             }
 
@@ -2816,8 +2814,7 @@ kernel void kernel_cpy_f32_f16(
     for (int64_t i00 = tpitg.x; i00 < ne00; i00 += ntg.x) {
         device const float * src = (device float *)((device char *) src0 + i03*nb03 + i02*nb02 + i01*nb01 + i00*nb00);
 
-        // TODO: is there a better way to handle -INFINITY?
-        dst_data[i00] = src[0] == -INFINITY ? -MAXHALF : src[0];
+        dst_data[i00] = src[0];
     }
 }