diff --git a/hopper/tile_size.h b/hopper/tile_size.h
index 555ff06f698..2e437559fc5 100644
--- a/hopper/tile_size.h
+++ b/hopper/tile_size.h
@@ -51,24 +51,43 @@ constexpr std::tuple<int, int, bool, bool> tile_size_fwd_sm90(
             return {128, is_local ? 64 : 80, true, true};  // 128 x 80 hits the limit of smem
         }
     } else {
-        if (headdim <= 64) {
-            if (use_one_mma_wg) {
+        // FP8 path
+        if (use_one_mma_wg) {
+            // Decode tiles — independent of two-level accumulation setting
+            if (headdim <= 96) {
                 return {64, 128, true, true};
             } else {
-                return {192, 160, true, true};
-            }
-        } else if (headdim <= 96) {
-            return {192, 128, true, true};
-        } else if (headdim <= 128) {
-            if (use_one_mma_wg) {
                 return {64, 96, true, true};
-            } else{
-                return {128, paged_kv_non_TMA ? 160 : (v_colmajor || (softcap && is_local) ? 192 : 224), true, true};
             }
-        } else if (headdim <= 192) {
-            return {128, (paged_kv_non_TMA || softcap) && is_local ? 128 : 160, true, true};
         } else {
-            return {128, is_local ? 64 : 128, true, !paged_kv_non_TMA};  // PagedKV uses more registers so we disabled IntraWGOverlap
+            // Prefill tiles — two-level accumulation needs smaller tiles to reduce
+            // register pressure from the separate fp32 accumulator (tOrO_accum).
+            // Currently just optimized for causal case.
+#ifndef FLASHATTENTION_DISABLE_FP8_TWO_LEVEL_ACCUMULATION
+            if (headdim <= 64) {
+                return {192, 128, true, true};
+            } else if (headdim <= 96) {
+                return {128, 128, true, true};
+            } else if (headdim <= 128) {
+                return {128, 192, true, true};
+            } else if (headdim <= 192) {
+                return {128, 96, true, true};
+            } else {
+                return {128, is_local ? 64 : 128, true, !paged_kv_non_TMA};  // TODO: FP8 prefill ~0.54x of BF16 with two-level accum at hd256
+            }
+#else
+            if (headdim <= 64) {
+                return {192, 160, true, true};
+            } else if (headdim <= 96) {
+                return {192, 128, true, true};
+            } else if (headdim <= 128) {
+                return {128, paged_kv_non_TMA ? 160 : (v_colmajor || (softcap && is_local) ? 192 : 224), true, true};
+            } else if (headdim <= 192) {
+                return {128, (paged_kv_non_TMA || softcap) && is_local ? 128 : 160, true, true};
+            } else {
+                return {128, is_local ? 64 : 128, true, !paged_kv_non_TMA};
+            }
+#endif
         }
     }
 }