clean up

flash_attn/flash_attn_triton_interface_amd.py

@@ -2,10 +2,8 @@
 import torch
 import triton
 
-# /////////////////////////////////////////// Interface //////////////////////////////////////////////////////////
 DEBUG=False
 
-
 def fwd(q,
         k,
         v,
@@ -35,32 +33,34 @@ def fwd(q,
     if dropout_p != 0.0:
         raise ValueError("dropout is not supported on HIP")
 
-
     if o is None:
         o = torch.empty_like(q)
 
+    # Setup metadata
+    input_metadata = MetaData(sm_scale=softmax_scale)
+    input_metadata.max_seqlens_q = q.shape[1]
+    input_metadata.max_seqlens_k = k.shape[1]
+    input_metadata.layout = "bshd"
 
-    # Create metadata object
-    metadata = MetaData(sm_scale=softmax_scale)
-    metadata.max_seqlens_q = q.shape[1]
-    metadata.max_seqlens_k = k.shape[1]
-    metadata.layout = "bshd"
+    batch, nheads_q, nheads_k, head_size = get_shape_from_layout(q, k, input_metadata)
 
-    # Setup metadata
     if causal:
-        metadata.need_causal()
+        input_metadata.need_causal()
+
     # if bias is not None:
-    #     metadata.need_bias(bias, q.shape[0], q.shape[1], q.shape[2], k.shape[2])
+    #     input_metadata.need_bias(bias, batch, nheads_q, input_metadata.max_seqlens_q, input_metadata.max_seqlens_k)
+
     if alibi_slopes is not None:
-        metadata.need_alibi(alibi_slopes, q.shape[0], q.shape[2])
+        input_metadata.need_alibi(alibi_slopes, batch, nheads_q)
+
     if dropout_p > 0.0:
-        metadata.need_dropout(dropout_p, return_softmax)
+        input_metadata.need_dropout(dropout_p, return_softmax)
 
     # Check arguments
-    metadata.check_args(q, k, v, o)
+    input_metadata.check_args(q, k, v, o)
 
     # Perform the forward attention computation
-    tri_out, encoded_softmax = attention(q, k, v, o, metadata)
+    tri_out, encoded_softmax = attention(q, k, v, o, input_metadata)
 
     softmax_lse = encoded_softmax
     softmax_p = encoded_softmax
@@ -96,28 +96,26 @@ def varlen_fwd(
 
     if dropout_p != 0.0:
         raise ValueError("dropout is not supported on HIP")
-
-
 
     if o is None:
         o = torch.empty_like(q)
 
-
-
-    # create metadata object
+    # Setup metadata
     input_metadata = MetaData(sm_scale=softmax_scale)
     input_metadata.set_varlen_params(cu_seqlens_q, cu_seqlens_k)
 
     # get shapes
     batch, nheads_q, nheads_k, head_size = get_shape_from_layout(q, k, input_metadata)
 
-    # Setup metadata
     if causal:
         input_metadata.need_causal()
+
     # if bias is not None:
-    #     metadata.need_bias(bias, q.shape[0], q.shape[1], q.shape[2], k.shape[2])
+    #     input_metadata.need_bias(bias, batch, nheads_q, q.shape[2], k.shape[2])
+
     if alibi_slopes is not None:
         input_metadata.need_alibi(alibi_slopes, batch, nheads_q)
+
     if dropout_p > 0.0:
         input_metadata.need_dropout(dropout_p, return_softmax)
 
@@ -132,7 +130,25 @@ def varlen_fwd(
 
     return tri_out, q , k , v, o, softmax_lse, softmax_p, torch.get_rng_state()
 
-def fwd_kvcache(*args, **kwargs):
+def fwd_kvcache( 
+        q,
+        k_cache,
+        v_cache,
+        k,
+        v,
+        cache_seqlens,
+        rotary_cos,
+        rotary_sin,
+        cache_batch_idx,
+        block_table,
+        alibi_slopes,
+        out,
+        softmax_scale,
+        causal,
+        window_size_left,
+        window_size_right,
+        rotary_interleaved,
+        num_splits):
     pass
 
 
@@ -157,12 +173,9 @@ def bwd(dout, q, k, v, out, softmax_lse, dq, dk, dv, alibi_slopes, dropout_p, so
         print("gen_:", gen_)
         print("rng_state:", rng_state)
 
-
-
     if out is None:
         out = torch.empty_like(q)
 
-
     # Ensure the tensors have requires_grad=True
     q.requires_grad_()
     k.requires_grad_()
@@ -187,25 +200,20 @@ def bwd(dout, q, k, v, out, softmax_lse, dq, dk, dv, alibi_slopes, dropout_p, so
     # Setup metadata
     if causal:
         metadata.need_causal()
+
     # if bias is not None:
     #     metadata.need_bias(bias, q.shape[0], q.shape[1], q.shape[2], k.shape[2])
 
     return_softmax = True
     if alibi_slopes is not None:
         metadata.need_alibi(alibi_slopes, batch, nheads_q)
+
     if dropout_p > 0.0:
         metadata.need_dropout(dropout_p, return_softmax)
 
     # Check arguments
     metadata.check_args(q, k, v, out)
 
-
-
-    # tri_out, _ = attention(q, k, v, out, metadata)
-    # tri_out.requires_grad_()
-    # dout.requires_grad_()
-    # tri_out.backward(dout)
-
     # write your own version backward
     M = torch.empty((batch, nheads_q, metadata.max_seqlens_q), device=q.device, dtype=torch.float32) # this passed from