diff --git a/tests/unit/test_d1_smem128.py b/tests/unit/test_d1_smem128.py
index b83f4061..29603784 100644
--- a/tests/unit/test_d1_smem128.py
+++ b/tests/unit/test_d1_smem128.py
@@ -1,4 +1,4 @@
-"""D1: Test SMEM-P at hd=128 with corrected coordinate store."""
+"""D1: Test SMEM-P at hd=128 via gP→TMA path."""
 import torch, math
 import cutlass.cute as cute
 import cutlass.torch as ct
@@ -6,7 +6,7 @@ import cuda.bindings.driver as cuda
 from dsv4.kernels.attention.fmha import FmhaKernel
 
 
-def test_smem_p(hd, n_kv=128):
+def test_smem_p_gP(hd, n_kv=128):
     m = 128
     torch.manual_seed(42)
     q = torch.randn(m, hd, 1, dtype=torch.bfloat16, device='cuda')
@@ -20,6 +20,7 @@ def test_smem_p(hd, n_kv=128):
     attn_exp = torch.exp(qf @ kf.T * scale - attn_max)
     attn_sum = attn_exp.sum(dim=-1, keepdim=True)
     ref_unnorm = attn_exp @ v.float()
+    ref_lse = (torch.log(attn_sum.squeeze(-1)) + attn_max.squeeze(-1))[0].item()
 
     lse_tensor = torch.zeros(m, 1, 1, dtype=torch.float32, device='cuda')
     kernel = FmhaKernel(head_dim=hd, s_k=n_kv, use_smem_p=True)
@@ -28,22 +29,32 @@ def test_smem_p(hd, n_kv=128):
 
     v_tile = v[:, 0:pv_n_tile].contiguous().unsqueeze(-1)
     c_tile = torch.zeros(m, pv_n_tile, 1, dtype=torch.bfloat16, device='cuda')
+    # gP: global buffer for P matrix (128 x s_k), partitioned by QK C-fragment
+    gP = torch.zeros(m, n_kv, 1, dtype=torch.bfloat16, device='cuda')
+
     mQ = ct.from_dlpack(q).mark_layout_dynamic(leading_dim=ct.get_leading_dim(q))
     mK = ct.from_dlpack(k).mark_layout_dynamic(leading_dim=ct.get_leading_dim(k))
     mV = ct.from_dlpack(v_tile).mark_layout_dynamic(leading_dim=ct.get_leading_dim(v_tile))
     mC = ct.from_dlpack(c_tile).mark_layout_dynamic(leading_dim=ct.get_leading_dim(c_tile))
     mLSE = ct.from_dlpack(lse_tensor).mark_layout_dynamic(leading_dim=ct.get_leading_dim(lse_tensor))
+    mGP = ct.from_dlpack(gP).mark_layout_dynamic(leading_dim=ct.get_leading_dim(gP))
 
-    print(f'hd={hd} SMEM-P: Compiling...', flush=True)
-    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream, mLSE)
-    compiled(mQ, mK, mV, mC, stream, mLSE)
+    print(f'hd={hd} SMEM-P (gP→TMA): Compiling...', flush=True)
+    compiled = cute.compile(kernel, mQ, mK, mV, mC, stream, mLSE, mGP)
+    compiled(mQ, mK, mV, mC, stream, mLSE, mGP)
     torch.cuda.synchronize()
 
     out = c_tile[:, :, 0].float()
     cos = torch.nn.functional.cosine_similarity(out.flatten().unsqueeze(0), ref_unnorm.flatten().unsqueeze(0)).item()
-    print(f'hd={hd} SMEM-P: cos={cos:.6f}  {"PASS" if cos >= 0.99 else "FAIL"}')
+    kernel_lse = lse_tensor[0, 0, 0].item()
+    lse_err = abs(kernel_lse - ref_lse)
+    print(f'hd={hd} SMEM-P (gP→TMA): cos={cos:.6f}  lse_err={lse_err:.6f}  {"PASS" if cos >= 0.99 else "FAIL"}')
+    if cos < 0.99:
+        print(f'  out[0,:4]={out[0,:4].tolist()}')
+        print(f'  ref[0,:4]={ref_unnorm[0,:4].tolist()}')
     return cos
 
 
 if __name__ == '__main__':
-    test_smem_p(128)
+    print("=== SMEM-P via gP→TMA ===\n")
+    test_smem_p_gP(128)