diff --git a/config/gpu/h100.toml b/config/gpu/h100.toml
index 9ca38f725..6c63ca852 100644
--- a/config/gpu/h100.toml
+++ b/config/gpu/h100.toml
@@ -4,7 +4,7 @@ batch_size = 8192
 num_atoms = 52
 buffer_size = 500000
 hidden_dim_base = 256
-replay_buffer_vram_fraction = 0.45  # state_dim=96, 86 tensors + IQN 8.6GB + attention 250MB + Phase 2 128MB
+replay_buffer_vram_fraction = 0.0  # disabled — use exact buffer_size=500K (was 0.45 → 15.8M entries = slow PER prefix scan in graph)
 
 [experience]
 gpu_timesteps_per_episode = 5000