Why ZeRO-2 use more CUDA Memory than ZeRO-1?

Follow the bing_bert tutorial, my deepspeed_config is:

{
  "train_batch_size": 4096,
  "train_micro_batch_size_per_gpu": 32,
  "steps_per_print": 1000,
  "prescale_gradients": false,
  "optimizer": {
    "type": "Adam",
    "params": {
      "lr": 6e-3,
      "betas": [
        0.9,
        0.99
      ],
      "eps": 1e-8,
      "weight_decay": 0.01
    }
  },

  "zero_optimization": {
    "stage": 1,
    "allgather_partitions": true,
    "allgather_bucket_size": 5e8,
    "overlap_comm": false,
    "reduce_scatter": true,
    "reduce_bucket_size": 5e8,
    "contiguous_gradients": true,
    "grad_hooks": true,
    "round_robin_gradients": false
  },


  "scheduler": {
    "type": "WarmupLR",
    "params": {
        "warmup_min_lr": 1e-8,
        "warmup_max_lr": 6e-3
    }
  },
  "gradient_clipping": 1.0,

  "wall_clock_breakdown": false,

  "fp16": {
    "enabled": true,
    "loss_scale": 0
  },
  "sparse_attention": {
    "mode": "fixed",
    "block": 16,
    "different_layout_per_head": true,
    "num_local_blocks": 4,
    "num_global_blocks": 1,
    "attention": "bidirectional",
    "horizontal_global_attention": false,
    "num_different_global_patterns": 4
  }
}

The CUDA Memory usage for stage 1 is 8900MB per GPU
The CUDA Memory usage for stage 2 is 9600MB per GPU

And the ZeRO-2 is much slower than ZeRO-1 in training speed.

Any help will be appreciate~

Start by reproducing the bing_bert tutorial with the posted DeepSpeed configuration, comparing ZeRO stage 1 and stage 2 memory use and training speed. Investigate the configuration and runtime behavior behind the difference; done means providing a supported explanation for both measurements and identifying any configuration-related cause.