_ = x[ErrReplicationValidationError-61]
	_ = x[ErrReplicationPermissionCheckError-62]
	_ = x[ErrObjectRestoreAlreadyInProgress-63]
	_ = x[ErrNoSuchKey-64]
	_ = x[ErrNoSuchUpload-65]
	_ = x[ErrInvalidVersionID-66]
	_ = x[ErrNoSuchVersion-67]
	_ = x[ErrNotImplemented-68]
	_ = x[ErrPreconditionFailed-69]
	_ = x[ErrRequestTimeTooSkewed-70]
	_ = x[ErrSignatureDoesNotMatch-71]
	_ = x[ErrMethodNotAllowed-72]
	_ = x[ErrInvalidPart-73]

              }
            ]
          }
        ]
      },
      "gridPos": {
        "h": 6,
        "w": 6,
        "x": 0,
        "y": 18
      },
      "id": 66,
      "options": {
        "legend": {
          "calcs": [],
          "displayMode": "list",
          "placement": "bottom",
          "showLegend": true
        },
        "tooltip": {

          "unit": "short"
        },
        "overrides": []
      },
      "gridPos": {
        "h": 3,
        "w": 3,
        "x": 6,
        "y": 6
      },
      "id": 66,
      "maxDataPoints": 100,
      "options": {
        "colorMode": "value",
        "graphMode": "area",
        "justifyMode": "auto",
        "orientation": "horizontal",
        "reduceOptions": {

    python version via ./configure.

*   Some improvements to GPU performance and memory usage:
    [convnet benchmarks](https://github.com/soumith/convnet-benchmarks/issues/66)
    roughly equivalent with native cudnn v2 performance. Improvements mostly due
    to moving to 32-bit indices, faster shuffling kernels. More improvements to
    come in later releases.

## Bug Fixes