[object Object]

Note: This API will be deprecated in later versions. Use aclnnWeightQuantMatmulAllReduce and aclnnAddRmsNorm instead.

[object Object][object Object]undefined

Note: When using this API, ensure that the driver firmware package and CANN package are in the 8.0.RC2 version or later. Otherwise, an error, such as BUS ERROR, will be reported.

[object Object]

• Operator function: Performs the MatMul, AllReduce, addition, and RMSNorm computation in sequence.
• Formula:$$mm\_out = allReduce(x1 @ (x2*antiquantScale + antiquantOffset) + bias)$$ $$y = mm\_out + residual$$ $$normOut = \frac{y}{RMS(y)} * gamma, RMS(y) = \sqrt{\frac{1}{d} \sum_{i=1}^{d} y_{i}^{2} + epsilon}$$

[object Object]

• [object Object] and [object Object] implement the same function in different ways. Select a proper operator based on your requirements.

  • [object Object]: Output tensor objects [object Object] and [object Object] need to be created to store the computation result.
  • [object Object]: Output tensor [object Object] needs to be created. The results that would have been stored in the output tensor [object Object] in the original non-inplace scenario are directly written to the memory of the input tensor [object Object].

• Each operator has calls. First, [object Object] is called to obtain the workspace size required for computation and the executor that contains the operator computation flow. Then, [object Object] is called to perform computation.

[object Object]

• Parameters

  [object Object]

• Returns:

  [object Object]: status code. For details, see .

  The first-phase API implements input parameter verification. The following errors may be thrown.

  [object Object]

[object Object]

• Parameters

  [object Object]

• Returns:

  [object Object]: status code. For details, see .

[object Object]

• Deterministic computation:

  • By default, aclnnWeightQuantMatmulAllReduceAddRmsNorm is implemented in non-deterministic mode. You can enable deterministic computing by setting the HCCL_DETERMINISTIC environment variable to true.

• The application scenario is the same as that of [object Object]. MC2 is disabled in incremental scenarios but enabled in full scenarios.

• The input [object Object] can be two-dimensional or three-dimensional, with shape (b, s, k) or (s, k), respectively. x2 must be two-dimensional, and its shape is (k, n). The axis meets the input parameter requirements of the MM operator. The k axis is the same, and the ranges of b*s and s are [1, 2147483647], and the ranges of k and n are [1, 65535]. If [object Object] is not empty, it must be one-dimensional, with shape (n).

• The input [object Object] must be three-dimensional, with its shape being (b, s, n). When [object Object] is two-dimensional, (b*s) of [object Object] is equal to s of [object Object]. The input [object Object] must be one-dimensional, with shape (n).

• In the pertensor scenario, the shape of antiquantScale is (1); in the perchannel scenario, the shape is (1, n)/(n); in the pergroup scenario, the shape is (ceil(k, antiquantGroupSize), n). If antiquantOffset is not empty, its shape is the same as that of antiquantScale.

• The dimensions and data types of the outputs [object Object] and [object Object] are the same as those of [object Object]. If [object Object] is not empty, its shape is equal to the last dimension of [object Object].

• The data type of [object Object] must be int8 or int4. The data types of the inputs [object Object], [object Object], [object Object], [object Object], [object Object], and [object Object] must be the same.

• The [object Object] matrix cannot be transposed. The [object Object] matrix can be transposed or not transposed.

• The value of [object Object] falls in the range [32, min(k-1, INT_MAX)] and must be a multiple of 32.

• 1, 2, 4, and 8 ranks are supported, and only the all-mesh networking of HCCS links is supported.

• Empty tensors with (b*s) and n being 0 are supported. Empty tensors with k being 0 are not supported.

• [object Object]Atlas A2 training products/Atlas A2 inference products[object Object]: Only one communication domain for MC2 operators within a model is supported.

[object Object]

The following example is for reference only. For details, see .