GetDataBlockSizeInBytes
Applicability
Product |
Supported |
|---|---|
Atlas 350 Accelerator Card |
√ |
√ |
|
√ |
|
x |
|
√ |
|
x |
|
√ |
Function Usage
Obtains the size (in byte) of a data block of the current chip version. You can compute the values of parameters, such as repeatTime, Data Block Stride, and Repeat Stride, to be passed in the API command based on the data block size.
Prototype
1 | __aicore__ inline constexpr int16_t GetDataBlockSizeInBytes() |
Parameters
None
Returns
Size (in byte) of a data block of the current chip version
Constraints
None
Example
In the following example, the value of repeatTime is computed based on the data block value obtained through GetDataBlockSizeInBytes.
1 2 3 4 5 6 7 8 | int16_t dataBlockSize = AscendC::GetDataBlockSizeInBytes(); // Each repeat involves eight data blocks. 8 * dataBlockSize/sizeof(half) can be calculated. mask is configured to enable all elements in the iteration to participate in the calculation. uint64_t mask = 8 * dataBlockSize / sizeof(half); // Divide a total of 512 elements by the number of elements involved in each repeat to obtain the value of repeatTime. uint8_t repeatTime = 512 / mask; // dstBlkStride, src0BlkStride, src1BlkStride = 1. Data is continuously read and written in a single iteration. // dstRepStride, src0RepStride, src1RepStride = 8. Data is continuously read and written between adjacent iterations. AscendC::Add(dstLocal, src0Local, src1Local, mask, repeatTime, { 1, 1, 1, 8, 8, 8 }); |
Parent topic: TPosition