Differences between Inter IPP and Microsoft v100 Compilers

Recently, two compilers were used in the project. Because it is an accurate algorithm computing, the computing accuracy is very high and float type is used at the same time, the result of Inter is much better than that of. But I don't know why. The test finally found that it was a problem with two compilers. After the test, we found that the float type is valid for 6th bits in the v100 compiler, and 6 bits appear to be random and not so accurate, but in the Inter processor, the float type is also calculated based on the double type, so the final result is much better. Mark it strongly. As an experience, you can remember the existence of this bug when doing computation in the future.

From Weizhi note (wiz)

Differences between Inter IPP and Microsoft v100 Compilers