DI, AX ADCQ $0x00, DX MOVQ DX, DI MOVQ AX, 216(CX) // Iteration 28 MOVQ 224(BX), AX MULQ SI ADDQ 224(CX), AX ADCQ $0x00, DX ADDQ DI, AX ADCQ $0x00, DX MOVQ DX, DI MOVQ AX, 224(CX) // Iteration 29 MOVQ 232(BX), AX MULQ SI ADDQ 232(CX), AX ADCQ $0x00, DX ADDQ DI, AX ADCQ $0x00, DX MOVQ DX, DI MOVQ AX, 232(CX) // Iteration 30 MOVQ 240(BX), AX MULQ SI ADDQ 240(CX), AX ADCQ $0x00, DX ADDQ DI, AX ADCQ $0x00, DX MOVQ DX, DI MOVQ AX, 240(CX) // Iteration 31 MOVQ 248(BX), AX MULQ SI ADDQ 248(CX), AX ADCQ $0x00,...

pkg syscall (darwin-arm64), const SYS_WAIT4 ideal-int
pkg syscall (darwin-arm64), const SYS_WAIT4_NOCANCEL = 400
pkg syscall (darwin-arm64), const SYS_WAIT4_NOCANCEL ideal-int
pkg syscall (darwin-arm64), const SYS_WAITEVENT = 232
pkg syscall (darwin-arm64), const SYS_WAITEVENT ideal-int
pkg syscall (darwin-arm64), const SYS_WAITID = 173
pkg syscall (darwin-arm64), const SYS_WAITID ideal-int
pkg syscall (darwin-arm64), const SYS_WAITID_NOCANCEL = 416

*   Enable JIT-compiled i64-indexed kernels on GPU for large tensors with more than 2**32 elements.
    *   Unary GPU kernels: Abs, Atanh, Acos, Acosh, Asin, Asinh, Atan, Cos, Cosh, Sin, Sinh, Tan, Tanh.

DI, AX ADCQ $0x00, DX MOVQ DX, DI MOVQ AX, 216(CX) // Iteration 28 MOVQ 224(BX), AX MULQ SI ADDQ 224(CX), AX ADCQ $0x00, DX ADDQ DI, AX ADCQ $0x00, DX MOVQ DX, DI MOVQ AX, 224(CX) // Iteration 29 MOVQ 232(BX), AX MULQ SI ADDQ 232(CX), AX ADCQ $0x00, DX ADDQ DI, AX ADCQ $0x00, DX MOVQ DX, DI MOVQ AX, 232(CX) // Iteration 30 MOVQ 240(BX), AX MULQ SI ADDQ 240(CX), AX ADCQ $0x00, DX ADDQ DI, AX ADCQ $0x00, DX MOVQ DX, DI MOVQ AX, 240(CX) // Iteration 31 MOVQ 248(BX), AX MULQ SI ADDQ 248(CX), AX ADCQ $0x00,...

* Separate development scripts `test.sh`, `lint.sh`, and `format.sh`. PR [#232](https://github.com/tiangolo/fastapi/pull/232).

* Re-enable `black` formatting checks for Python 3.7. PR [#229](https://github.com/tiangolo/fastapi/pull/229) by [@zamiramir](https://github.com/zamiramir).

## 0.21.0