Almost all modern uses follow the IEEE 754-2008 standard, where the 16-bit base-2 format is referred to as binary16, and the exponent uses 5 bits. This can express values in the range ±65,504, with the minimum value above 1 being 1 + 1/1024.
Depending on the computer, half-precision can be over an order of magnitude faster than double precision, e.g. 550 PFLOPS for half-precision vs 37 PFLOPS for double precision on one cloud provider.[1]
Several earlier 16-bit floating point formats have existed including that of Hitachi's HD61810 DSP of 1982 (a 4-bit exponent and a 12-bit mantissa),[2] Thomas J. Scott's WIF of 1991 (5 exponent bits, 10 mantissa bits)[3] and the 3dfx Voodoo Graphics processor of 1995 (same as Hitachi).[4]
ILM was searching for an image format that could handle a wide dynamic range, but without the hard drive and memory cost of single or double precision floating point.[5] The hardware-accelerated programmable shading group led by John Airey at SGI (Silicon Graphics) used the s10e5 data type in 1997 as part of the 'bali' design effort. This is described in a SIGGRAPH 2000 paper[6] (see section 4.3) and further documented in US patent 7518615.[7] It was popularized by its use in the open-source OpenEXR image format.
Nvidia and Microsoft defined the halfdatatype in the Cg language, released in early 2002, and implemented it in silicon in the GeForce FX, released in late 2002.[8] However, hardware support for accelerated 16-bit floating point was later dropped by Nvidia before being reintroduced in the Tegra X1 mobile GPU in 2015.
The F16C extension in 2012 allows x86 processors to convert half-precision floats to and from single-precision floats with a machine instruction.
The format is assumed to have an implicit lead bit with value 1 unless the exponent field is stored with all zeros. Thus, only 10 bits of the significand appear in the memory format but the total precision is 11 bits. In IEEE 754 parlance, there are 10 bits of significand, but there are 11 bits of significand precision (log10(211) ≈ 3.311 decimal digits, or 4 digits ± slightly less than 5 units in the last place).
Exponent encoding
The half-precision binary floating-point exponent is encoded using an offset-binary representation, with the zero offset being 15; also known as exponent bias in the IEEE 754 standard.[9]
Thus, as defined by the offset binary representation, in order to get the true exponent the offset of 15 has to be subtracted from the stored exponent.
The stored exponents 000002 and 111112 are interpreted specially.
The minimum strictly positive (subnormal) value is
2−24 ≈ 5.96 × 10−8.
The minimum positive normal value is 2−14 ≈ 6.10 × 10−5.
The maximum representable value is (2−2−10) × 215 = 65504.
Half precision examples
These examples are given in bit representation
of the floating-point value. This includes the sign bit, (biased) exponent, and significand.
Binary
Hex
Value
Notes
0 00000 0000000000
0000
0
0 00000 0000000001
0001
2−14 × (0 + 1/1024 ) ≈ 0.000000059604645
smallest positive subnormal number
0 00000 1111111111
03ff
2−14 × (0 + 1023/1024 ) ≈ 0.000060975552
largest subnormal number
0 00001 0000000000
0400
2−14 × (1 + 0/1024 ) ≈ 0.00006103515625
smallest positive normal number
0 01101 0101010101
3555
2−2 × (1 + 341/1024 ) ≈ 0.33325195
nearest value to 1/3
0 01110 1111111111
3bff
2−1 × (1 + 1023/1024 ) ≈ 0.99951172
largest number less than one
0 01111 0000000000
3c00
20 × (1 + 0/1024 ) = 1
one
0 01111 0000000001
3c01
20 × (1 + 1/1024 ) ≈ 1.00097656
smallest number larger than one
0 11110 1111111111
7bff
215 × (1 + 1023/1024 ) = 65504
largest normal number
0 11111 0000000000
7c00
∞
infinity
1 00000 0000000000
8000
−0
1 10000 0000000000
c000
(−1)1 × 21 × (1 + 0/1024 ) = −2
1 11111 0000000000
fc00
−∞
negative infinity
By default, 1/3 rounds down like for double precision, because of the odd number of bits in the significand. The bits beyond the rounding point are 0101... which is less than 1/2 of a unit in the last place.
Precision limitations
Min
Max
interval
0
2−13
2−24
2−13
2−12
2−23
2−12
2−11
2−22
2−11
2−10
2−21
2−10
2−9
2−20
2−9
2−8
2−19
2−8
2−7
2−18
2−7
2−6
2−17
2−6
2−5
2−16
2−5
2−4
2−15
2−4
1/8
2−14
1/8
1/4
2−13
1/4
1/2
2−12
1/2
1
2−11
1
2
2−10
2
4
2−9
4
8
2−8
8
16
2−7
16
32
2−6
32
64
2−5
64
128
2−4
128
256
1/8
256
512
1/4
512
1024
1/2
1024
2048
1
2048
4096
2
4096
8192
4
8192
16384
8
16384
32768
16
32768
65520
32
65520
∞
∞
65520 and larger numbers round to infinity. This is for round-to-even; other rounding strategies will change this cut-off.
ARM alternative half-precision
ARM processors support (via a floating-point control register bit) an "alternative half-precision" format, which does away with the special case for an exponent value of 31 (111112).[10] It is almost identical to the IEEE format, but there is no encoding for infinity or NaNs; instead, an exponent of 31 encodes normalized numbers in the range 65536 to 131008.
Uses of half precision
Half precision is used in several computer graphics environments to store pixels, including MATLAB, OpenEXR, JPEG XR, GIMP, OpenGL, Vulkan,[11]Cg, Direct3D, and D3DX. The advantage over 8-bit or 16-bit integers is that the increased dynamic range allows for more detail to be preserved in highlights and shadows for images, and avoids gamma correction. The advantage over 32-bit single-precision floating point is that it requires half the storage and bandwidth (at the expense of precision and range).[5]
Half precision can be useful for mesh quantization. Mesh data is usually stored using 32-bit single-precision floats for the vertices, however in some situations it is acceptable to reduce the precision to only 16-bit half-precision, requiring only half the storage at the expense of some precision. Mesh quantization can also be done with 8-bit or 16-bit fixed precision depending on the requirements.[12]
Hardware and software for machine learning or neural networks tend to use half precision: such applications usually do a large amount of calculation, but don't require a high level of precision. Due to hardware typically not supporting 16-bit half-precision floats, neural networks often use the bfloat16 format, which is the single precision float format truncated to 16 bits.
If the hardware has instructions to compute half-precision math, it is often faster than single or double precision. If the system has SIMD instructions that can handle multiple floating-point numbers within one instruction, half precision can be twice as fast by operating on twice as many numbers simultaneously.[13]
Support by programming languages
Zig provides support for half precisions with its f16 type.[14]
.NET 5 introduced half precision floating point numbers with the System.Half standard library type.[15][16] As of January 2024[update], no .NET language (C#, F#, Visual Basic, and C++/CLI and C++/CX) has literals (e.g. in C#, 1.0f has type System.Single or 1.0m has type System.Decimal) or a keyword for the type.[17][18][19]
Swift introduced half-precision floating point numbers in Swift 5.3 with the Float16 type.[20]
OpenCL also supports half-precision floating point numbers with the half datatype on IEEE 754-2008 half-precision storage format.[21]
As of 2024[update], Rust is currently working on adding a new f16 type for IEEE half-precision 16-bit floats.[22]
Julia provides support for half-precision floating point numbers with the Float16 type.[23]
C++ introduced half-precision since C++23 with the std::float16_t type.[24]GCC already implements support for it.[25]
Support for half precision in the x86instruction set is specified in the F16C instruction set extension, first introduced in 2009 by AMD and fairly broadly adopted by AMD and Intel CPUs by 2012. This was further extended up the AVX-512_FP16 instruction set extension implemented in the Intel Sapphire Rapids processor.[27]
On RISC-V, the Zfh and Zfhminextensions provide hardware support for 16-bit half precision floats. The Zfhmin extension is a minimal alternative to Zfh.[28]
On Power ISA, VSX and the not-yet-approved SVP64 extension provide hardware support for 16-bit half-precision floats as of PowerISA v3.1B and later.[29][30]
Support for half precision on IBM Z is part of the Neural-network-processing-assist facility that IBM introduced with Telum. IBM refers to half precison floating point data as NNP-Data-Type 1 (16-bit).
^Scott, Thomas J. (March 1991). "Mathematics and computer science at odds over real numbers". Proceedings of the twenty-second SIGCSE technical symposium on Computer science education - SIGCSE '91. Vol. 23. pp. 130–139. doi:10.1145/107004.107029. ISBN0897913779. S2CID16648394.
^Ho, Nhut-Minh; Wong, Weng-Fai (September 1, 2017). "Exploiting half precision arithmetic in Nvidia GPUs"(PDF). Department of Computer Science, National University of Singapore. Retrieved July 13, 2020. Nvidia recently introduced native half precision floating point support (FP16) into their Pascal GPUs. This was mainly motivated by the possibility that this will speed up data intensive and error tolerant applications in GPUs.
