Skip to content

nxtpow2

Returns a power of two at least as large as the integer part of its input, with a minimum result of 2.

Use nxtpow2 when choosing a buffer or function-table size that must be a power of two. It returns a number and does not allocate storage.

Syntax

isize = nxtpow2(ivalue)

isize nxtpow2 ivalue

Initialization

ivalue is the number to convert. The opcode runs at initialization only.

isize is the smallest power of two that is at least 2 and at least the input's integer part. An exact power of two stays unchanged, except that 1 becomes 2.

The fractional part truncates toward zero before the calculation. For example, 8.9 becomes 8, so the result is 8. If the result must cover the full input value, use nxtpow2(ceil(ivalue)).

Input Result
-5.8 2
0 2
1 2
2 2
3 4
8 8
8.9 8
9 16
1500000000 2147483648

Input limits

The input must be finite, at least -2147483648 and less than 2147483648. Values outside that range, infinity and NaN cause an initialization error. The range check happens before truncation.

The largest possible result is 2147483648, or 2 to the power of 31. That value is a valid result but is outside the input range. The minimum result is 2, including for valid negative inputs and zero.

Csound builds with 32-bit samples cannot represent every large integer. For example, 2147483647 rounds to 2147483648 and fails the input check in those builds. floatsize reports the sample type's size in bytes.

A result within these limits may still exceed the size supported by a table, FFT or available memory. Check the requirements of the opcode that will use it.

Examples

The example chooses a buffer size for a duration in seconds. It first rounds the sample count up with ceil, then rounds to a power of two. At 48000 Hz, 0.1 seconds needs 4800 samples and uses an 8192-element buffer. A quarter of a second needs 12000 samples and uses 16384 elements.

It uses nxtpow2.csd.

Choose a buffer size from a duration
 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
<CsoundSynthesizer>
<CsOptions>
-n -d -m0
</CsOptions>
<CsInstruments>
sr = 48000
ksmps = 32
nchnls = 1
0dbfs = 1

instr ChooseBufferSize
  iDuration = p4

  // Round up so the buffer can hold the full duration.
  iSamples = ceil(iDuration * sr)
  iSize = nxtpow2(iSamples)

  // Allocate a zero-filled buffer with the chosen power-of-two length.
  iBuffer[] init iSize
  prints("%g seconds needs %.0f samples. Buffer length is %.0f.\n", \
         iDuration, iSamples, lenarray(iBuffer))
endin
</CsInstruments>
<CsScore>
i "ChooseBufferSize" 0   0.01 0.1
i "ChooseBufferSize" 0.1 0.01 0.25
e
</CsScore>
</CsoundSynthesizer>

See also

ceil, int, powoftwo, ftgen, floatsize, Mathematical Functions

Credits

Author Victor Lazzarini, 2016.

New in Csound 6.08.