dlmread — Read numeric data from delimiter-separated text files in MATLAB and RunMat.

dlmread(filename) reads numeric data from delimiter-separated ASCII text files. Custom delimiter, range, and offset forms follow MATLAB and RunMat compatibility semantics.

Syntax

M = dlmread(filename)
M = dlmread(filename, delimiter)
M = dlmread(filename, row, col)
M = dlmread(filename, delimiter, range)
M = dlmread(filename, delimiter, row, col)
M = dlmread(filename, delimiter, row, col, range)

Inputs

Returns

Errors

How dlmread works

• Accepts character vectors or string scalars for filename. String arrays must contain exactly one element.
• The delimiter is optional; when omitted, a comma (,) is used. Supply a character (for example ';'), a string scalar ("\t"), or a numeric ASCII code (9 for tab). Empty delimiter inputs are not allowed.
• dlmread(filename, R, C) and dlmread(filename, delimiter, R, C) start reading at zero-based offsets R and C, skipping any earlier rows or columns.
• dlmread(filename, range) and dlmread(filename, delimiter, range) accept a numeric vector [r1 c1 r2 c2] or [r1 c1] describing a rectangular slice. The indices are zero-based and inclusive. Excel-style addresses such as "B2:D6" are also accepted for compatibility with csvread.
• Empty fields (for example two adjacent delimiters) are interpreted as 0. Tokens such as NaN, Inf, and -Inf are accepted (case-insensitive).
• Any other nonnumeric token raises an error that identifies the offending row and column using one-based indices (matching MATLAB diagnostic messages).
• Results are dense double-precision tensors laid out in column-major order. Empty files yield a 0×0 tensor.
• Leading UTF-8 byte order marks (BOM) are stripped automatically to match MATLAB's handling of spreadsheets that emit BOM-prefixed text files.
• Paths may include ~ to reference the home directory; RunMat expands the token before opening the file.

Does RunMat run dlmread on the GPU?

dlmread performs file I/O and parsing on the CPU. Arguments are gathered from the GPU when needed, and the output tensor lives in host memory. Wrap the result in gpuArray if you need residency on a device. No provider hooks are involved.

GPU memory and residency

dlmread always creates a CPU-resident tensor because the function performs file I/O synchronously on the host. If you need the data on the GPU, call gpuArray(dlmread(...)) or switch to readmatrix with the 'Like' option to direct the result to a device.

Examples

Reading comma-delimited data by default

writematrix([1 2 3; 4 5 6], "samples.csv");
M = dlmread("samples.csv")
delete("samples.csv");

Expected output:

M =
     1     2     3
     4     5     6

Importing semicolon-separated values

fid = fopen("scores.txt", "w");
fprintf(fid, "1;2;3\n4;5;6\n");
fclose(fid);

M = dlmread("scores.txt", ";")
delete("scores.txt");

Expected output:

M =
     1     2     3
     4     5     6

Using tab characters as the delimiter

fid = fopen("tabs.txt", "w");
fprintf(fid, "10\t11\t12\n13\t14\t15\n");
fclose(fid);

M = dlmread("tabs.txt", "\t")
delete("tabs.txt");

Expected output:

M =
    10    11    12
    13    14    15

Skipping a header row and column (zero-based offsets)

fid = fopen("with_header.txt", "w");
fprintf(fid, "Label,Jan,Feb\nalpha,1,2\nbeta,3,4\n");
fclose(fid);

M = dlmread("with_header.txt", ",", 1, 1)
delete("with_header.txt");

Expected output:

M =
     1     2
     3     4

Extracting a rectangular range

fid = fopen("block.txt", "w");
fprintf(fid, "10,11,12,13\n14,15,16,17\n18,19,20,21\n");
fclose(fid);

sub = dlmread("block.txt", ",", [1 1 2 3])
delete("block.txt");

Expected output:

sub =
    15    16    17
    19    20    21

Treating empty fields as zeros

fid = fopen("blanks.txt", "w");
fprintf(fid, "1,,3\n,5,\n7,8,\n");
fclose(fid);

M = dlmread("blanks.txt")
delete("blanks.txt");

Expected output:

M =
     1     0     3
     0     5     0
     7     8     0

Using a numeric ASCII code for the delimiter

fid = fopen("pipe.txt", "w");
fprintf(fid, "5|6|7\n8|9|10\n");
fclose(fid);

M = dlmread("pipe.txt", double('|'))
delete("pipe.txt");

Expected output:

M =
     5     6     7
     8     9    10

Using dlmread with coding agents

Open a RunMat example with live inputs, then ask the agent to explain how dlmread changes the result.

FAQ

Related Io functions

Open-source implementation

Unlike proprietary runtimes, every RunMat function is open-source. Read exactly how dlmread is executed, line by line, in Rust.

• View the source for dlmread in Rust on GitHub
• Learn how the RunMat runtime works
• Found a bug? Open an issue with a minimal reproduction.

About RunMat

RunMat is an open-source runtime that executes MATLAB-syntax code blazing on any GPU. It is licensed under the Apache 2.0 license.

• RunMat automatically optimizes your math for GPU execution on Apple, Nvidia, and AMD hardware. No code changes needed. Simulations that took hours now take minutes.
• Start running code in seconds. RunMat runs in the browser, on the desktop, or from the CLI. No license server, no IT ticket.

Getting started · Benchmarks · Pricing