Skip to content

Understanding GCC's Target Specific Options

Jump to bottom
Yuriy Kolerov edited this page Jul 13, 2023 · 16 revisions

Overview

In addition to the common options, GCC also supports a set of -m<value> target specific options. Each of them configures a particular feature or an extension. For example, -mno-code-density option disables generating code density instructions.

-mcpu=<core> selects a particular ISA and CPU family and enables/disables a set of other -m options. Each -mcpu value leads to linking with a prebuilt library which corresponds to this particular -mcpu=<value> and is built using this -mcpu=<value>.

For example, option combination -mcpu=em4 -mno-code-density will generate code that doesn't use code density instructions, however it will be linked with standard library that has been built with just -mcpu=em4, which uses code density instructions - therefore a final application still may use code density instructions.

Table of Target Options

There are several configurations files in GCC's source tree which describe target options for all targets. All of them may be found in gcc/config/arc directory:

Here is a comparison table for all ARCv1 and ARCv2 CPU families.

Options ARC HS ARC EM ARC 700 ARC 600
-mbarrel-shifter Default Yes Default Yes
-mnorm Default Yes Default Yes
-mswap Default Yes Default Yes
-rf16 Yes Yes Yes Yes
-mfpu=<...> Yes Yes — —
-mmpy-option=<...> Yes Yes — —
-mcode-density Default Yes — —
-mdiv-rem Yes Yes — —
-matomic Default — Yes —
-mll64 Yes — — —
-mspfp — Yes Yes Yes
-mdpfp — Yes Yes Yes
-msimd — Yes Yes —
-mmargonaut — — Yes Yes
-mea — — Yes —
-mmul64 — — — Yes
-mmul32x16 — — — Yes
-munaligned-access

You can find a short description for all target options using --target-help option. For example:

$ arc-elf32-gcc --target-help
...
  -mbranch-index              Enable use of BI/BIH instructions when available.
  -mcase-vector-pcrel         Use pc-relative switch case tables - this enables case table shortening.
  -mcmem                      Enable use of NPS400 xld/xst extension.
  -mcode-density              Enable code density instructions for ARCv2.
  -mcode-density-frame        Enable ENTER_S and LEAVE_S opcodes for ARCv2.
  -mcompact-casesi            Enable compact casesi pattern.  Uses of this option are diagnosed.
  -mcpu=CPU                   Compile code for ARC variant CPU.
  -mcrc                       Enable variable polynomial CRC extension.  Uses of this option are diagnosed.
  -mdiv-rem                   Enable DIV-REM instructions for ARCv2.
  -mdpfp                      FPX: Generate Double Precision FPX (compact) instructions.
  -mdpfp-compact              FPX: Generate Double Precision FPX (compact) instructions.
...

You can find which target options are enabled or disabled by a particular -mcpu=<...> using -Q options. For example:

$ arc-elf32-gcc -Q --target-help -mcpu=archs
...
  -mbranch-index                        [disabled]
  -mcase-vector-pcrel                   [disabled]
  -mcmem                                [disabled]
  -mcode-density                        [enabled]
  -mcode-density-frame                  [disabled]
  -mcompact-casesi                      [disabled]
  -mcpu=CPU                             archs
  -mcrc                                 [disabled]
  -mdiv-rem                             [enabled]
  -mdpfp                                [disabled]
  -mdpfp-compact                        [disabled]
...

Floating Point Options for ARC EM, HS3x and HS4x Families

Here is a table which matches -mfpu= values to corresponding features for ARC EM and ARC HS as they defined in ARChitect.

-mfpu= Families -has_fpu -fpu_dp_option -fpu_div_option -fpu_fma_option -fpu_dp_assist
fpus EM, HS On — — — —
fpus_div EM, HS On — On — —
fpus_fma EM, HS On — — On —
fpus_all EM, HS On — On On —
fpud HS On On — — —
fpud_div HS On On On — —
fpud_fma HS On On — On —
fpud_all HS On On On On —
fpuda EM On — — — On
fpuda_div EM On — On — On
fpuda_fma EM On — — On On
fpuda_all EM On — On On On

Multiplication Options for ARC EM, HS3x and HS4x Families

Here is a table which matches -mmpy-option= values to corresponding features for ARC EM and ARC HS as they defined in ARChitect.

-mmpy-option= Aliases ARChitect option for ARC HS ARChitect option for ARC EM
0 none -mpy_option=none -mpy_option=none
1 w — —
2 wlh1, mpy -mpy_option=mpy -mpy_option=wlh1
3 wlh2 — -mpy_option=wlh2
4 wlh3 — -mpy_option=wlh3
5 wlh4 — -mpy_option=wlh4
6 wlh5 — -mpy_option=wlh5
7 plus_dmpy -mpy_option=plus_dmpy —
8 plus_macd -mpy_option=plus_macd —
9 plus_qmacw -mpy_option=plus_qmacw —

Values of -mcpu for ARC HS5x and ARC HS6x Families

-mcpu= -msimd -m128 -mll64
hs5x Y N/A
hs58 Y N/A Y
hs6x Y N/A
hs68 Y Y N/A

Values of -mcpu for ARC HS3x and HS4x Families

-mcpu= -mdiv-rem -matomic -mll64 -mmpy-option= -mfpu=
hs Y
hs34 Y mpy
archs Y Y Y mpy
hs38 Y Y Y plus_qmacw
hs4x Y Y Y plus_qmacw
hs4xd Y Y Y plus_qmacw
hs38_linux Y Y Y plus_qmacw fpud_all

The above -mcpu values correspond to specific ARC EM Processor templates presented in the ARChitect tool.

  • -mcpu=hs corresponds to a basic ARC HS with only atomic instructions enabled. It corresponds to the following ARC HS templates in ARChitect: hs34_base, hs36_base and hs38_base.
  • -mcpu=hs34 is like hs but with with additional support for standard hardware multiplier. It corresponds to the following ARC HS templates in ARChitect: hs34, hs36 and hs38.
  • -mcpu=archs is a generic configuration, which corresponds to the default configuration in older GNU toolchain versions.
  • -mcpu=hs38 is a full-featured ARC HS. It corresponds to the following ARC HS template in ARChitect: hs38_full.
  • -mcpu=hs4x and -mcpu=hs4xd have the same option set as -mcpu=hs38, but compiler will optimize instruction scheduling for specified processors.
  • -mcpu=hs38_linux is a full-featured ARC HS with additional support of double-precision FPU.

Values of -mcpu for ARC EM Family

-mcpu -mcode-density -mnorm -mswap -mbarrel-shifter -mdiv-rem -mmpy-option= -mfpu= -mrf16 -mspfp -mdpfp
em
em_mini Y
em4 Y
arcem Y Y wlh1
em4_dmips Y Y Y Y Y wlh1
em4_fpus Y Y Y Y Y wlh1 fpus
em4_fpuda Y Y Y Y Y wlh1 fpuda
quarkse_em Y Y Y Y Y wlh2 quark Y Y

The above -mcpu values correspond to specific ARC EM Processor templates presented in the ARChitect tool. It should be noted however that some ARC features are not currently supported in the GNU toolchain, for example DSP instruction support.

  • -mcpu=em doesn't correspond to any specific template, it simply defines the base ARC EM configuration without any optional instructions.
  • -mcpu=em_mini is same as em, but uses reduced register file with only 16 core registers.
  • -mcpu=em4 is a base ARC EM core configuration with -mcode-density option. It corresponds to the following ARC EM templates in ARChitect: em4_mini, em4_sensor, em4_ecc, em6_mini, em5d_mini, em5d_mini_v3, em5d_nrg, em7d_nrg, em9d_mini. Note that those mini templates have a reduced core register file, but -mcpu=em4 doesn't enable this feature.
  • -mcpu=arcem doesn't correspond to any specific template, it is a legacy flag preserved for compatibility with older GNU toolchain versions, where -mcpu used to select only a CPU family, while optional features were enabled or disabled by individual -m<something> options.
  • -mcpu=em4_dmips is a full-featured ARC EM configuration for integer operations. It corresponds to the following ARC EM templates in ARChitect: em4_dmips, em4_rtos, em6_dmips, em4_dmips_v3, em4_parity, em6_dmips_v3, em6_gp, em5d_voice_audio, em5d_nrg_v3, em7d_nrg_v3, em7d_voice_audio, em9d_nrg, em9d_voice_audio, em11d_nrg and em11d_voice_audio.
  • -mcpu=em4_fpus is like em4_dmips but with additional support for single-precision floating point unit. It corresponds to the following ARC EM templates in ARChitect: em4_dmips_fpusp, em4_dmips_fpusp_v3, em5d_nrg_fpusp and em9d_nrg_fpusp.
  • -mcpu=em4_fpuda is like em4_fpus but with additional support for double-precision assist instructions. It corresponds to the following ARC EM templates in ARChitect: em4_dmips_fpuspdp and em4_dmips_fpuspdp_v3.
  • -mcpu=quarkse_em is a configuration for ARC processor in Intel Quark SE chip. It enables additional floating point instructions which cannot be enable using a particular target option.

Values of -mcpu for ARC 600 and 700 Families

-mcpu= -mnorm -mswap -mbarrel-shifter Multiplier
arc700 Y Y Y -mmpy
arc600 Y
arc600_norm Y Y
arc600_mul64 Y Y -mmul64
arc600_mul32x16 Y Y -mmul32x16
arc601
arc601_norm Y
arc601_mul64 Y -mmul64
arc601_mul32x16 Y -mmul32x16

Controlling the Memory Model

-mcmodel Memory size for Data/Code
small 1MB region
medium 4GB region (used by -fpic)
large Full memory (used by -fPIC)
Clone this wiki locally