Difference between revisions of "emFloat"
(→Arithmetic performance) |
m |
||
(11 intermediate revisions by one other user not shown) | |||
Line 1: | Line 1: | ||
− | emFloat is the performance-optimized floating-point library for microcontrollers. |
+ | [https://www.segger.com/products/development-tools/runtime-library/technology/floating-point-library/?mtm_campaign=kb&mtm_kwd=emFloat emFloat] is the performance-optimized floating-point library for microcontrollers. |
− | This wiki page explains and links to other articles providing information too specific for the user manual or [https://www.segger.com/products/development-tools/runtime-library/technology/floating-point-library/ product pages on www.segger.com] |
+ | This wiki page explains and links to other articles providing information too specific for the user manual or [https://www.segger.com/products/development-tools/runtime-library/technology/floating-point-library/?mtm_campaign=kb&mtm_kwd=emFloat product pages on www.segger.com] |
<div class="toclimit-3"> |
<div class="toclimit-3"> |
||
Line 8: | Line 8: | ||
== Arithmetic performance == |
== Arithmetic performance == |
||
+ | |||
+ | The following table summarizes the benchmarks for floating-point arithmetic and conversion |
||
+ | when forced to use only scalar, integer instructions in the RISC-V core, even if the core |
||
+ | implementation provides the single-precision (F) or double-precision (D) instruction set extension. |
||
{| class="wikitable sortable" |
{| class="wikitable sortable" |
||
Line 14: | Line 18: | ||
! scope="col" | Device |
! scope="col" | Device |
||
! scope="col" | Architecture |
! scope="col" | Architecture |
||
+ | ! scope="col" | Core |
||
! scope="col" | emFloat |
! scope="col" | emFloat |
||
! scope="col" | Configuration |
! scope="col" | Configuration |
||
Line 19: | Line 24: | ||
! scope="col" class="unsortable" | Link |
! scope="col" class="unsortable" | Link |
||
|- |
|- |
||
− | | |
+ | | FE310-G002 || RV32IMAC || SiFive E310-G002 || 2.6.0 || Speed || Run from ITIM, data in DTIM || [[Bench_Arith_FE310-G002_2p6p0_Speed|Results]] |
+ | |- |
||
+ | | FE310-G002 || RV32IMAC || SiFive E310-G002 || 2.6.0 || Balanced || Run from ITIM, data in DTIM || [[Bench_Arith_FE310-G002_2p6p0_Balanced|Results]] |
||
+ | |- |
||
+ | | Arty A7-100T || RV32IMAFC_Zba_Zbb || SiFive E76 || 2.6.0 || Speed || Run from ITIM, data in DTIM || [[Bench_Arith_Arty_A7_100T_E76_2p6p0_Speed|Results]] |
||
+ | |- |
||
+ | | Arty A7-100T || RV32IMAFC_Zba_Zbb || SiFive E76 || 2.6.0 || Balanced || Run from ITIM, data in DTIM || [[Bench_Arith_Arty_A7_100T_E76_2p6p0_Balanced|Results]] |
||
+ | |- |
||
+ | | GD32VF103 || RV32IMAC || Nuclei Bumblebee || 2.6.0 || Speed || Run from flash, data in internal RAM || [[Bench_Arith_GD32VF103_2p6p0_Speed|Results]] |
||
+ | |- |
||
+ | | GD32VF103 || RV32IMAC || Nuclei Bumblebee || 2.6.0 || Balanced || Run from flash, data in internal RAM || [[Bench_Arith_GD32VF103_2p6p0_Balanced|Results]] |
||
+ | |- |
||
+ | | TLSR9518A || RV32IMAC_P_xv5 || Andes N25F || 2.6.0 || Speed || Run from ILM, data in DLM || [[Bench_Arith_TLSR9518A_2p6p0_Speed|Results]] |
||
+ | |- |
||
+ | | TLSR9518A || RV32IMAC_P_xv5 || Andes N25F || 2.6.0 || Balanced || Run from ILM, data in DLM || [[Bench_Arith_TLSR9518A_2p6p0_Balanced|Results]] |
||
|- |
|- |
||
|} |
|} |
Latest revision as of 13:15, 31 May 2024
emFloat is the performance-optimized floating-point library for microcontrollers.
This wiki page explains and links to other articles providing information too specific for the user manual or product pages on www.segger.com
Contents
Arithmetic performance
The following table summarizes the benchmarks for floating-point arithmetic and conversion when forced to use only scalar, integer instructions in the RISC-V core, even if the core implementation provides the single-precision (F) or double-precision (D) instruction set extension.
Device | Architecture | Core | emFloat | Configuration | Notes | Link |
---|---|---|---|---|---|---|
FE310-G002 | RV32IMAC | SiFive E310-G002 | 2.6.0 | Speed | Run from ITIM, data in DTIM | Results |
FE310-G002 | RV32IMAC | SiFive E310-G002 | 2.6.0 | Balanced | Run from ITIM, data in DTIM | Results |
Arty A7-100T | RV32IMAFC_Zba_Zbb | SiFive E76 | 2.6.0 | Speed | Run from ITIM, data in DTIM | Results |
Arty A7-100T | RV32IMAFC_Zba_Zbb | SiFive E76 | 2.6.0 | Balanced | Run from ITIM, data in DTIM | Results |
GD32VF103 | RV32IMAC | Nuclei Bumblebee | 2.6.0 | Speed | Run from flash, data in internal RAM | Results |
GD32VF103 | RV32IMAC | Nuclei Bumblebee | 2.6.0 | Balanced | Run from flash, data in internal RAM | Results |
TLSR9518A | RV32IMAC_P_xv5 | Andes N25F | 2.6.0 | Speed | Run from ILM, data in DLM | Results |
TLSR9518A | RV32IMAC_P_xv5 | Andes N25F | 2.6.0 | Balanced | Run from ILM, data in DLM | Results |