diff options
Diffstat (limited to 'recipes-demo/de10-nano-fftsw-apps/files/README_TARGET_BUILD.TXT')
-rw-r--r-- | recipes-demo/de10-nano-fftsw-apps/files/README_TARGET_BUILD.TXT | 151 |
1 files changed, 151 insertions, 0 deletions
diff --git a/recipes-demo/de10-nano-fftsw-apps/files/README_TARGET_BUILD.TXT b/recipes-demo/de10-nano-fftsw-apps/files/README_TARGET_BUILD.TXT new file mode 100644 index 0000000..a69cae3 --- /dev/null +++ b/recipes-demo/de10-nano-fftsw-apps/files/README_TARGET_BUILD.TXT @@ -0,0 +1,151 @@ +To build the applications in this directory, you should be able to run these +target scripts in this order: + +./target_build_all.sh + this will build all the applications +./archive_for_target.sh + this will archive all applications and target scripts into fft.tgz for + installation on the target + +At this point everything should be built. If you'd like to install and run +everything you can run this script: + +./setup_target_fft_env.sh + this target script will ensure that the fft_driver is installed, ensure + that /mnt/ram has a 100MB tmpfs ram disk mounted, and then extract the + fft.tgz archive into /mnt/ram/fft and run the create_input_waveforms.sh + script and then runs the run_all.sh script + +NOTE: see the README_TARGET.TXT file for additional information + +Description of directory contents: + +target_build_all.sh + target script to build all applications and libraries + +target_build_app.sh + target script to build one application + +target_build_lib.sh + target script to build one library + +clean_all.sh + target script to clean all build objects + +archive_for_target.sh + target script to archive all applications and scripts for target + installation + +create_input_waveforms.sh + target script that executes waveform creation applications and + duplication scripts to create all input waveforms for the applications + to consume + +duplicate_x128.sh +duplicate_x32.sh +duplicate_x8.sh + these target scripts duplicate a file x8, x32 or x128 times. These are + used to replicate the fundamental waveforms into longer patterns + +images/ + this directory contains host scripts to create PNG images of waveform + data text files using gnu plot and montage from imagemagick + +run_all.sh +run_fft_256.sh +run_fft_256x32.sh +run_fft_256x32x128.sh +run_fft_4096.sh +run_stream_256x32x128.sh +run_stream_256x16x1.sh +run_stream_256x1x1.sh + these target scripts execute all of the various permutations of the demo + applications mentioned below + +setup_target_fft_env.sh + this target script is used to extract the fft.tgz archive onto the ram + disk mounted at /mnt/ram + +overhead.c + this is the source for the overhead library that all of the applications + are linked against + +c16_256.c +c16_256x32.c +c16_256x32x128.c +c16_4096.c + these applications use the NE10 C functions to implement the FFT + algorithm for a variety of input sample sizes with 16-bit resolution + +c32_256.c +c32_256x32.c +c32_256x32x128.c +c32_4096.c + these applications use the NE10 C functions to implement the FFT + algorithm for a variety of input sample sizes with 32-bit resolution + +fft_256.c +fft_256x32.c +fft_256x32x128.c +fft_4096.c + these applications use the FPGA logic to implement the FFT + algorithm for a variety of input sample sizes. This variant pushes + the input samples into the FPGA and pulls the results back out using + programmed I/O from the HPS Cortex A9 processor + +fftdma_256.c +fftdma_256x32.c +fftdma_256x32x128.c +fftdma_4096.c + these applications use the FPGA logic to implement the FFT + algorithm for a variety of input sample sizes. This variant moves + the input samples into the FPGA and the results back out using DMA + through the ACP port of the Cortex A9 processor + +neon16_256.c +neon16_256x32.c +neon16_256x32x128.c +neon16_4096.c + these applications use the NE10 NEON functions to implement the FFT + algorithm for a variety of input sample sizes with 16-bit resolution + +neon32_256.c +neon32_256x32.c +neon32_256x32x128.c +neon32_4096.c + these applications use the NE10 NEON functions to implement the FFT + algorithm for a variety of input sample sizes with 32-bit resolution + +stream_fpga_256x32x128.c +stream_fpga_256x16x1.c +stream_fpga_256x1x1.c +stream_neon32_256x32x128.c +stream_neon32_256x16x1.c +stream_neon32_256x1x1.c +stream_raw_256x32x128.c +stream_raw_256x16x1.c +stream_raw_256x1x1.c + these applications stream data from the FPGA into ramdisk files, the + fpga variant DMAs the input stream thru the FPGA FFT logic before it + is DMAed through the ACP port of the Cortex A9, the neon variant DMAs + the raw input data stream through the ACP port and then computes the FFT + algorithm on it before saving to the ram disk, the raw variant DMAs the + raw input data stream through the ACP port and then directly into the + ram disk + +create_real_short_sine32.c +create_real_short_square32.c +create_real_short_triangle32.c + these applications create the initial 32 samples of their respective + waveform, sine, square and triangle + +real_short_to_ne10cpx_long.c +real_short_to_ne10cpx_short.c + these applications translate the real short data format into ne10cpx + long and short format which is consumed by the demo applications + +ne10cpx_long_to_text.c +ne10cpx_short_to_text.c + these applications translate the ne10cpx long and short format binary + data into text format output which may be read by gnu plot for example + |