Image Greyscaler Benchmark Results

Input Image

Image dimensions (pixels): 28000x28000 (FHD == 1920x1080)
Image size (bytes): 119,828,615

Hardware/Software Specifications

Hardware

CPU

CPU:       Info: Quad Core
           model: Intel Core i7-4790K
           bits: 64
           type: MT MCP
           arch: Haswell
           rev: 3
           cache:
              L2: 8 MiB
           flags: avx avx2 lm nx pae sse sse2 sse3 sse4_1 sse4_2 ssse3 vmx
           bogomips: 63873
           Speed: 4333 MHz
           min/max: 800/4400 MHz
           Core speeds (MHz):
              1: 4333 2: 4043 3: 4323 4: 4126
              5: 4346 6: 4282 7: 4277 8: 4339

RAM

Memory:    RAM: total: 11.54 GiB
           Array-1: capacity: 32 GiB
           slots: 4
           EC: None
           max-module-size: 8 GiB
           note: est.
           Device-1: ChannelA-DIMM0 size: No Module Installed
           Device-2: ChannelA-DIMM1 size: 4 GiB speed: 1333 MT/s type: DDR3
           Device-3: ChannelB-DIMM0 size: No Module Installed
           Device-4: ChannelB-DIMM1 size: 8 GiB speed: 1333 MT/s type: DDR3

CUDA

CUDA Device Query (Runtime API) version (CUDART static linking)

Detected 1 CUDA Capable device(s)

Device 0:
  CUDA Driver Version / Runtime Version          11.3 / 11.3
  CUDA Capability Major/Minor version number:    5.0
  Total amount of global memory:                 3998 MBytes (4192337920 bytes)
  (005) Multiprocessors, (128) CUDA Cores/MP:    640 CUDA Cores
  GPU Max Clock rate:                            1110 MHz (1.11 GHz)
  Memory Clock rate:                             2700 Mhz
  Memory Bus Width:                              128-bit
  L2 Cache Size:                                 2097152 bytes
  Total amount of constant memory:               65536 bytes
  Total amount of shared memory per block:       49152 bytes
  Total shared memory per multiprocessor:        65536 bytes
  Total number of registers available per block: 65536
  Warp size:                                     32
  Maximum number of threads per multiprocessor:  2048
  Maximum number of threads per block:           1024
  Max dimension size of a thread block (x,y,z): (1024, 1024, 64)
  Max dimension size of a grid size    (x,y,z): (2147483647, 65535, 65535)

deviceQuery
CUDA Driver = CUDART
CUDA Driver Version = 11.3
CUDA Runtime Version = 11.3
NumDevs = 1
Result = PASS

Software

Operating System

Kernel: 5.12.14 x86_64

Compilers and Tools

gcc:
```
gcc (GCC) 11.1.0
```

clang:

clang version 12.0.1
Target: x86_64-pc-linux-gnu
Thread model: posix

cmake:
```
cmake version 3.20.5
```

Algorithm

The overall algorithm of all implementation models is as the following:

Read the input image
Load the entire image into memory (RAM or GPU DRAM)
Start timer
Transform image (launch threads in multi-threaded implementation)
Gather results
Stop timer
Print elapsed execution time
Write results to output image file
Clean any allocated memory

Limitations

Current implementation uses libjpeg for reading, manipulating, and writing jpeg images.

Results

Results are average of only processing time of executing the same implementation with different configuration parameters 2000 times:

OpenMP
pthread
CUDA

Speedup

The following is the best execution time of different parallelism model implementations compared in one view (+ single-threaded serial execution):

xeptore/grayscaler-benchmarks