WhyChips

A professional platform focused on electronic component information and knowledge sharing.

WAVE-J: High-Performance JPEG Codec IP for 8K Imaging

Wide banner of futuristic tiled glowing blue digital data storage panel array

Chips&Media has a new flagship codec IP: WAVE-J JPEG codec IP. It is a high-performance JPEG codec core for next-generation imaging systems. It replaces CODAJ12V, stays compatible with existing JPEG workflows, and boosts speed, configurability, and efficiency.

WAVE-J targets demanding imaging workloads. It supports ultra-large images, fast processing, many pixel formats, and built-in pre/post processing. In most cases, you won’t need a separate imaging accelerator.

Industry-Leading 8K Throughput Performance

WAVE-J is built for 8K JPEG processing. It encodes 8K at 60 fps at 430 MHz and decodes 8K at 60 fps at 500 MHz.

Throughput depends on chroma format and bit depth. For 8-bit 4:0:0, encoding reaches ~2,950 Mpixels/s and decoding reaches ~2,690 Mpixels/s. For 4:2:0, encoding reaches ~2,060 Mpixels/s and decoding reaches ~2,020 Mpixels/s. Versus CODAJ12V, WAVE-J is up to 3.5× faster for encoding and 2.7× faster for decoding.

Architectural Upgrades for Modern SoC Integration

WAVE-J moves from a 64-bit AMBA AXI3 interface to a 128-bit AMBA AXI4 interface. It supports AXI burst lengths of 8, 16, or 32 to improve memory efficiency on modern SoCs.

Quantization and Huffman tables can be loaded via APB or AXI to reduce control-path bottlenecks. A generalized DMA engine supports bitstream concatenation (useful for 90°/270° rotated images) and constant-value filling.

Expanded Ultra-Large Image Support

Maximum image size increases from 32K × 32K to 64K × 64K. This helps with panoramas, digital pathology, aerial mapping, AI dataset prep, and other ultra-large still-image workloads.

Internal memory is redesigned for higher density. An optional 28 KB SRAM configuration can improve latency tolerance in memory-constrained or heavily loaded systems.

Flexible Pixel Format & Integrated Image Processing

WAVE-J supports 8-bit, 10-bit, and 12-bit paths. It works with planar and packed formats such as YUV400/420/422/444, RGB, and ARGB. Bit conversion lets you use 10-bit I/O even when the JPEG path runs at 8-bit or 12-bit.

The built-in pre/post pipeline supports fixed downscaling (1/2 to 1/32). It also supports chroma conversion, color space conversion, cropping, mirroring, and rotation for both encode and decode. These features reduce external memory traffic and often remove the need for a separate image-processing accelerator.

Boosted Submodule Throughput

Submodules are faster across the board. Pre/post processing is ~3.2× faster. Transform and quantization throughput is doubled. Variable-length coding is 2–4× faster. Variable-length decoding is up to 2.5× faster.

Entropy processing is optimized for 8×8 block operation. This improves sustained throughput and reduces slowdowns on vertically oriented images.

Low-Power, Small-Footprint TSMC 7nm Implementation

The design uses one external codec clock. Internal clock gating separates encoder and decoder domains to save power.

In a TSMC 7nm JPEG core at 500 MHz, estimated power is ~18.75 mW (encode) and ~14.80 mW (decode) for an 8-bit configuration with pre/post enabled. Area is ~0.05–0.1 mm² depending on configuration. A full-featured setup uses ~1.03M logic gates and 16.64 KB SRAM, totaling ~1.33M gate equivalents.

Modular Configuration for Custom Use Cases

WAVE-J is configurable. You can choose encoder-only, decoder-only, or both. You can select 8-bit or 12-bit processing. You can enable only the pre-processing features you need. You can also trade memory capacity for latency tolerance.

This modular approach helps control area and power across product lines. Although a full WAVE-J build is larger than CODAJ12V, it delivers much higher throughput and more features. It also improves performance per imaging task and makes SoC integration easier for demanding platforms.

Final Verdict

WAVE-J fits smartphones, cameras, surveillance, AI servers, edge devices, and home appliances. It combines 8K60 performance, ultra-large image support, AXI4 connectivity, broad format support, and integrated image processing. It is a compact imaging subsystem, not just a JPEG engine.

Related Articles

  • WAVE-P: Hardware Acceleration for the APV Professional Video Codec
  • Chips&Media Strengthens Codec Leadership With Next-Gen AV2 Licensing Deal
  • Chips&Media’s Next-Generation Video CODEC IP Powers Ambarella’s Expanding Edge AI Portfolio

发表回复