The Ziptilion Bandwidth IP accelertes the main memory bandwidth with up to 50%. The IP core packages a novel and proprietary technology that accelerates the limited off-chip bandwidth of main memories through real-time, general-purpose and on-the-fly memory data compression. The product benefit is significantly more main memory bandwidth at unmatched power efficiency.
Ziptilion Bandwidth IP is integrated in the memory subsystem of the SoC, close to the memory controller so that it can intercept all the memory traffic to/from DRAM to compress and decompress the data on-the-fly. The effect of compression is transparent to the CPU/accelerator subsystem as well as to the operating system and applications. Similarly, the memory controller is also unaware that the transmitted/received memory data is compressed. In essence, data compression and decompression, compaction as well as addressing the compressed memory space are handled automatically, transparently and hardware-accelerated by the IP.
Up to 50% main memory bandwidth acceleration
Overview
Key Features
- Bandwidth acceleration: 25-50%
- Performance acceleration: 10-25%
- Compression ratio: 2-3x across diverse data sets
- Frequency: DDR4/DDR5 DRAM speed
- IP area: Starting at 0.3mm2 (@5nm TSMC)
- Memory technologies supported: (LP)DDR4, (LP)DDR5, HBM
- Ziptilion Bandwidth IP is compatible with all DRAM technologies and supports standard interfaces such as AXI and CHI. Other proprietary interfaces can be supported upon request.
Benefits
- High performance and low latency main memory bandwidth acceleration 25% average, with peak of 50%
- Unmatched power efficiency
- Real-time compression, super-fast compaction and transparent memory management
- Operating at main memory speed and throughput
- Compatible to AXI4/CHI, both 128-b and 256-b bus interface
- Intelligent real-time analysis and tuning of the IP Block
Applications
- Server CPUs, Smart devices and Embedded systems all face the same challenge. The memory bandwidth is limiting the system scaling and the many cores and accelerators are fighting to serve their memory access requests. A wide range of data set from these different applications have been evaluated and they all verify that it is evident that bandwidth acceleration provides a very efficient and effective way to utilize the full memory potential.
Deliverables
- Synthesizable Verilog RTL (encrypted)
- Implementation constraints
- UVM testbench (self-checking)
- Vectors for testbench and expected results
- User Documentation
Technical Specifications
Foundry, Node
TSMC 7nm, 5nm, 3nm
Maturity
Tape-out
Availability
Immediate
Related IPs
- JPEG 2000 Encoder - Up to 16-bit per Component Lossy & Numerically Lossless Image & Video Compression
- Output Buffer for up to 12Gb/s
- PHY/PCS Logical Sub-Block IP Core for PCIe supporting PCIe 5.0, 4.0, 3.1 PHY/PMA and compliant to the PIPE 5.2 and 4.4.1 Specifications
- 32-kHz Bandwidth Reconfigurable Delta-Sigma ADC providing up to 13 ENOB
- 4x improvement to vector computation with 4x sustained bandwidth of prior generations
- 20 to 50 MHz crystal oscillator