DDR3/4 and LPDDR2/3/4/4x Combo PHY&MAC
Overview
With sophisticated architecture and advanced technology, this DDR3/4 and LPDDR2/3/4/4x IP combo solution with high performance and low power. In 12~28nm CMOS process, KNiulink could offer both controller and PHY IPs. DDR3/4 and LPDDR2/3/4/4x could be supported in this combo solution. The speed of LPDDR4/4x is up to 4266Mbps. In architecture, it supports Multiport AMBA AXI interface, configurable port number and support asynchronous or synchronous AXI port. For PHY interface, it will integrate DFI4.0 compatible design. One de-skew PLL is embedded inside the PHY to improve jitter performance.
Key Features
- Support DDR3/DDR4 and LPDDR3/LPDDR4/LPDDR4x
- Support DDR4 up to 3200Mbps speed, LPDDR4/4x up to 4266Mbps
- Support Multiport AMBA AXI2.0 interface, port number, asynchronous or synchronous AXI port, data width, FIFO depth, and command queue depth configurable
- Support DFI4.0 compliant interface between controller and PHY
- Support 2 ranks, max density up to 8GB
- Support automatic temperature monitor and refresh rate adjust
- Support CA, write, read VREF eye training and per-bit de-skew training, write leveling training
- Support Inline BIST and SIPI/LFSR/USER patterns
- Support DDPHY loopback test
- Support fully pin-mux easy for PKG/PCB routing
- Support mask write, write/read DBI
- Support 2T CA timing, CA parity, and gear down mode(DDR4)
- Support ECC (error correcting code), 32/64bit data+8bit ECC, correct 1bit-error and detect 2bit-error(DDR4)
Benefits
- High efficiency Command Queue scheduler to provide maximum bandwidth
- Low latency, DFI PIPE configurable on both controller and PHY sides, AXI async/sync configurable, DFI ratio adjustable with frequency;
- Complex QoS, high urgent, priority-based Round-robin port arbitration, timeout, bandwidth restriction supported
- Support AXI read out of order
- Support dual channel access with independent dual command queue, interleaving processed in controller; (LPDDR4)
- Support per-bank refresh and burst refresh to improve DRAM efficiency; (LPDDR4)
- Support automatic clock gate, and different levels of low power mode
- (1)Doze sleep, automatic power down or self-refresh, controller clock could be closed
- (2)Light sleep, hand shake with SOC, root clock could be closed
- (3)Deep sleep, root clock could be closed and power down, data retention supported in DDRIO
- Support hardware-based DDR frequency switch (DFS), and DFI 1:2/1:1 adjustable with frequency, to reduce latency in low frequency
- Support different power mode in different rank
- Support FSP (Frequency-Set-Point) to change working frequency quickly; (LPDDR4)
Applications
- PC
- Server
- Workstations
- Cell Phone
- Storage Devices
Deliverables
- Datasheet (Including Integration Guideline, Interface PINs, clock and reset description, all training flows etc.) (DDRMC & DDRPHY)
- Register Map files (register address and function description), timing calculation sheet (DDRMC & DDRPHY)
- Timing lib/db, Layout Frame (.LEF) (DDRPHY)
- Encrypted RTL and Netlist, SPEF/SDF (DDRPHY)
- Top Level GDS (DDRPHY)
- RTL Code & SDC Constraints (DDRMC)
- Verification environment and cases (testbench, DDRIO Verilog model, ddr3/ddr4 initial flow, training flow, bandwidth access, DFT pattern etc. (DDRMC & DDRPHY)
Technical Specifications
Foundry, Node
28/22/14/12/11nm
Maturity
Silicon Proven
Availability
Available
Related IPs
- Fast Access Controller – a plug-and-play IP solution for fast embedded Flash Programming and Memory Testing
- Embedded Configuration and Test Processor
- M13 Multiplexer and De-multiplexer (M13MD)
- Complete memory system supporting any combinations of SDR SDRAM, DDR, DDR2, Mobile SDR, FCRAM, Flash, EEPROM, SRAM and NAND Flash, all in one IP core
- Clock and Data Recovery of HDB3/B3ZS coded signals
- High-performance 2D (sprite graphics) GPU IP combining high pixel processing capacity and minimum gate count.