Moon2 improves Java applications
Moon2 improves Java applications
By Colin Holland, Embedded Systems
April 22, 2002 (7:48 a.m. EST)
URL: http://www.eetimes.com/story/OEG20020422S0005
VULCANmachines, has released it second-generation processor for Java-based technology. Moon2 is a 32bit microprocessor core designed for use in highperformance, low-power, Java-based embedded systems. The processor is available as Soft IP to be licensed and synthesized into ASIC and FPGA based system-on-chip designs. It is a native processor using a stack-based architecture designed from the ground up to execute Java bytecodes optimally. Bytecodes are executed directly, with the core of the instruction set implemented in hardware, and complex tasks microcoded or delegated to firmware. By eliminating the software overhead of a Java Virtual Machine or JIT solution, Moon2 gives an order of magnitude performance increase in Java execution.Moon2 can be employed as a coprocessor to a host core in a multi-lingual, multi-processor system. For example, where a heritage design has significant investment in existing code (which may be written in C or ass embler), it can be used to accelerate the execution of Java without subtracting from the processing bandwidth of the host core. Alternatively, where there is no significant legacy code Moon2 can be deployed as a stand-alone processor in single language (i.e. Java) system. Both VHDL and Verilog implementations available and its uses 27K gates plus 3K ROM and 1K single port RAM. Power consumption is typically 0.25mWMHz in 0.18mm process. There are interfaces to all major industry standard bus architecture and an optional code cache. The physical device access is in Java with no operating system or Virtual Machine required. The runtime environment is designed to support the J2ME CLDC specification and beyond Moon2 comprises an instruction execution unit and a datapath unit, which handles ALU operations and stack and memory access. The instruction execution unit fetches bytecodes from the memory location pointed to by the program counter, typically fetching more than one complete bytecode at a time since the width of the data bus is larger than most bytecode sequences. The optional code cache in Moon2 can be used to decrease the execution time of frequently used code. Furthermore the cache minimises bus contention and memory access times where it is deployed as a slave to a host processor. Bytecodes are stored in a small prefetch buffer. Since this buffer can hold several memory words at once, bytecode execution can usually be performed at a rate of one bytecode per clock tick. In cases where a particular bytecode requires multiple ALU operations, several microcode instructions are executed preceding the next bytecode fetch. The current bytecode is decoded by a microcode ROM which indicates, for example; the ALU operation required and the size of immediate arguments. The datapath unit consists of the ALU, hardware stack, and a number of internal registers. The hardware stack is implemented in memory on t he device upon which the Moon Core itself is implemented. In general ALU operations are executed on the top two elements of the stack; also data may be pushed onto the stack (e.g. by an immediate data instruction), or popped from it (e.g.as the result of a two-operand operation). The pushes and pops required for stack operations are merged together wherever possible. A generic bus architecture is used to simplify the integration of third-party IP with the Moon2 Core. Simple HDL wrappers are available to enable IP which is compliant with industry standard bus architectures such as AMBA AHB or PVCI (Peripheral Virtual Component Interface) to interface directly to Moon2. Kit aids development of Moon2-based designs Vulcan Machines has also developed the MoonLander development kit which provides an implementation of Moon2 as a stand-alone processor core. Based on Altera's SOPC development board, MoonLander forms a system prototyping tool for Java embedded system designers enabling both hardwa re and software integration and verification. The Moon2 reference design is pre-loaded into flash memory on the MoonLander board and boots up on power up. Altera's Quartus II software can be used to re-program flash memory using a Programmer Object File (.pof), alternatively the PLD itself can be programmed directly with a bitstream (eg .sof) file via a MasterBlaster or ByteBlaster programming cable. The reference design includes a number of peripherals including a UART, external SDRAM memory interface, LED interface and a JTAG port to enable debug and code download. HDL source code for the reference design is provided, along with a HDL simulation test bench which mimics the operation of the MoonLander development board. The development board includes an APEX 20K400E device, which is 22% utilised with the Moon2 reference design, 4Mbyte of Flash memory, pre-configured with the 32bit Moon2 reference design, and 64Mbyte SDRAM module in a DIMM socket. It has standard interconnects for 10/100 Ethernet, (USB) and Firewire as well as two RS-232 serial ports and two PS/2 ports. Debugging ports, including the SignalTap embedded logic analyser and JTAG. Multiple clocks, user configurable I/O lines, switches and LED's are provided.
Related News
- SiFive Expands and Improves Industry-Leading RISC-V Processor Portfolio
- Cadence Extends Battery Life and Improves User Experience for Next-Generation Hearables, Wearables and Always-On Devices
- Synopsys Improves Automotive Functional Safety with Fast Soft-Error Analysis
- Flex Logix Improves Deep Learning Performance By 10X With New EFLX4K AI eFPGA Core
- Averant's Solidify 6.5 Significantly Improves Combinational and Sequential Equivalency Checking and Clock Domain Crossing Checks
Breaking News
- Frontgrade Gaisler Unveils GR716B, a New Standard in Space-Grade Microcontrollers
- Blueshift Memory launches BlueFive processor, accelerating computation by up to 50 times and saving up to 65% energy
- Eliyan Ports Industry's Highest Performing PHY to Samsung Foundry SF4X Process Node, Achieving up to 40 Gbps Bandwidth at Unprecedented Power Levels with UCIe-Compliant Chiplet Interconnect Technology
- CXL Fabless Startup Panmnesia Secures Over $60M in Series A Funding, Aiming to Lead the CXL Switch Silicon Chip and CXL IP
- Cadence Unveils Arm-Based System Chiplet
Most Popular
- Cadence Unveils Arm-Based System Chiplet
- CXL Fabless Startup Panmnesia Secures Over $60M in Series A Funding, Aiming to Lead the CXL Switch Silicon Chip and CXL IP
- Esperanto Technologies and NEC Cooperate on Initiative to Advance Next Generation RISC-V Chips and Software Solutions for HPC
- Eliyan Ports Industry's Highest Performing PHY to Samsung Foundry SF4X Process Node, Achieving up to 40 Gbps Bandwidth at Unprecedented Power Levels with UCIe-Compliant Chiplet Interconnect Technology
- Arteris Selected by GigaDevice for Development in Next-Generation Automotive SoC With Enhanced FuSa Standards
E-mail This Article | Printer-Friendly Page |