International Business Machines Corporation

John Henry Westermann, Jr

Viktor Sandor Gyuris

Harrell Hoffman

Gernot Eberhard Guenther

Thomas John Tryt

Kevin Anthony Pasnik

Daniel R Crouse, Ii

Gernot E Günther

Vikto Gyuris

A system and method for design verification and, more particularly, a hardware simulation accelerator design and method that exploits a parallel structure of user models to support a large user model size. The method includes a computer including N number of logic evaluation units (LEUs) that share a common pool of instruction memory (IM). The computer infrastructure is operable to: partition a number of parallel operations in a netlist; and send a same instruction stream of the partitioned number of parallel operations to N number of LEUs from a single IM. The system is a hardware simulation accelerator having a computer infrastructure operable to provide a stream of instructions to multiple LEUs from a single IM. The multiple LEUs are clustered together with multiple IMs such that each LEU is configured to use instructions from any of the multiple IMs thereby allowing a same instruction stream to drive the multiple LEUs.

Hardware simulation accelerator design and method that exploits a parallel structure of user models to support a larger user model size

A circuit arrangement and method detect external requests to access a memory array in a hardware simulation accelerator during performance of a simulation on a simulation model and access the memory array without halting the simulation in response to detecting the external request. Such functionality may be provided, for example, by detecting such external requests in response to processing a predetermined instruction in an instruction stream associated with the simulation model, where the predetermined instruction is configured to ensure a predetermined period of inactivity for the memory array. By doing so, the memory array can be accessed from outside of the hardware simulation accelerator during the processing of a simulation, and without requiring that the simulation be halted, thus reducing overhead and improving simulation efficiency.

Concealment of external array accesses in a hardware simulation accelerator

A hardware simulation accelerator to simulate logic designs, a method to encode instructions for use in the hardware simulation accelerator, and a computer program product having code of the method by which the hardware simulation accelerator can read encoded instructions to simulate the logic design, and computer program product of the encoded instructions to simulate a logic design in a hardware accelerator. Each instruction has one of a plurality of opcodes, the opcodes select which of the hardware resources of the hardware simulation accelerator will implement and use the values set forth in other programmable bits of the encoded instruction. The encoded instruction may be a routing and/or a gate evaluation instruction.

Instruction encoding in a hardware simulation accelerator

A hardware accelerator includes hardware support for a combinational only cycle and a latch only cycle in a simulation model with a single partition of latches and combinational logic. Preferred embodiments use a special 4-input 1-output function unit in the hardware accelerator in place of the normal latch function that write back the old latch value for combinational only cycles. Other embodiments include hardware support for separate array write disables for arrays and transparent latches depending on whether the cycle is a combinational only cycle and a latch only cycle. A conditional array write disable dependent on the occurrence of a hardware breakpoint is also included that supports switching from a latch plus combinational cycle to a latch only cycle, to give control to the user before evaluating the combinational logic if a breakpoint occurs on a latch.

Hardware accelerator with a single partition for latches and combinational logic

A method and system are described in a logic simulator machine for efficiently creating a trace of an array which includes a plurality of storage locations. The logic simulator machine executes a test routine. Prior to executing the test routine, an initial copy of all data included within each of the storage locations of the array is stored as a first trace of the array. During execution of a first cycle the test routine, all of the write control inputs into the array are read to identify ones of the storage locations which were modified during the execution of the first cycle. A new trace of the array is generated which includes a copy of all of the data of the first trace. In addition, only those ones of the storage locations in the first trace which were modified during the first cycle are updated. A trace is thus generated by updating only those ones of the storage locations which were modified during execution of a cycle of the test routine.

Growing community of inventors

Endicott, NY, United States of America

John Henry Westermann, Jr