Abstract: Embodiments relate to a system operation queue for a transaction. An aspect includes determining whether a system operation is part of an in-progress transaction of a central processing unit (CPU). Another aspect includes based on determining that the system operation is part of the in-progress transaction, storing the system operation in a system operation queue corresponding to the in-progress transaction. Yet another aspect includes, based on the in-progress transaction ending, processing the system operation in the system operation queue.

Abstract: A self-timed parallelized multi-core processor has an instruction decoder unit for receiving a program code instruction, determining an operating code and latency for the instruction, and assigning a loop index to the instruction. An instruction decomposer creates a primitive by decomposing the instruction, replacing the loop index with a core index, and broadcasting the primitive. Self-timed processing cores each having a unique core index compare the core index to their unique processing core index. The processing cores act on the primitive when their processing core index is within a threshold of the core index.

Abstract: A compute instruction to be executed is to use a memory operand in a computation. An address associated with the memory operand is to be used to locate a portion of memory from which data is to be obtained and placed in the memory operand. A determination is made as to whether the portion of memory extends across a specified memory boundary. Based on the portion of memory extending across the specified memory boundary, the portion of memory includes a plurality of memory units and a check is made as to whether at least one specified memory unit is accessible and whether at least one specified memory unit is inaccessible.

Abstract: Methods and apparatuses relating to accelerating vector multiplication. In one embodiment, an apparatus includes a first buffer to store a first cache line of indices for elements of a first vector, a second buffer to store a second cache line of indices for elements of a second vector, a comparison unit to compare each index of the first cache line of indices with each index of the second cache line of indices, a plurality of multipliers to each multiply an element from the first vector and an element from the second vector for an index match from the comparison unit to produce a product, and an adder to add together the product from each of the plurality of multipliers.

Abstract: Method and system for managing a speculative transaction in a processing unit is provided. The speculative transaction is initiated by dispatching a first instruction indicating start of the speculative transaction. One or more register file (RF) entries are marked as pre-transaction memory (PTM), in response to the initiating. At least one second instruction targeting at least one of the marked RF entries is dispatched, while the transaction is active, wherein the at least one second instruction writes new result data into the at least one RF entry. Previous result data evicted from the at least one RF entry by the new result data, is saved into a history buffer (HB) entry. The HB entry is marked as PTM, in response to the saving, wherein the processing unit, upon detecting a trigger, is rolled back to a state before the initiating the transaction by restoring the previous result data to the at least one RF entry.

Abstract: A method for improving power, performance, area (PPA) for mixed precision computations in a processing environment. The method includes determining a braiding factor as a number of units of work encoded into a physical thread. A value of the braiding factor is determined based on a mix of precision requirements presented for individual units of work. Units of work are classified as instructions for applied code transformation based on associated precision requirements for the processing environment. Instruction inputs from specified registers are packed together into a destination register according to the determined value of the braiding factor. The packed instructions presented in vector form are executed with an instruction set architecture configured for executing packed instructions of different precisions.