Abstract: Efficient techniques are described for controlling ordered accesses in a weakly ordered storage system. A stream of memory requests is split into two or more streams of memory requests and a memory access counter is incremented for each memory request. A memory request requiring ordered memory accesses is identified in one of the two or more streams of memory requests. The memory request requiring ordered memory accesses is stalled upon determining a previous memory request from a different stream of memory requests is pending. The memory access counter is decremented for each memory request guaranteed to complete. A count value in the memory access counter that is different from an initialized state of the memory access counter indicates there are pending memory requests. The memory request requiring ordered memory accesses is processed upon determining there are no further pending memory requests.

Abstract: Techniques for controllably allocating a portion of a plurality of memory banks as cache memory are disclosed. To this end, a configuration tracker and a bank selector are employed. The configuration tracker configures whether each memory bank is to operate in a cache or not. The bank selector has a plurality of bank distributing functions. Upon receiving an incoming address, the bank selector determines the configuration of memory banks currently operating as the cache and applies an appropriate bank distributing function based on the configuration of memory banks. The applied bank distributing function utilizes bits in the incoming address to access one of the banks configured as being in the cache.

Abstract: Whenever a link address is written to the link stack, the prior value of the link stack entry is saved, and is restored to the link stack after a link stack push operation is speculatively executed following a mispredicted branch. This condition is detected by maintaining an incrementing tag register which is incremented by each link stack write instruction entering the pipeline, and a snapshot of the incrementing tag register, associated with each branch instruction. When a branch is evaluated and determined to have been mispredicted, the snapshot associated with it is compared to the incrementing tag register. A discrepancy indicates a link stack write instruction was speculatively issued into the pipeline after the mispredicted branch instruction, and pushed a link address onto the link stack, thus corrupting the link stack. The prior link address is restored to the link stack from the link stack restore buffer.

Abstract: Whenever a link address is written to the link stack, the prior value of the link stack entry is saved, and is restored to the link stack after a link stack push operation is speculatively executed following a mispredicted branch. This condition is detected by maintaining a count of the total number of uncommitted link stack write instructions in the pipeline, and a count of the number of uncommitted link stack write instructions ahead of each branch instruction. When a branch is evaluated and determined to have been mispredicted, the count associated with it is compared to the total count. A discrepancy indicates a link stack write instruction was speculatively issued into the pipeline after the mispredicted branch instruction, and pushed a link address onto the link stack. The prior link address is restored to the link stack from the link stack restore buffer.

Abstract: Techniques and methods are used to control allocations of cache lines to a higher level cache that have been displaced from a lower level cache. The allocations of the displaced cache lines are prevented for displaced cache lines that are determined to be redundant in the next level cache, whereby the displaced cache line castouts are not allocated to the higher level cache. To such ends, a line is selected to be displaced in a lower level cache. Information associated with the selected line is identified which indicates that the selected line is present in a higher level cache. An allocation of the selected line in the higher level cache is prevented based on the identified information.

Abstract: Efficient techniques are described for enforcing order of memory accesses. A memory access request is received from a device which is not configured to generate memory barrier commands. A surrogate barrier is generated in response to the memory access request. A memory access request may be a read request. In the case of a memory write request, the surrogate barrier is generated before the write request is processed. The surrogate barrier may also be generated in response to a memory read request conditional on a preceding write request to the same address as the read request. Coherency is enforced within a hierarchical memory system as if a memory barrier command was received from the device which does not produce memory barrier commands.

Abstract: A method for retrieving a return address from a link stack when returning from a procedure in a pipeline processor is disclosed. The method identifies a retrieve instruction operable to retrieve a return address from a software stack. The method further identifies a branch instruction operable to branch to the return address. The method retrieves the return address from the link stack, in response to both the instruction and the branch instruction being identified and fetches instructions using the return address.

Abstract: Address translation performance within a processor is improved by identifying an address that causes a boundary crossing between different pages in memory and linking address translation information associated with both memory pages. According to one embodiment of a processor, the processor comprises circuitry configured to recognize an access to a memory region crossing a page boundary between first and second memory pages. The circuitry is also configured to link address translation information associated with the first and second memory pages. Thus, responsive to a subsequent access the same memory region, the address translation information associated with the first and second memory pages is retrievable based on a single address translation.

Abstract: Techniques and methods are used to reduce allocations to a higher level cache of cache lines displaced from a lower level cache. The allocations of the displaced cache lines are prevented for displaced cache lines that are determined to be redundant in the next level cache, whereby castouts are reduced. To such ends, a line is selected to be displaced in a lower level cache. Information associated with the selected line is identified which indicates that the selected line is present in a higher level cache. An allocation of the selected line in the higher level cache is prevented based on the identified information. Preventing an allocation of the selected line saves power that would be associated with the allocation.

Abstract: A method of processing a plurality of instructions in multiple pipeline stages within a pipeline processor is disclosed. The method partially or wholly executes a stalled instruction in a pipeline stage that has a function other than instruction execution prior to the execution stage within the processor. Partially or wholly executing the instruction prior to the execution stage in the pipeline speeds up the execution of the instruction and allows the processor to more effectively utilize its resources, thus increasing the processor's efficiency.

Abstract: A processor implements an apparatus and a method for predicting an indirect branch address. A target address generated by an instruction is automatically identified. A predicted next program address is prepared based on the target address before an indirect branch instruction utilizing the target address is speculatively executed. The apparatus suitably employs a register for holding an instruction memory address that is specified by a program as a predicted indirect address of an indirect branch instruction. The apparatus also employs a next program address selector that selects the predicted indirect address from the register as the next program address for use in speculatively executing the indirect branch instruction.

Abstract: Whenever a link address is written to the link stack, the prior value of the link stack entry is saved, and is restored to the link stack after a link stack push operation is speculatively executed following a mispredicted branch. This condition is detected by maintaining an incrementing tag register which is incremented by each link stack write instruction entering the pipeline, and a snapshot of the incrementing tag register, associated with each branch instruction. When a branch is evaluated and determined to have been mispredicted, the snapshot associated with it is compared to the incrementing tag register. A discrepancy indicates a link stack write instruction was speculatively issued into the pipeline after the mispredicted branch instruction, and pushed a link address onto the link stack, thus corrupting the link stack. The prior link address is restored to the link stack from the link stack restore buffer.

Abstract: A method of resolving simultaneous branch predictions prior to validation of the predicted branch instruction is disclosed. The method includes processing two or more predicted branch instructions, with each predicted branch instruction having a predicted state and a corrected state. The method further includes selecting one of the corrected states. Should one of the predicted branch instructions be mispredicted, the selected corrected state is used to direct future instruction fetches.

Abstract: Techniques and methods are used to control allocations to a higher level cache of cache lines displaced from a lower level cache. The allocations of the displaced cache lines are prevented for displaced cache lines that are determined to be redundant in the next level cache, whereby castouts are controlled. To such ends, a line is selected to be displaced in a lower level cache. Information associated with the selected line is identified which indicates that the selected line is present in a higher level cache. An allocation of the selected line in the higher level cache is prevented based on the identified information.

Abstract: When misses occur in an instruction cache, prefetching techniques are used that minimize miss rates, memory access bandwidth, and power use. One of the prefetching techniques operates when a miss occurs. A notification that a fetch address missed in an instruction cache is received. The fetch address that caused the miss is analyzed to determine an attribute of the fetch address and based on the attribute a line of instructions is prefetched. The attribute may indicate that the fetch address is a target address of a non-sequential operation. Another attribute may indicate that the fetch address is a target address of a non-sequential operation and the target address is more than X % into a cache line. A further attribute may indicate that the fetch address is an even address in the instruction cache. Such attributes may be combined to determine whether to prefetch.

Abstract: Whenever a link address is written to the link stack, the prior value of the link stack entry is saved, and is restored to the link stack after a link stack push operation is speculatively executed following a mispredicted branch. This condition is detected by maintaining a count of the total number of uncommitted link stack write instructions in the pipeline, and a count of the number of uncommitted link stack write instructions ahead of each branch instruction. When a branch is evaluated and determined to have been mispredicted, the count associated with it is compared to the total count. A discrepancy indicates a link stack write instruction was speculatively issued into the pipeline after the mispredicted branch instruction, and pushed a link address onto the link stack. The prior link address is restored to the link stack from the link stack restore buffer.

Abstract: Techniques and methods are used to reduce allocations to a higher level cache of cache lines displaced from a lower level cache. The allocations of the displaced cache lines are prevented for displaced cache lines that are determined to be redundant in the next level cache, whereby castouts are reduced. To such ends, a line is selected to be displaced in a lower level cache. Information associated with the selected line is identified which indicates that the selected line is present in a higher level cache or the selected line is a write-through line. An allocation of the selected line in the higher level cache is prevented based on the identified information. Preventing an allocation of the selected line saves power that would be associated with the allocation.

Abstract: Techniques and methods are used to control allocations to a higher level cache of cache lines displaced from a lower level cache. The allocations of the displaced cache lines are prevented for displaced cache lines that are determined to be redundant in the next level cache, whereby castouts are controlled. To such ends, a line is selected to be displaced in a lower level cache. Information associated with the selected line is identified which indicates that the selected line is present in a higher level cache. An allocation of the selected line in the higher level cache is prevented based on the identified information.

Abstract: A method of processing branch history information is disclosed. The method retrieves branch instructions from an instruction cache and executes the branch instructions in a plurality of pipeline stages. The method verifies that a branch instruction has been identified. The method further receives branch history information during a first pipeline stage and loads the branch history information into a first register. The method further loads the branch history information into the second register during the second pipeline stage.

Abstract: Whenever a link address is written to the link stack, the prior value of the link stack entry is saved, and is restored to the link stack after a link stack push operation is speculatively executed following a mispredicted branch. This condition is detected by maintaining a count of the total number of uncommitted link stack write instructions in the pipeline, and a count of the number of uncommitted link stack write instructions ahead of each branch instruction. When a branch is evaluated and determined to have been mispredicted, the count associated with it is compared to the total count. A discrepancy indicates a link stack write instruction was speculatively issued into the pipeline after the mispredicted branch instruction, and pushed a link address onto the link stack. The prior link address is restored to the link stack from the link stack restore buffer.