Patent Application
20170357507
The field of the invention is data processing, or, more specifically, methods and apparatus for operation of a multi-slice processor.
The development of the EDVAC computer system of 1948 is often cited as the beginning of the computer era. Since that time, computer systems have evolved into extremely complicated devices. Today's computers are much more sophisticated than early systems such as the EDVAC. Computer systems typically include a combination of hardware and software components, application programs, operating systems, processors, buses, memory, input/output devices, and so on. As advances in semiconductor processing and computer architecture push the performance of the computer higher and higher, more sophisticated computer software has evolved to take advantage of the higher performance of the hardware, resulting in computer systems today that are much more powerful than just a few years ago.
One area of computer system technology that has advanced is computer processors. As the number of computer systems in data centers and the number of mobile computing devices has increased, the need for more efficient computer processors has also increased. Speed of operation and power consumption are just two areas of computer processor technology that affect efficiency of computer processors.
Methods and apparatus for operation of a multi-slice processor are disclosed in this specification. Such a multi-slice processor includes a plurality of execution slices and a plurality of load/store slices, where the load/store slices are coupled to the execution slices via a results bus. Such a multi-slice processor may further include a load/store superslice, where a load/store superslice includes a set predict array, a first load/store slice, and a second load/store slice. Operation of such a multi-slice processor includes: receiving a two-target load instruction directed to the first load/store slice and a store instruction directed to the second load/store slice; determining a first subset of ports of the set predict array as inputs for an effective address for the two-target load instruction; determining a second subset of ports of the set predict array as inputs for an effective address for the store instruction; and generating, in dependence upon logic corresponding to the set predict array that is less than logic implementing an entire load/store slice, output for performing the two-target load instruction in parallel with generating output for performing the store instruction.
The foregoing and other objects, features and advantages of the invention will be apparent from the following more particular descriptions of exemplary embodiments of the invention as illustrated in the accompanying drawings wherein like reference numbers generally represent like parts of exemplary embodiments of the invention.
Exemplary methods and apparatus for operation of a multi-slice processor in accordance with the present invention are described with reference to the accompanying drawings, beginning with
The computer (152) of
The example computer processor (156) of
The example multi-slice processor (156) of
Although the multi-slice processor (156) in the example of
Stored in RAM (168) in the example computer (152) is a data processing application (102), a module of computer program instructions that when executed by the multi-slice processor (156) may provide any number of data processing tasks. Examples of such data processing applications may include a word processing application, a spreadsheet application, a database management application, a media library application, a web server application, and so on as will occur to readers of skill in the art. Also stored in RAM (168) is an operating system (154). Operating systems useful in computers configured for operation of a multi-slice processor according to embodiments of the present invention include UNIX, Linux™, Microsoft Windows™, AIX™, IBM's z/OS™, and others as will occur to those of skill in the art. The operating system (154) and data processing application (102) in the example of
The computer (152) of
The example computer (152) of
The exemplary computer (152) of
The arrangement of computers and other devices making up the exemplary system illustrated in
For further explanation,
The multi-slice processor in the example of
The general purpose registers (206) are configured to store the youngest instruction targeting a particular logical register and the result of the execution of the instruction. A logical register is an abstraction of a physical register that enables out-of-order execution of instructions that target the same physical register.
When a younger instruction targeting the same particular logical register is received, the entry in the general purpose register is moved to the history buffer, and the entry in the general purpose register is replaced by the younger instruction. The history buffer (208) may be configured to store many instructions targeting the same logical register. That is, the general purpose register is generally configured to store a single, youngest instruction for each logical register while the history buffer may store many, non-youngest instructions for each logical register.
Each execution slice (204) of the multi-slice processor of
The arithmetic logic unit (212) depicted in the example of
The results bus (220) may be configured in a variety of manners and be of composed in a variety of sizes. In some instances, each execution slice may be configured to provide results on a single bus line of the results bus (220). In a similar manner, each load/store slice may be configured to provide results on a single bus line of the results bus (220). In such a configuration, a multi-slice processor with four processor slices may have a results bus with eight bus lines—four bus lines assigned to each of the four load/store slices and four bus lines assigned to each of the four execution slices. Each of the execution slices may be configured to snoop results on any of the bus lines of the results bus. In some embodiments, any instruction may be dispatched to a particular execution unit and then by issued to any other slice for performance. As such, any of the execution slices may be coupled to all of the bus lines to receive results from any other slice. Further, each load/store slice may be coupled to each bus line in order to receive an issue load/store instruction from any of the execution slices. Readers of skill in the art will recognize that many different configurations of the results bus may be implemented.
The multi-slice processor in the example of
The example multi-slice processor of
During the flush and recovery operation, in prior art processors, the dispatcher was configured to halt dispatch of new instructions to an execution slice. Such instructions may be considered either target or source instructions. A target instruction is an instruction that targets a logical register for storage of result data. A source instruction by contrast has, as its source, a logical register. A target instruction, when executed, will result in data stored in an entry of a register file while a source instruction utilizes such data as a source for executing the instruction. A source instruction, while utilizing one logical register as its source, may also target another logical register for storage of the results of instruction. That is, with respect to one logical register, an instruction may be considered a source instruction and with respect to another logical register, the same instruction may be considered a target instruction.
The multi-slice processor in the example of
In some cases, a load/store unit receiving an issued instruction, such as a load/store slice, may not yet be able to handle the instruction, and the instruction sequencing unit (240) may keep the instruction queued until such time as the load/store slice may handle the instruction. After the instruction is issued, the instruction sequencing unit (240) may track progress of the instruction based at least in part on signals received from a load/store slice.
For further explanation,
For example, in some cases, a load instruction, such as a two-target load instruction, would entirely occupy two individual load/store slices without the additional port logic, which would prevent simultaneous execution of the two-target load instruction with another instruction. However, the two-target load instruction may not use all components of a second load/store slice, and therefore, with the addition of the port logic that includes much less silicon, space, and logic than the logic of an entire load/store slice, the two-target load instruction may be executed simultaneously with another instruction. In this way, different combinations of load and stores, which may have needed more than two load/store slices to execute simultaneously, may now execute simultaneously.
The simultaneous execution of different combinations of single-target and two-target loads and stores may depend, at least in part, on a store instruction not needing to use a data cache (232) simultaneously with a two-target load instruction—thereby allowing the two-target load instruction to use both data caches (232x, 232y). In this way, a two-target load instruction may be executed simultaneously with a store instruction without the addition of another data cache to support the store instruction.
A set predict array (332) generally includes logic for inputting, at a given port, an effective address and outputting, on a corresponding port, an effective address hash value that may be used to index into a data cache. While the effective address hash value may be highly accurate, the comparison logic (352x, 352y, 352z) receiving the output values of the set predict array may validate, or confirm to be correct, the accuracy of the output values by referencing a corresponding data cache directory (“DDIR”) (354x, 354y) lookup for the same effective address input to the set predict array (332).
A load instruction may be a single-target load instruction or a two-target load instruction, where a single-target load instruction may be executed by a single load/store slice (222a), and where a two-target load instruction may be executed by two load/store slices (222a, 222b). For example, a two-target load instruction may specify bytes of data to load that are more than may be processed by a single load/store slice at a single time due to the data not being able to be loaded into a single load/store slice data cache at once. Similarly, a two-target store instruction may specify bytes of data to store that are more than may be processed by a single load/store slice at a single time.
The load/store superslice (222xy) includes components allowing for the implementation of two load/store slices that may each execute single-target loads and stores simultaneously. However, the additional port logic in excess of the logic for implementing the two load/store slices is configured such that the additional port logic, while being much less than the logic for implementing another entire load/store slice, allows for simultaneous execution of load and store instructions that would not otherwise be possible using simply two load/store slices. In other words, without the additional port logic, the logic for two load/store slices in a load/store superslice (222xy) may simultaneously process any two single-target load or single-target store instructions, but not any combination of two-target loads and two-target stores.
The additional port logic may include logic corresponding to the addition of an input port (“LSSz In”) and an output port (“LSSz Out”) for the set predict array (332), a port control (340) for determining input combinations to the set predict array (332), a MUX (350z) for handling a portion of a given two-target instruction, and comparison logic (352z) for analyzing output from the set predict array output port (“LSSz Out”).
As noted above, the load/store superslice (222xy) may include logic for implementing two load/store slices. A first load/store slice may include load/store slice (LSS) MUX (350x), data cache directory (DDIR) (354x), data cache (232x), store request queue (306x), load reorder queue (304x), load miss queue (308x), effective address to real address translation (ERAT) (330x) cache, and format (226x) logic.
A second load/store slice may include load/store slice (LSS) MUX (350y), data cache directory (DDIR) (354y), data cache (232y), store request queue (306y), load reorder queue (304y), load miss queue (308y), effective address to real address translation (ERAT) (330y) cache, and format (226y) logic.
The load/store superslice (222xy) may further include logic that is shared, including the load/store access queue (224), the set predict array (332), port control (340) logic, load/store slice z (LSSz) MUX (350z), and comparison logic for effective address hash value x (“EAHx Compare”) (352x), comparison logic for effective address hash value y (“EAHy Compare”) (352y), and comparison logic for effective address hash value z (“EAHz Compare”) (352z).
The load/store slice (222) may retrieve data from and store data to any tier of a memory hierarchy, beginning with a local data cache (232), and extending as far down in the hierarchy as needed to find requested data or to store data. Requested data, when received, may be provided to general purpose registers, virtual registers, or to some other destination. The received data may also be stored in a data cache (232) for subsequent access. The load/store slice (222) may also manage translations of effective addresses to real addresses to communicate with or access different levels of memory hierarchy.
A store reorder queue (306) may include entries for tracking the cache operations for sequential consistency and may reissue operations into the load/store pipeline for execution independent of an execution slice. For example, the store reorder queue (306) may signal the load/store access queue (224) to reissue an instruction.
A load miss queue (308) may issue requests for data to one or more data storage devices of a multi-tiered memory hierarchy, where a request for data may correspond to a load instruction for the data.
Responsive to the data being returned along the line (302) to the load/store slice (222), the data may be delivered to a destination such as the results bus (220 of
A load reorder queue (304) may track execution of cache operations issued to the load/store slice (222) and includes entries for tracking cache operations for sequential consistency, among other attributes. The load reorder queue (304) may also reissue operations into the load/store pipeline for execution, which provides operation that is independent of the execution slices. For example, the load reorder queue (304) may signal the load/store access queue (224) to reissue an instruction.
The effective address to real address translation (330) cache may be implemented on individual load/store slices, where the effective address to read address translation (330) cache may be indexed according to an effective address field of an instruction, where an indexed entry of the effective address to real address translation (330) cache may store a correspondingly mapped real address field.
A load/store slice (222), for a given load or store instruction specifying an effective or virtual address, may perform a translation from the effective address to a real address. Generally, translations from effective addresses to real addresses may go through multiple levels of translation—where a combination of the effective address to real address translation (330) cache, a translation lookaside buffer (not shown), and a page walk cache (not shown) may serve to store different results of previous address translations to speed future address translations. For example, effective address to real address translations may be stored within different tiers of memory having different performance characteristics, with a first tier being the effective address to real address translation (330) cache, a second tier being a translation lookaside buffer, a third tier being one or more page tables reachable through one or more page directories stored in a lower tier or tiers of processor memory. The effective address to real address translation (330) cache and the translation lookaside buffer may both cache results of previous address translations retrieved from stored page tables. The page walk cache may store results that allow the avoidance of performing a full page walk through every level of a multi-level address translation.
The load/store slice (222), in performing an effective address to real address translation for a load or store instruction, may first use an effective page number field of the effective address of the instruction to index the effective address to real address translation (330) cache. If there is a hit, then the entry for the hit provides the translation of an effective page number to a physical page number—where the real address used to access physical memory is comprised of the physical page number from the entry and an offset from an offset field of the effective address.
Otherwise, on a miss of the effective address to real address translation (330) cache, the load/store slice (222) may index the translation lookaside buffer using the effective page number of the effective address. If there is a hit, then the entry for the hit provides the translation for the effective page number of the effective address to physical page number.
Otherwise, on a miss of the translation lookaside buffer, the load/store slice (222) may perform a page walk of the multiple nested levels of address translation—where each level may correspond to a page directory indexed by a portion of the effective address ultimately leading to a page table storing effective to real translation addresses. The page directories and page tables may be read from lower tiers of processor memory.
For further explanation,
The method of
The method of
The outputs of LSSx MUX 350x and LSSz MUX 350z may be selected by the port control (340) logic identifying the load instruction as a two-target load instruction whose execution is dependent upon using the two data caches (232x, 232z)—where the output ports from the set predict array (332) corresponding to the two input ports for the two-target load instruction effective address, may feed into comparison logic components EAHx Compare (352x) and EAHz Compare (352z), and further where the comparison logic components (352x, 352z) validate the effective address hash value outputs from the set predict array (332) to ensure that an access to the data caches (232x, 232y) index into a correct location.
The port control logic (340) may select the input to the MUXs (350x, 350z) for the two-target load instruction through bypass line (360x) directly from the results bus, or, after one or more cycles, through an output (361) from the load/store access queue (224) that also receives the two-target load instruction. The port logic (340) may select the MUXs to receive the two-target load instruction simultaneously with selecting a MUX to receive a store instruction.
The method of
The output of LSSy MUX 350y may be selected by the port control (340) logic determining the store instruction as an instruction whose execution is not dependent upon using the two data caches (232x, 232y), and consequently, determining that the instruction may be executed simultaneously, or at least partially simultaneously, with the two-target load instruction—where the output port from the set predict array (332) corresponding to the input port for the store instruction effective address, may feed into comparison logic components EAHy Compare (352y), and further where the comparison logic component (352y) may validate the effective address hash value output from the set predict array (332).
The port control logic (340) may select the input to the MUX (350z) for the store instruction through bypass line (360y) directly from the results bus, or, after one or more cycles, through an output (361) from the load/store access queue (224) that also receives the store instruction. The port logic (340) may select the MUX to receive the two-target load instruction simultaneously with selecting MUXs to receive the two-target instruction. While in this example, the store instruction is a one-target store instruction, because neither a one-target store instruction or a two-target store instruction, would conflict with a one-target or two-target load instruction for use of the data caches, a two-target store instruction may also be processed similarly to a one-target store instruction.
The method of
In this example, the generated hash value outputs for the two-target load instruction effective address are propagated to EAHx Compare (352x) and EAHz Compare (352z) based on inputs corresponding to input ports (LSSx In, LSSz In) corresponding to the effective address for the two-target load instruction. Similarly, the generated hash value output for the store instruction effective address is propagated to EAHy Compare (352y) based on inputs corresponding to input port (LSSy In) corresponding to the effective address for the store instruction. Further, the output values may be used for loading data into the data caches and for storing data to a memory location.
In this way, in dependence upon additional port logic corresponding to the extra ports on the set predict array (332), the load/store superslice (222xy) may simultaneously process a two-target load instruction and a store instruction.
For further explanation,
The method of
The method of
Receiving (502) the plurality of load instructions may be carried out by the load/store access queue (224) receiving, and storing or queuing, load instructions received from the results bus. Further, the load instructions may be received from respective instruction sequencing units corresponding to respective execution slices, or the instructions may be received from a single instruction sequencing unit.
Receiving (504) the plurality of store instructions may be carried out by the load/store access queue (224) receiving, and storing or queuing, store instructions received from the results bus. Further, the store instructions may be received from respective instruction sequencing units corresponding to respective execution slices, or the instructions may be received from a single instruction sequencing unit. In this way, the load/store access queue (224) may store multiple different types of load and store instructions.
Determining (506) the plurality of pairings of load and store instructions may be carried out by the port logic (340) accessing the stored load and store instructions within the load/store access queue and pairing two-target (2TL) and one-target (1TL) load instructions with two-target (2TS) and one-target (1TS) store instructions, and within constraints of any possible dependencies, pairing possible combinations that may include {2TL, 1TS}, {2TL, 2TS}, {1TL, 1TL}—where the order is not relevant.
Some considerations in determining a pairing of instructions from the queued instructions in the load/store access queue (224) may include selecting pairs of oldest stores and oldest loads, selecting the oldest load with any age store. In other cases, the port logic (340) may select as many loads, regardless of age, within a cycle as possible, for example, two different one-target loads may be selected to launch before an older store instruction. In other examples, different pairings that result in a maximum, or in a relatively more efficient instruction throughput relative to other pairings, may be selected to increase an amount of instructions that may be launched an processed simultaneously. Such selection methodologies may prevent single-slice operations from being starved out due to servicing dual-slice operations.
Further considerations in determining a pairing of instructions from the queued instructions in the load/store access queue (224) may include prioritizing recycled load or store operations received from the load reorder queues or from the store reorder queues along lines (358) to be included in a pairing with another queued load or store operation. In some cases, the recycled operation may be selected based on an already determined operation. In another case, the recycled operation may be selected first, and the other operation in the pairing may be selected in dependence upon the selected recycled operation. In this way, recycled operations may be prevented from languishing, or being delayed, for too long.
The present invention may be a system, a method, and/or a computer program product. The computer program product may include a computer readable storage medium (or media) having computer readable program instructions thereon for causing a processor to carry out aspects of the present invention.
The computer readable storage medium can be a tangible device that can retain and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but is not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. A non-exhaustive list of more specific examples of the computer readable storage medium includes the following: a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), a static random access memory (SRAM), a portable compact disc read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanically encoded device such as punch-cards or raised structures in a groove having instructions recorded thereon, and any suitable combination of the foregoing. A computer readable storage medium, as used herein, is not to be construed as being transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission media (e.g., light pulses passing through a fiber-optic cable), or electrical signals transmitted through a wire.
Computer readable program instructions described herein can be downloaded to respective computing/processing devices from a computer readable storage medium or to an external computer or external storage device via a network, for example, the Internet, a local area network, a wide area network and/or a wireless network. The network may comprise copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. A network adapter card or network interface in each computing/processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium within the respective computing/processing device.
Computer readable program instructions for carrying out operations of the present invention may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like, and conventional procedural programming languages, such as the “C” programming language or similar programming languages. The computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider). In some embodiments, electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present invention.
Aspects of the present invention are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer readable program instructions.
These computer readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function/act specified in the flowchart and/or block diagram block or blocks.
The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions/acts specified in the flowchart and/or block diagram block or blocks.
The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems that perform the specified functions or acts or carry out combinations of special purpose hardware and computer instructions.
It will be understood from the foregoing description that modifications and changes may be made in various embodiments of the present invention without departing from its true spirit. The descriptions in this specification are for purposes of illustration only and are not to be construed in a limiting sense. The scope of the present invention is limited only by the language of the following claims.
