Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0806—Multiuser, multiprocessor or multiprocessing cache systems
  • G06F12/0842—Multiuser, multiprocessor or multiprocessing cache systems for multiprocessing or multitasking
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F9/00—Arrangements for program control, e.g. control units
  • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
  • G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
  • G06F9/30003—Arrangements for executing specific machine instructions
  • G06F9/3004—Arrangements for executing specific machine instructions to perform operations on memory
  • G06F9/30043—LOAD or STORE instructions; Clear instruction
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F9/00—Arrangements for program control, e.g. control units
  • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
  • G06F9/30—Arrangements for executing machine instructions, e.g. instruction decode
  • G06F9/38—Concurrent instruction execution, e.g. pipeline or look ahead
  • G06F9/3824—Operand accessing
  • G06F9/3834—Maintaining memory consistency
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
  • G06F2212/10—Providing a specific technical effect
  • G06F2212/1016—Performance improvement
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F2212/00—Indexing scheme relating to accessing, addressing or allocation within memory systems or architectures
  • G06F2212/10—Providing a specific technical effect
  • G06F2212/1032—Reliability improvement, data loss prevention, degraded operation etc

Definitions

• Embodiments of the present invention relate generally to data processing system operation and more particularly to a method, system, apparatus, and article of manufacture for performing cachelme polling utilizing a store and reserve instruction.
• information handling e.g., computer
• I/O input/output
• display devices keyboards
• cursor control devices e.g., mice, trackballs, or the like
• Modern, conventional information handling systems may therefore include a wide variety of system components (e.g., multiple processors using SMP, ASMP, NUMA, or similar configurations, coprocessors, direct memory access controllers, and I/O devices each of which may include additional processors, registers, and memory) .
• system components e.g., multiple processors using SMP, ASMP, NUMA, or similar configurations, coprocessors, direct memory access controllers, and I/O devices each of which may include additional processors, registers, and memory.
• Interrupts coupled with interrupt service routines or handlers may be utilized by information handling system components to communicate and/or to indicate the occurrence of an event.
• memory-mapped I/O and port or "port-mapped" I/O may be utilized to provide communication between system components (e.g., processors and I/O devices) .
• the coordination of activity among elements of an information handling system is of particular importance in the transfer of data between elements for the purposes of performing input/output (I/O) operations.
• the data providing processor will typically notify the I/O device or data-receiving processor that the transfer of data to the buffer is complete.
• notification is typically performed by writing a specific data value into a memory mapped input/output (MMIO) register within the I/O device or data-receiving processor.
• MMIO memory mapped input/output
• the I/O device or data-receiving processor may retrieve data from the buffer via a direct memory access (DMA) .
• DMA direct memory access
• MMIO register polling In some conventional information handling systems the completion of DMA retrieval of data can be detected via MMIO register polling or via interrupts. Neither MMIO register polling nor interrupts is an efficient mechanism for detecting the completion of the DMA however because interrupt overhead is typically too great for relatively small buffers and MMIO register polling inefficiently utilizes bus bandwidth which could otherwise be used for DMA transfers, increasing overall system throughput.
• a predetermined "busy” indicator data value is written into a cacheable memory location, typically known as a buffer flag or semaphore, prior to notifying an 1/0 device (e.g., via MMIO) of a buffer's availability.
• the processor then polls the buffer flag for a predetermined "not busy” indicator data value to detect the completion of a corresponding DMA. Since the data is already modified in the processor's cache, cacheline polling does not generate any additional bus activity.
• the I/O device or receiving processor writes a "not busy" completion data value to the buffer flag.
• the new buffer flag value can then be accessed by the data-providing processor via a normal cache coherency protocol during which the "busy"-indicatmg buffer flag data in cache memory is invalidated or replaced by a new completion value.
• cachelme polling is an efficient polling mechanism.
• the data- providing processor executes a set of "polling" instructions repeatedly until the DMA transfer is complete and the buffer flag value is updated, thus wasting valuable system resources (e.g., processor cycles, bus cycles, electrical power, instruction or thread dispatch slots, or the like) .
• a first process initially requests an action to be performed by a second process.
• a reservation is set at a cacheable memory location via a store operation.
• the first process reads the cacheable memory location via a load operation to determine whether or not the requested action has been completed by the second process.
• the load operation of the first process is stalled until the reservation on the cacheable memory location is lost.
• the reservation in the cacheable memory location is reset by the second process.
• FIG. 1 illustrates a communications network including an information handling system according to an embodiment of the present invention
• Fig. 2 illustrates a high-level block diagram of an information handling system according to an embodiment of the present invention
• Fig. 3 illustrates a block diagram representation of a selected portion of an information handling system capable of performing cachelme polling utilizing a store and reserve instruction according to an embodiment of the present invention
• Fig. 4 illustrates a state diagram of a state machine for managing a reservation for a store and reserve instruction according to an embodiment of the present invention
• Fig. 5 illustrates a flow diagram of process to perform cachelme polling utilizing a store and reserve instruction according to an embodiment of the present invention.
• Embodiments of the present invention provide a store and reserve instruction which may be utilized for performing cachelme polling embodied within a method, information handling system, and machine- readable medium article of manufacture as described herein.
• Fig. 1 illustrates a communications network including an information handling system according to an embodiment of the present invention.
• a conventional load (LD) instruction executed subsequent to a store and reserve (STAR) instruction as described will complete execution and return data after a previously- set load reservation, which sets the load reservation, has been "lost" or reset and will otherwise typically stall.
• a STAR instruction may be utilized within a single cachelme polling routine loop to correctly manage reservations.
• system 101 includes a network 104 to which a plurality of information handling systems (e.g., computers and computing devices) are coupled.
• network 104 may comprise a LAN, a global network, such as the Internet, or any other communications network.
• information handling systems coupled to network 104 include client computers 106, server computers 108, personal digital assistants (PDAs) 110, digital television (DTV) 112 and may further comprise other wired or wireless computers and computing devices not shown.
• the processing elements employed by the member information handling systems of network 104 are constructed from a common computing module. These processing elements also preferably all have the same instruction set architecture (ISA) and perform processing in accordance with a common processor instruction set.
• ISA instruction set architecture
• each computing module contains a processing controller and a plurality of identical processing units for performing parallel processing of the data and applications transmitted over network 104.
• This homogeneous configuration for system 101 facilitates adaptability, processing speed and processing efficiency. Because each member of system 101 performs processing using one or more (or some fraction) of the same computing module, the particular computer or computing device performing the actual processing of data and applications is less relevant than in conventional systems. The processing of a particular application and data, moreover, can be shared among the network's members. By uniquely identifying the cells comprising the data and applications processed by system 101 throughout the system, the processing results can be transmitted to the computer or computing device requesting the processing irrespective of where this processing occurred. Because the modules performing this processing have a common structure and employ a common ISA, the computational burdens of an added layer of software to achieve compatibility among the processing elements is avoided. This architecture and programming model facilitates the processing speed necessary to execute, e.g., real-time, multimedia applications.
• each software cell 102 contains, or can contain, both applications and data.
• Each software cell 102 also contains an ID to globally identify the cell throughout network 104 and system 101. This uniformity of structure for the software cells, and the software cells' unique identification throughout the network, facilitates the processing of applications and data on any computer or computing device of network 104.
• a client 106 may formulate a software cell 102 but, because of the limited processing capabilities of client 106, transmit this software cell to a server 108 for processing.
• Software cells can migrate, therefore, throughout network 104 for processing on the basis of the availability of processing resources on the network.
• the homogeneous structure of processing elements and software cells of system 101 also avoids many of the problems of today's heterogeneous networks. For example, inefficient programming models which seek to permit processing of applications on any ISA using any instruction set, e.g., virtual machines such as the Java virtual machine, are avoided. System 101, therefore, can implement broadband processing far more effectively and efficiently than conventional networks.
• Fig. 2 illustrates a high-level block diagram of an information handling system according to an embodiment of the present invention. While a particular number and arrangement of elements have been illustrated with respect to the information handling system of Fig. 2, it should be appreciated that embodiments of the present invention are not limited to data processing systems having any particular number, type, or arrangement of components and so many encompass a wide variety of data processing system types, architectures, and form factors (e.g., network elements or nodes, personal computers, workstations, servers, or the like) .
• data processing system types, architectures, and form factors e.g., network elements or nodes, personal computers, workstations, servers, or the like
• CBE 200 includes a power processor element (PPE) 210 and multiple synergistic processor elements (SPEs) 220-234 communicatively coupled together and with additional system elements described further herein via a high bandwidth internal element interconnect bus (EIB) 286.
• PPE power processor element
• SPEs synergistic processor elements
• CBE 200 of the illustrated embodiment further includes one or more external buses or devices 290 coupled to EIB 286 via a bus interface controller (BIC) 287 and a shared memory 289 coupled to EIB 286 via a memory interface controller (MIC) 288 as shown.
• BIC bus interface controller
• MIC memory interface controller
• CBE 200 may be a system-on-a-chip such that each of the elements depicted in Figure 2 may be provided on a single microprocessor chip. Moreover, in one embodiment CBE 200 is provided as a heterogeneous processing environment in which each of SPEs 220-234 may receive different instructions from each of the other SPEs in the system. Moreover, the instruction set for each of the SPEs is different from that of PPE 210, e.g., PPE 210 may execute Reduced Instruction Set Computer (RISC) based instructions while SPEs 220-234 execute vectorized instructions.
• RISC Reduced Instruction Set Computer
• SPEs 220-234 are coupled to each other and to PPE 210 via EIB 286. Additionally, SPEs 220-234 are each coupled to MIC 288 and BIC 287 via EIB 286.
• MIC 288 provides a communication interface to shared memory 289. Shared memory 289 may comprise any of a number of system memory-type storage elements such as random access memory (RAM) , read-only memory (ROM) , flash memory, or the like.
• BIC 287 provides a communication interface between CBE 200 and other external buses and devices 290.
• Exemplary external devices may include traditional I/O devices such as keyboards, displays, printers, cursor control devices (e.g., trackballs, mice, tablets, etc.), speakers, and microphones; storage devices such as fixed or “hard” magnetic media storage devices, optical storage devices (e.g., CD or DVD ROMs), solid state storage devices (e.g., USB, Secure Digital SDTM, CompactFlashTM, MMC, or the like) , removable magnetic medium storage devices such as floppy disks and tape, or other storage devices or mediums; and wired or wireless communication devices or media (e.g., communication networks accessed via modem or direct network interface) .
• I/O devices such as keyboards, displays, printers, cursor control devices (e.g., trackballs, mice, tablets, etc.), speakers, and microphones
• storage devices such as fixed or “hard” magnetic media storage devices, optical storage devices (e.g., CD or DVD ROMs), solid state storage devices (e.g., USB, Secure Digital SDTM, CompactF
• PPE 210 is a dual threaded processing element.
• the combination of this dual threaded PPE 210 and the eight SPEs 220-134 makes the CBE 200 capable of handling 10 simultaneous threads and over 228 outstanding memory requests.
• PPE 210 acts as a controller for the eight SPEs 220-234 which handle most of the computational workload.
• PPE 210 may be used to execute one or more conventional operating systems while SPEs 220-234 perform vectorized floating point code execution, for example.
• PPE 210 comprises a power processor unit (PPU) or core and associated level 1 (Ll) and level 2 (L2) caches (not shown) and each of SPEs 220-234 comprise a synergistic processing unit (SPU) , memory flow control units, local memory or store, and a bus interface unit comprising a combination direct memory access (DMA) controller, memory management unit (MMU) , and bus interface unit (not shown) .
• the described local memory or store comprises a 256 KB instruction and data memory which is visible to PPE 210 and can be addressed directly by software.
• PPE 210 may load SPEs 220-134 with small programs or threads, chaining the SPEs together to handle each step in a complex operation.
• a set-top box incorporating CBE 200 may load programs for reading a DVD, video and audio decoding, and display, and the data would be passed off from SPE to SPE until it finally ended up on the output display.
• each SPE 220-234 gives a theoretical 32 GFLOPS of performance with PPE 210 having a similar level of performance.
• PPE 210 may also execute instructions and handle or process data retrieved from shared memory 289 into its local registers or caches via MIC 288.
• an external device 290 may access shared memory 289, for example via BIC 287 and one or more DMA controllers within SPEs 220-234.
• Fig. 3 illustrates a block diagram representation of a selected portion of an information handling system capable of performing cachelme polling utilizing a store and reserve instruction according to an embodiment of the present invention.
• PPE 310 of the illustrated embodiment of Fig. 3 corresponds to PPE 210 of Fig. 2.
• Information handling system 300 of Fig. 3 includes a PPE 310 which, via EIB 386, is coupled to shared memory 389 and an external device 390 utilizing MIC 388 and BIC 387, respectively.
• shared memory 389 includes a cacheable memory location 336 including data which specifies a buffer flag data value as shown.
• PPE 310 includes a power processor unit (PPU) 316 hierarchically coupled to an Ll cache 312 and L2 cache 314 as shown.
• PPU 316 includes a number of functional units and data storage elements.
• PPU 316 comprises a load/store unit 318 utilized to execute memory accessing instructions (e.g., loads from memory and stores to memory) and a condition register 320 which stores data in the form of bits or flags indicating the current state of PPU 316, reflecting the result of certain data processing or information handling operations (e.g., data overflow or underflow, positive or negative result, or the like) .
• Ll cache 312 and L2 cache 314 include a cache management unit (CMU) (e.g., CMU 322 of Ll cache 312 and CMU 328 of L2 cache 314) as well as a storage element (e.g., storage element 324 of Ll cache 312 and storage element 332 of L2 cache 314) .
• CMU cache management unit
• CMUs 322 and 328 are each used to control the storage of data and/or instructions within a corresponding one of storage elements 324 and 332, implementing, for example, cachelme replacement algorithms, updating cachelme state or status metadata, or the like.
• Storage elements 324 and 332 in turn are utilized to store lines or blocks of data comprising application data and/or instructions as well as accompanying metadata (e.g., cache tags, status bits, or the like) .
• CMUs 322 and 328 has been depicted as integral units or modules of their respective caches, in alternative embodiments of the present invention CMUs 322 and/or 328 or the functionality thereof may be provided in other configurations (e.g., within a single one of Ll cache 312 and L2 cache 314, within PPU 316, as a separate unit or module, or a combination thereof) .
• PPU 316 may be utilized to perform cachelme polling via the execution of a store and reserve (STAR) instruction as will now be described.
• load/store unit 318 of PPU 316 may be initially used to execute a STAR instruction to cause data specifying a "busy" buffer flag indicator data value (e.g., OxBB) to be stored within cacheable memory location 336 of shared memory 389 as indicated by dashed line 338.
• the described STAR instruction is utilized to signal an associated external device 390 (e.g., a graphics device) that data to be retrieved by the device has been stored within an associated buffer (e.g., a dedicated portion of shared memory 389 or other memory within or external to, information handling system 300.
• an associated external device 390 e.g., a graphics device
• additional operations may be utilized to notify external device 390 that data has been written to the buffer flag stored within cacheable memory location 336 and that the retrieval of data from the described buffer may commence.
• cachelme polling may continue to be performed utilizing PPE 310 in which a LDRL instruction may be executed to cause data stored within cacheable memory location 336 of shared memory 389 to be stored within a register (e.g., a general purpose register) within PPU 316 (not shown) .
• a register e.g., a general purpose register
• the STAR instruction causes a reservation to be set by storing a specific data value within a reservation register 330 within L2 cache 314.
• a reservation is set by storing a logical ⁇ l' within a reservation bit of reservation register 330 and a memory address of a corresponding region of memory (e.g., a specific memory location or region associated with cacheable memory location 336 of shared memory 389) with which the reservation is associated.
• the setting and resetting of a reservation within reservation register 330 causes a corresponding reservation status bit to be set or reset within condition register 320.
• reservation register 330 has been illustrated as within CMU 328 of L2 cache 314, in alternative embodiments of the present invention such a reservation register or data may be stored elsewhere within information handling system 300 (e.g., within Ll cache 312, PU 316, a separate bus/EIB interface unit, or the like) .
• a reservation may be reset by CMU 328 of L2 cache 314 using a cache "snoop" operation following the detection of an attempt, request, or performance of a write (e.g., by external device 390) to cacheable memory location 336 as indicated by dashed line 340.
• the retrieved buffer flag data value may be compared to known "busy” and/or "not busy" buffer flag indicator data values. The described comparison may be utilized to determine whether external device 390' s retrieval (e.g., via DMA transfer) of data previously stored within an associated buffer has completed such that, for example, the associated buffer may be reused for additional data transfer to external device 390.
• Fig. 4 illustrates a state diagram of a state machine for managing a reservation for a store and reserve instruction according to an embodiment of the present invention.
• the illustrated "state machine” therefore represents operations to be performed by, or functionality incorporated into, one or more elements of an information handling system (e.g., information handling system 300 of Fig. 3) .
• such functionality is incorporated into a processing element or unit such as PPU 316 of Fig. 3, in other embodiments, such functionality maybe embodied within a standalone or additional system element capable of monitoring and controlling the operations of an associated information handling system.
• the functionality represented by the state diagram of Fig. 4 is instantiated for each hardware thread initiated or supported.
• state machine 400 includes five separate states, namely, state SO, state Sl, state S2, state S3 and state S4.
• a STAR instruction is utilized to set a buffer flag "busy" indicator data value, to initialize the state of buffer data to be accessed, and to set a load reservation as described herein.
• state machine 400 records the address of the cachelme for buffer flag after the receipt of a load instruction, and enters state Sl.
• the described load targets the address corresponding to a cachelme storing a buffer flag. Multiple load reservations can exist concurrently.
• a cache state check state Sl the state of the cache memory is checked. If the cachelme storing a buffer flag is invalid, state machine 400 enters state S4. If the cachelme storing a buffer flag is modified or shared, state machine 400 enters state S2.
• state machine 400 In a wait on reservation to be lost state S2, state machine 400 remains idle while the load reservation exist for the load operation. After the load reservation has been lost, state machine 400 enters state S4.
• a cache memory has a mechanism to detect if another processor is accessing one of its cachelmes. This mechanism is commonly referred to as a snoop machine.
• a similar process can be used by state machine 400 to determine if a cachelme is being modified by another processor or device.
• state machine 400 watches for store instructions to the buffer flag by another thread on the same processor or other processors sharing the cache memory.
• state machine 400 may be utilized to initiate or cause a transition from state S2 to state S4. If an interrupt is directed towards a stalled processor or processor thread in the described embodiment, state machine 400 will exit to state S4 to allow the interrupt to be processed. If the interrupt is not processed, the processor or other device may never update the buffer flag.
• a timeout may be utilized to cause a transition from state S2 to state S4.
• software may be utilized to trigger a timeout for the cachelme polling period.
• state machine 400 With the timeout option, state machine 400 will exit to state S4 after a specified amount of time.
• a timeout value can be set prior to or as a parameter of the load when reservation lost instruction.
• a wait on reservation to be lost (buffer flag not in processor's cache) state S3 may be entered if a cachelme storing a buffer flag is castout due to other load instructions or store instructions requiring the same cachelme.
• State S3 is substantially similar to the previously-described state S2 with the exception that the buffer flag is not stored in the processor's cache memory. The same exit conditions exist for state S3 as they are for state S2 however, the cachelme storing a buffer flag is marked as being no longer valid (i.e., modified or shared) in the processor's cache in state S3.
• cachelme castout operations may cause a reservation to be reset, State S3 may be eliminated.
• a transition from state S3 back to state S2 may occur if the cache memory preloads data from the system bus (e.g., cache injection) .
• a data forward state S4 the state of the cachelme storing a buffer flag is examined to determine if the buffer flag-storing cachelme contains valid data (i.e., shared, modified, etc.) . If so, the data is forwarded to the processor and state machine 400 reenters state SO. This is the case when the buffer flag data has not been modified by a processor or device because the loss of the load reservation is caused by another event (e.g., an interrupt or timeout) .
• the buffer flag comprises data specifying a "busy" indicator data value and a compare instruction will then be executed. If an interrupt is pending, the interrupt will be processed next. When the interrupt returns, a branch will be taken if the data returned was the "busy" indicator data value.
• a "load miss" is generated and buffer flag data is requested from shared memory via the bus.
• the buffer flag data is returned, the data is forwarded to the processor, and the state of the cache memory is updated.
• State machine 400 then enters state SO. This process is the same as a normal load operation that misses in a cache memory.
• the data has been modified by the device to comprise a "not busy" indicator data value (e.g., OxBC or any value not equal to OxBB) .
• the invalidation of a cachelme can be caused by a device or other processor reading the cachelme with the intention to modify (RWITM) the data stored therein, or by a device or other processor writing the cachelme (write with flush) .
• buffer flag or semaphore has been described herein as stored within a cacheable memory location, in alternative embodiments the memory location need not be cacheable. More specifically, methods or processes of the present invention may be applied to a buffer flag located in a non- cacheable memory location as long as the processor has a means to snoop for a device updating the buffer flag-containing memory address. Locating the buffer flag within a non-cacheable memory location is made apparent by the existence of state S3. In state S3, the buffer flag is not valid in the processor's cache, which is the same state that would exist for a non- cacheable flag.
• Fig. 5 illustrates a flow diagram of process to perform cachelme polling utilizing a store and reserve instruction according to an embodiment of the present invention.
• load/store unit 318 of PPU 316 fills a buffer within shared memory 389 with data (e.g., data to be handled or processed by external device 390) (process block 502) .
• data e.g., data to be handled or processed by external device 390
• the transfer of data to the described buffer may be performed utilizing a DMA controller or engine provided within one or more of SPEs 220-234.
• load/store unit 318 executes a store and reserve instruction (process block 503) .
• execution of the described STAR instruction stores data indicating or specifying a "busy” indicator data value within cacheable memory location 336 of shared memory 389 (process block 504) and "sets" a reservation utilizing reservation register 330 (process block 506) .
• external (e.g., I/O) device 390 may be notified (e.g., by writing to a MMIO register, generating an exception, interrupt, trap, or the like) that the associated data buffer is ready to be accessed (not show) .
• I/O I/O
• Load/store unit 318 then makes a determination whether the reservation has been reset (process block 508) (e.g., by checking one or more bits of flags within condition register 320) . While a continuous loop has been utilized to depict the determination of when a reservation is reset, it should be appreciated that no actual instructions are executed by either the processor (in a singly-threaded uni or multi-processor system) or an associated thread (in a multi-threaded processor system) , thus saving valuable processing and electrical power otherwise wasted utilizing conventional cachelme polling techniques. Once the reservation is reset (e.g., via the storage of a
• the buffer flag data stored within cacheable memory location 336 is loaded by load/store unit 318 into a register of PPU 316 (process block 510) .
• a fixed-point execution unit (not shown) of PPU 316 compares the contents of the register of a PPU 316 to which the buffer flag data of cacheable memory location 336 has been loaded to a specified "busy" indicator data value (process block 512) .
• a branch unit (not shown) of PPU 316 then utilizes the comparison result, determining whether the register contents matched "busy” indicator data value (process block 514). Thereafter, the depicted process embodiment is either restarted completely (process block 502) in response to a determination that the register contents do not match the known "busy" indicator data value or reentered (process block 503) at the point at which the STAR instruction was executed and the reservation set.
• Fig. 5 Although the operations depicted in Fig. 5 have been described with respect to specific system elements, the actual elements utilized to perform such operations is immaterial to process embodiments of the present invention. Moreover, in alternative embodiments, such operations may be performed by any information handling system elements. Similarly, while the flow diagram depicted in Fig. 5 indicates a particular order of operation and a specific granularity of process operations, in alternative embodiments the illustrated order may be varied (e.g., process operations may be performed in another order or performed substantially in parallel) and one or more of the process operations may be coalesced or fragmented. Similarly, addition process operations may be added where necessary in alternative embodiments of the present invention.
• Embodiments of the present invention may include software, information processing hardware, and various processing operations further described herein.
• the features and process operations of various invention embodiments may be embodied in executable instructions embodied within a machine-readable medium such as shared memory 289, a storage device, a communication device or medium, or the like.
• a machine-readable medium may include any mechanism that provides (i.e., stores and/or transmits) data in a form readable by a machine (e.g., CBE 200) .
• a machine-readable medium includes but is not limited to: random access memory (RAM) ; read only memory (ROM) ; magnetic storage media; optical storage media; flash memory devices; electrical, optical, and/or acoustical propagated signals (e.g., carrier waves, infrared signals, digital signals, etc.); or the like.
• the described executable instructions can be used to cause a general or special purpose processor such as PPU 316, programmed with the instructions, to perform operations, methods or processes of the present invention.
• the features or operations of the present invention may be performed by specific hardware components that contain hard-wired logic for performing such operations, or by any combination of programmed data processing components and custom hardware components .

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Software Systems (AREA)
• Physics & Mathematics (AREA)
• General Engineering & Computer Science (AREA)
• General Physics & Mathematics (AREA)
• Memory System Of A Hierarchy Structure (AREA)

Priority Applications (3)

Applications Claiming Priority (2)

Publications (2)

Family

ID=38509831

Family Applications (1)

Country Status (5)

Cited By (1)

Families Citing this family (10)

Citations (2)

Family Cites Families (15)

• 2006
  • 2006-03-16 US US11/377,505 patent/US9390015B2/en active Active
• 2007
  • 2007-02-26 CN CN2007800089935A patent/CN101401071B/zh active Active
  • 2007-02-26 JP JP2008558761A patent/JP4566264B2/ja active Active
  • 2007-02-26 WO PCT/EP2007/051810 patent/WO2007104638A2/en not_active Ceased
  • 2007-02-26 EP EP07726510.6A patent/EP1994469B1/en active Active

Patent Citations (2)

Also Published As

Similar Documents

Legal Events