International Business Machines Corporation

Apple Inc.

Jason Nathaniel Dale

Jonathan James DeMent

Clark M O'Niell

Steven Leonard Roberts

Timothy Hume Heil

Kimberly Marie Fernsler

Jeffrey Powers Bradford

Christopher John Spandikow

Jeffrey A Stuecheli

Francis Patrick O'Connell

Miles Robert Dooley

Gavin Balfour Meil

Gregory Howard Bellows

James Allen Rose

Adam P Burns

William John Starke

Richard James Eickemeyer

Bradly George Frey

Yaoqing Gao

Joaquin Madruga

Robert D Kenney

Dean J Burdick

John Barry Griswell, Jr

Philip Lee Vitale

Chris Abernathy

Brian H Horton

Alex E Mericas

Douglas Robert Logan

Greg H Bellows

Jim A Kahle

Richard J Eickemeyer, Jr

Techniques are disclosed relating to sharing datapath circuitry among multiple SIMD groups. In some embodiments, pipeline circuitry is configured to perform operations specified by instructions of first and second assigned SIMD groups. The pipeline circuitry may include first and second front-end circuitry configured to decode instructions of the respective SIMD groups. The pipeline circuitry may include shared execution circuitry configured to perform operations specified by the first and second assigned SIMD groups and arbitration circuitry configured to select an instruction from among at least the first and second front-end circuitry for assignment to the shared execution circuitry in a current cycle. The arbitration circuitry may select an instruction based on one or more of: stall counts, whether available instructions are being speculatively executed, whether ones of available instructions target a particular portion of the shared execution circuitry, numbers of execution cycles, and SIMD group ages.

Multi-channel data path circuitry

A technique for data prefetching for a multi-core chip includes determining memory utilization of the multi-core chip. In response to the memory utilization of the multi-core chip exceeding a first level, data prefetching for the multi-core chip is modified from a first data prefetching arrangement to a second data prefetching arrangement to minimize unused prefetched cache lines. In response to the memory utilization of the multi-core chip not exceeding the first level, the first data prefetching arrangement is maintained. The first and second data prefetching arrangements are different.

Data prefetch ramp implemenation based on memory utilization

A system and method operable to manage misaligned direct memory access (DMA) data transfers is provided. This method involves determining a delta between N bytes of data to be copied from within a local side buffer (source location) to a remote buffer (destination location). After the delta is determined a tail of the same length is copied to temporary storage. Then the N bytes of data on the local side buffer minus the tail will be shifted to align the N bytes of data to be copied from within the local side buffer to the starting address of the destination location in the remote buffer. The pre-shifted N bytes of data within the local side buffer may be DMA transferred to the remote buffer. The tail transferred to temporary storage may then be copied from temporary storage to the remote buffer.

Managing misaligned DMA addresses

A system and method operable to manage a message queue is provided. This management may involve out-of-order asynchronous heterogeneous remote direct memory access (RDMA) to the message queue. This system includes a pair of processing devices, a primary processing device and an additional processing device, a memory in storage location and a data bus coupled to the processing devices. The processing devices cooperate to process queue data within a shared message queue wherein when an individual processing device successfully accesses queue data the queue data is locked for the exclusive use of the processing device. When the processing device acquires the queue data, the queue data is locked and the queue data acquired by the acquiring processing device includes the queue data for both the primary processing device and additional processing device such that the processing device has all queue data necessary to process the data and return processed queue data.

Managing an out-of-order asynchronous heterogeneous remote direct memory access (RDMA) message queue

A prefetch unit includes a transience register and a length register. The transience register hosts an indication of transient for data stream prefetching. The length register hosts an indication of a stream length for data stream prefetching. The prefetch unit monitors the transience register and the length register. The prefetch unit generates prefetch requests of data streams with a transient property up to the stream length limit when the transience register indicates transient and the length register indicates the stream length limit for data stream prefetching. A cache controller coupled with the prefetch unit implements a cache replacement policy and cache coherence protocols. The cache controller writes data supplied from memory responsive to the prefetch requests into cache with an indication of transient. The cache controller victimizes cache lines with an indication of transient independent of the cache replacement policy.

Persistent prefetch data stream settings

A mechanism is provided in a multi-core environment for assigning a globally unique core identifier. A Power PC&#xae; processor unit (PPU) determines an index alias corresponding to a natural index to a location in local storage (LS) memory. A synergistic processor unit (SPU) corresponding to the PPU translates the natural index to a first address in a core's memory, as well as translates the index alias to a second address in the core's memory. Responsive to the second address exceeding a physical memory size, the load store unit of the SPU truncates the second address to a usable range of address space in systems that do not map an address space. The second address and the first address point to the same physical location in the core's memory. In addition, the aliasing using index aliases also preserves the ability to combine persistent indices with relative indices without creating holes in a relative index map.

Assigning efficiently referenced globally unique identifiers in a multi-core environment

A system, method, and program product are provided that identifies a cache set using Translation LookAside Buffer (TLB) attributes. When a virtual address is requested, the method, system, and program product identifies a cache set using buffer attributes. When a virtual address is received, an attempt is made to load the received virtual address from a cache. When the attempt results in a cache miss, a page is identified within a Translation LookAside Buffer that includes the virtual address. A class identifier is then retrieved from the identified page, with the class identifier identifying a cache set that is selected from the cache.

System and method for cache-locking mechanism using translation table attributes for replacement class ID determination

A system, method, and program product are provided that identifies a cache set using Segment LookAside Buffer attributes. When an effective address is requested, an attempt is made to load the received effective address from an L2 cache. When this attempt results in a cache miss, the system identifies a segment within the Segment LookAside Buffer that includes the effective address. A class identifier is retrieved from the identified segment within the Segment LookAside Buffer. This class identifier identifies a cache set selected from the cache for replacement. Data is then reloaded into the cache set of the cache by using the retrieved class identifier that corresponds to the effective address.

System and method for cache-locking mechanism using segment table attributes for replacement class ID determination

One of the processors of a multiprocessor system is chosen to be a boot processor. The other processors of the multiprocessor system execute masking code that generates electromagnetic and/or thermal signatures that mask the electromagnetic and/or thermal signatures of the actual boot processor. Such masking may involve running the same boot code as the boot processor but without obtaining access to security information, such as the security key for accessing the system. The electromagnetic and/or thermal signatures generated by the execution of the masking code preferably approximate the electromagnetic and/or thermal signatures of the actual boot code executing on the boot processor. In this way, it is difficult to distinguish which processor is the actual boot processor.

Masking a hardware boot sequence

Boot code is partitioned into a plurality of boot code partitions. Processors of a multiprocessor system are selected to be boot processors and are each provided with a boot code partition to execute in a predetermined boot code sequence. Each processor executes its boot code partition in accordance with the boot code sequence and signals to a next processor the successful and uncompromised execution of its boot code partition. If any of the processors does not signal successful completion and/or uncompromised execution of its boot code partition, the boot operation fails. The processors may be arranged, with regard to the boot operation, in a daisy chain, ring, or master/slave arrangement, for example.

Secure boot across a plurality of processors

Pervasive logic is provided that includes a random event generator. The random event generator randomly selects which processor of a plurality of processors in the multiprocessor system is to be a boot processor for the multiprocessor system. A corresponding configuration bit for the randomly selected processor is set to identify the processor as a boot processor. Based on the setting of the configuration bits for each processor in the plurality of processors, a selection of a security key is made. The security key is then used to decrypt the boot code for booting the multiprocessor system. Only the randomly selected boot processor is able to select the correct security key for correctly decrypting the boot code, which it then executes to bring the system to an operational state.

Selecting a random processor to boot on a multiprocessor system

A mechanism is provided for booting a multiprocessor device based on selection of encryption keys to be provided to the processors. With the mechanism, a security key and one or more randomly generated key values are provided to a selector mechanism of each processor of the multiprocessor device. A random selection mechanism is provided in pervasive logic that randomly selects one of the processors to be a boot processor and thereby, provides a select signal to the selector of the boot processor such that the boot processor selects the security key. All other processors select one of the one or more randomly generated key values. As a result, only the randomly selected boot processor is able to use the proper security key to decrypt the boot code for execution.

Booting a multiprocessor device based on selection of encryption keys to be provided to processors

A mechanism is provided for masking a boot sequence by providing a dummy processor. With the mechanism, one of the processors of a multiprocessor system is chosen to be a boot processor. The other processors of the multiprocessor system execute masking code that generates electromagnetic and/or thermal signatures that mask the electromagnetic and/or thermal signatures of the actual boot processor. The execution of the masking code on the non-boot processors preferably generates electromagnetic and/or thermal signatures that approximate the signatures of the actual boot code execution on the boot processor. One of the non-boot processors is selected to execute masking code that is different from the other masking code sequence to thereby generate a electromagnetic and/or thermal signature that appears to be unique from an external monitoring perspective.

Masking a boot sequence by providing a dummy processor

Masking a boot sequence by providing a dummy processor is provided. One of the processors of a multiprocessor system is chosen to be a boot processor. The other processors of the multiprocessor system execute masking code that generates electromagnetic and/or thermal signatures that mask the electromagnetic and/or thermal signatures of the actual boot processor. The execution of the masking code on the non-boot processors preferably generates electromagnetic and/or thermal signatures that approximate the signatures of the actual boot code execution on the boot processor. One of the non-boot processors is selected to execute masking code that is different from the other masking code sequence to thereby generate an electromagnetic and/or thermal signature that appears to be unique from an external monitoring perspective.

Page size prediction is used to predict a page size for a page of memory being accessed by a memory access instruction such that the predicted page size can be used to access an address translation data structure. By doing so, an address translation data structure may support multiple page sizes in an efficient manner and with little additional circuitry disposed in the critical path for address translation, thereby increasing performance.

Growing community of inventors

Austin, TX, United States of America

Jason Nathaniel Dale