Abstract: Technologies are generally described herein for accelerating a cache state transfer in a multicore processor. The multicore processor may include first, second, and third tiles. The multicore processor may initiate migration of a thread executing on the first core at the first tile from the first tile to the second tile. The multicore processor may determine block addresses of blocks to be transferred from a first cache at the first tile to a second cache at the second tile, and identify that a directory at the third tile corresponds to the block addresses. The multicore processor may update the directory to reflect that the second cache shares the blocks. The multicore processor may transfer the blocks from the first cache in the first tile to the second cache in the second tile effective to complete the migration of the thread from the first tile to the second tile.

Abstract: Technologies are generally described for methods and systems to assign threads in a multi-core processor. In an example, a method to assign threads in a multi-core processor may include determining data relating to memory controllers fetching data in response to cache misses experienced by a first core and a second core. Threads may be assigned to cores based on the number of cache misses processed by respective memory controllers. Methods may further include determining that a thread is latency-bound or bandwidth-bound. Threads may be assigned to cores based on the determination of the thread as latency-bound or bandwidth-bound. In response to the assignment of the threads to the cores, data for the thread may be stored in the assigned cores.

Abstract: Techniques described herein generally include methods and systems related to cache partitioning in a chip multiprocessor. Cache-partitioning for a single thread or application between multiple data sources improves energy or latency efficiency of a chip multiprocessor by exploiting variations in energy cost and latency cost of the multiple data sources. Partition sizes for each data source may be selected using an optimization algorithm that minimizes or otherwise reduces latencies or energy consumption associated with cache misses.

Abstract: Technologies are generally described for methods, systems, and processors effective to migrate a thread. The thread may be migrated from the first core to the second core. The first and the second core may be configured in communication with a first cache. The first core may generate a request for a first data block from the first cache. In response to a cache miss in the first cache for the first data block, the first core may generate a request for the first data block from a memory. The first core may coordinate with a second cache to store the first data block in the second cache. The thread may be migrated from the second core to a third core. The second core and third core may be configured in communication with the second cache.

Abstract: Techniques are generally described for cache management in a processor with a cache. In response to receiving a bulk memory modification instruction, data blocks of the cache associated with the bulk memory modification instruction may be identified. A cache coherence state of the identified data blocks may also be identified. The updated cache coherence state may be indicative of a zero value of the data blocks and the cache coherence state of the identified data blocks may be updated without modification to a cache data array.

Abstract: Technologies are generally described for systems, devices, and methods effective to dynamically select at least one power supply rail for a router. In some examples, a power control unit may be configured to determine a buffer occupancy level of one or more buffers of the router. In some further examples, the buffer occupancy level may be compared to a threshold value. In various other examples, the at least one power supply rail of the router may be switched from a first power rail to a second power rail based on the results of the comparison.

Abstract: Techniques described herein generally include methods and systems related to dynamic cache-sizing used to reduce the energy consumption of a DRAM cache in a chip multiprocessor. Dynamic cache sizing may be performed by adjusting the refresh interval of a DRAM cache or by combining way power-gating of the DRAM cache with adjusting the refresh interval.

Abstract: Techniques described herein generally include methods and systems related to designing and operating a DRAM device that has significantly reduced refresh energy use. A method for designing a DRAM optimizes or otherwise improves the DRAM for energy efficiency based on a measured or predicted failure probability of memory cells in the DRAM. The DRAM may be configured to operate at an increased refresh interval, thereby reducing DRAM refresh energy but causing a predictable portion of the memory cells in the DRAM to leak electrical energy too quickly to retain data. The DRAM is further configured with a selected number of spare memory cells for replacing the “leaky” memory cells, so that operation of the DRAM at the increased refresh interval may result in little or no reduction in capacity of the DRAM.

Abstract: Technologies are generally described for methods and systems to assign threads in a multi-core processor. In an example, a method to assign threads in a multi-core processor may include determining data relating to memory controllers fetching data in response to cache misses experienced by a first core and a second core. Threads may be assigned to cores based on the number of cache misses processed by respective memory controllers. Methods may further include determining that a thread is latency-bound or bandwidth-bound. Threads may be assigned to cores based on the determination of the thread as latency-bound or bandwidth-bound. In response to the assignment of the threads to the cores, data for the thread may be stored in the assigned cores.

Abstract: Techniques described herein generally include methods and systems related to cache partitioning in a chip multiprocessor. Cache-partitioning for a single thread or application between multiple data sources improves energy or latency efficiency of a chip multiprocessor by exploiting variations in energy cost and latency cost of the multiple data sources. Partition sizes for each data source may be selected using an optimization algorithm that minimizes or otherwise reduces latencies or energy consumption associated with cache misses.

Abstract: A cache coherence mechanism may comprise a bit-to-cache map for processor cores operable up to a maximum frequency for cores of a multicore processor. Entries in a cache coherence directory may include a bit field identifying cores operable at or near the maximum frequency that share a memory block corresponding to the entry. An additional field may indicate sharing by cores operating at lower frequencies. The additional field may be indicative of the bit-field corresponding to a bit-to-cache map representative of cores other than those operating at or near the maximum frequency.

Abstract: A processor may comprise a plurality of cores operating at heterogeneous frequencies communicatively coupled by a network of routers also operating at heterogeneous frequencies. A core may be prioritized for thread execution based on operating frequencies of routers on a path from the core to a memory controller. Relatively higher priority may be assigned to cores having a path comprising only routers operating at a relatively higher frequency. A combined priority for thread execution may be based on core frequency, router frequency, and the frequency of routers on a path from the core to a memory controller. A core may be selected based primarily on core operating frequency when cache misses fall below a threshold value.

Abstract: Techniques described herein generally include methods and systems related to designing and operating a DRAM device that has significantly reduced refresh energy use. A method for designing a DRAM optimizes or otherwise improves the DRAM for energy efficiency based on a measured or predicted failure probability of memory cells in the DRAM. The DRAM may be configured to operate at an increased refresh interval, thereby reducing DRAM refresh energy but causing a predictable portion of the memory cells in the DRAM to leak electrical energy too quickly to retain data. The DRAM is further configured with a selected number of spare memory cells for replacing the “leaky” memory cells, so that operation of the DRAM at the increased refresh interval may result in little or no reduction in capacity of the DRAM.

Abstract: Technologies are described herein for improving performance and energy efficiency in a computing system while using a large memory page size. Some example technologies may configure a main memory of the computing system to include a page-to-chunk table and a data area. The page-to-chunk table may include multiple entries such as a first entry. The first entry may correspond to a page that is made up of multiple chunks. The first entry may include pointers to the multiple chunks stored in the data area.

Abstract: Technologies are generally described for methods and systems effective to implement a memory allocation accelerator. A processor may generate a request for allocation of a requested chunk of memory. The request may be received by a memory allocation accelerator configured to be in communication with the processor. The memory allocation accelerator may process the request to identify an address for a particular chunk of memory corresponding to the request and may return the address to the processor.

Abstract: Technologies are generally described herein to detect unidirectional resistance drift errors in a multilevel cell of a phase change memory. The resistance levels of the multilevel cell of the phase change memory may be encoded to detect unidirectional resistance drift errors. In some examples, Berger Code-compatible encoding may be used. When a word is written to the multilevel cell, a write check code may be generated. The write check code may be a binary representation of the number of zeroes contained in the word as written. When the word is read from the multilevel cell, a read check code may be generated. The read check code may be a binary representation of the number of zeroes contained in the word as read. An error can be detected if a comparison indicates that the write check code and the read check code are different.

Abstract: Technologies are generally described for a method, device and architecture effective to allocate resources. In an example, the method may include associating first and second resources with first and second resource identifiers and mapping the first and second resource identifiers to first and second sets of addresses in a memory, respectively. The method may include identifying that the first resource is at least partially unavailable. The method may include mapping the second resource identifier to at least one address of the first set of addresses in the memory when the first resource is identified as at least partially unavailable. The method may include receiving a request for the first resource, wherein the request identifies a particular address of the addresses in the first set of addresses. The method may include analyzing the particular address to identify a particular resource and allocating the request to the particular resource.

Abstract: Technologies are generally described for methods and systems effective to implement hybrid routers in multicore architectures. A first tile may include a processor core, a cache configured to be in communication with the processor core and a router configured to be in communication with the cache. The router may be effective to move data with a packet switching channel or a circuit switching channel. The first tile may include switching logic configured to be in communication with the cache and the router. The switching logic may be effective to receive a routing objective that may relate to energy or delay costs in routing data through the network. The switching logic may select one of the packet switching channel or the circuit switching channel to move the data through the network based on the routing objective.

Abstract: Technologies are generally described for methods and systems effective to implement a memory allocation accelerator. A processor may generate a request for allocation of a requested chunk of memory. The request may be received by a memory allocation accelerator configured to be in communication with the processor. The memory allocation accelerator may process the request to identify an address for a particular chunk of memory corresponding to the request and may return the address to the processor.

Abstract: Technologies are generally described for methods and systems to assign threads in a multi-core processor. In an example, a method to assign threads in a multi-core processor may include determining data relating to memory controllers fetching data in response to cache misses experienced by a first core and a second core. Threads may be assigned to cores based on the number of cache misses processed by respective memory controllers. Methods may further include determining that a thread is latency-bound or bandwidth-bound. Threads may be assigned to cores based on the determination of the thread as latency-bound or bandwidth-bound. In response to the assignment of the threads to the cores, data for the thread may be stored in the assigned cores.