Patent application number | Description | Published |
20140115210 | Multi Processor Multi Domain Conversion Bridge with Out of Order Return Buffering - An asynchronous dual domain bridge is implemented between the cache coherent master and the coherent system interconnect. The bridge has 2 halves, one in each clock/powerdown domain—master and interconnect. The asynchronous bridge is aware of the bus protocols used by each individual processor within the attached subsystem, and can perform the appropriate protocol conversion on each processor's transactions to adapt the transaction to/from the bus protocol used by the interconnect. | 04-24-2014 |
20140115220 | SYNCHRONIZING BARRIER SUPPORT WITH ZERO PERFORMANCE IMPACT - The barrier-aware bridge tracks all outstanding transactions from the attached master. When a barrier transaction is sent from the master, it is tracked by the bridge, along with a snapshot of the current list of outstanding transactions, in a separate barrier tracking FIFO. Each barrier is separately tracked with whatever transactions that are outstanding at that time. As outstanding transaction responses are sent back to the master, their tracking information is simultaneously cleared from every barrier FIFO entry. | 04-24-2014 |
20140115265 | OPTIMUM CACHE ACCESS SCHEME FOR MULTI ENDPOINT ATOMIC ACCESS IN A MULTICORE SYSTEM - The MSMC (Multicore Shared Memory Controller) described is a module designed to manage traffic between multiple processor cores, other mastering peripherals or DMA, and the EMIF (External Memory InterFace) in a multicore SoC. The invention unifies all transaction sizes belonging to a slave previous to arbitrating the transactions in order to reduce the complexity of the arbitration process and to provide optimum bandwidth management among all masters. The two consecutive slots assigned per cache line access are always in the same direction for maximum access rate. | 04-24-2014 |
20140115266 | OPTIONAL ACKNOWLEDGEMENT FOR OUT-OF-ORDER COHERENCE TRANSACTION COMPLETION - To enable efficient tracking of transactions, an acknowledgement expected signal is used to give the cache coherent interconnect a hint for whether a transaction requires coherent ownership tracking. This signal informs the cache coherent interconnect to expect an ownership transfer acknowledgement signal from the initiating master upon read/write transfer completion. The cache coherent interconnect can therefore continue tracking the transaction at its point of coherency until it receives the acknowledgement from the initiating master only when necessary. | 04-24-2014 |
20140115267 | Hazard Detection and Elimination for Coherent Endpoint Allowing Out-of-Order Execution - A coherence maintenance address queue tracks each memory access from receipt until the memory reports the access complete. The address of each new access is compared against the address of all entries in the queue. This check is made when the access is ready to transmit to the memory. If there is no address match, then the current access does not conflict with any pending access. If there is an address match, the current access is stalled. The multi-core shared memory controller would then typically proceed to another access waiting a slot to the endpoint memory. Stored addresses in the coherence maintenance address queue are retired when the endpoint memory reports completion of the operation. At this point the access is no longer a hazard to following operations. | 04-24-2014 |
20140115270 | MULTI PROCESSOR BRIDGE WITH MIXED ENDIAN MODE SUPPORT - An asynchronous dual domain bridge is implemented between the cache coherent master and the coherent system interconnect. The bridge has 2 halves, one in each clock/powerdown domain—master and interconnect. The asynchronous bridge is aware of the endian view used by each individual processor within the attached subsystem, and can perform the appropriate endian conversion on each processor's transactions to adapt the transaction to/from the endian view used by the interconnect. | 04-24-2014 |
20140115271 | COHERENCE CONTROLLER SLOT ARCHITECTURE ALLOWING ZERO LATENCY WRITE COMMIT - This invention speeds operation for coherence writes to shared memory. This invention immediately commits to the memory endpoint coherence write data. Thus this data will be available earlier than if the memory controller stalled this write pending snoop responses. This invention computes write enable strobes for the coherence write data based upon the cache dirty tags. This invention initiates a snoop cycle based upon the address of the coherence write. The stored write enable strobes enable determination of which data to write to the endpoint memory upon a cached and dirty snoop response. | 04-24-2014 |
20140115272 | Deadlock-Avoiding Coherent System On Chip Interconnect - This invention mitigates these deadlocking issues by a adding a separate non-blocking pipeline for snoop returns. This separate pipeline would not be blocked behind coherent requests. This invention also repartitions the master initiated traffic to move cache evictions (both with and without data) and non-coherent writes to the new non-blocking channel. This non-blocking pipeline removes the need for any coherent requests to complete before the snoop request can reach the memory controller. Repartitioning cache initiated evictions to the non-blocking pipeline prevents deadlock when snoop and eviction occur concurrently. The non-blocking channel of this invention combines snoop responses from memory controller initiated requests and master initiated evictions/non-coherent writes. | 04-24-2014 |
20140115273 | DISTRIBUTED DATA RETURN BUFFER FOR COHERENCE SYSTEM WITH SPECULATIVE ADDRESS SUPPORT - The MSMC (Multicore Shared Memory Controller) described is a module designed to manage traffic between multiple processor cores, other mastering peripherals or DMA, and the EMIF (External Memory InterFace)in a multicore SoC. Each processor has an associated return buffer allowing out of order responses of memory read data and cache snoop responses to ensure maximum bandwidth at the endpoints, and all endpoints receive status messages to simplify the return queue. | 04-24-2014 |
20140115279 | Multi-Master Cache Coherent Speculation Aware Memory Controller with Advanced Arbitration, Virtualization and EDC - This invention is an integrated memory controller/interconnect that provides very high bandwidth access to both on-chip memory and externally connected off-chip memory. This invention includes an arbitration for all memory endpoints including priority, fairness, and starvation bounds; virtualization; and error detection and correction hardware to protect the on-chip SRAM banks including automated scrubbing. | 04-24-2014 |
20140143486 | FLEXIBLE ARBITRATION SCHEME FOR MULTI ENDPOINT ATOMIC ACCESSES IN MULTICORE SYSTEMS - The MSMC (Multicore Shared Memory Controller) described is a module designed to manage traffic between multiple processor cores, other mastering peripherals or DMA, and the EMIF (External Memory InterFace) in a multicore SoC. The invention unifies all transaction sizes belonging to a slave previous to arbitrating the transactions in order to reduce the complexity of the arbitration process and to provide optimum bandwidth management among all masters. Two consecutive slots are assigned per cache line access to automatically guarantee the atomicity of all transactions within a single cache line. The need for synchronization among all the banks of a particular SRAM is eliminated, as synchronization is accomplished by assigning back to back slots. | 05-22-2014 |
20140143849 | Secure Master and Secure Guest Endpoint Security Firewall - This invention is a security firewall having a security hierarchy including: secure master (SM); secure guest (SG); and non-secure (NS). There is one secure master and n secure guests. The firewall includes one secure region for secure master and one secure region for secure guests. The SM region only allows access from the secure master and the SG region allows accesses from any secure transaction. Finally, the non-secure region can be implemented two ways. In a first option, non-secure regions may be accessed only upon non-secure transactions. In a second option, non-secure regions may be accessed any processing core. In this second option, the access is downgraded to a non-secure access if the security identity is secure master or secure guest. If the two security levels are not needed the secure master can unlock the SM region to allow any secure guest access to the SM region. | 05-22-2014 |
20140149690 | Multi-Processor, Multi-Domain, Multi-Protocol Cache Coherent Speculation Aware Shared Memory Controller and Interconnect - This invention combines a multicore shared memory controller and an asynchronous protocol converting bridge to create a very efficient heterogeneous multi-processor system. After traversing the protocol converting bridge the commands travel through the regular processor port. This allows the interconnect to remain unchanged while having any combination of different processors connected. This invention tightly integrates all of the processors into the same memory controller/interconnect. | 05-29-2014 |
20140156951 | MULTICORE, MULTIBANK, FULLY CONCURRENT COHERENCE CONTROLLER - This invention optimizes non-shared accesses and avoids dependencies across coherent endpoints to ensure bandwidth across the system even when sharing. The coherence controller is distributed across all coherent endpoints. The coherence controller for each memory endpoint keeps a state around for each coherent access to ensure the proper ordering of events. The coherence controller of this invention uses First-In-First-Out allocation to ensure full utilization of the resources before stalling and simplicity of implementation. The coherence controller provides Snoop Command/Response ID Allocation per memory endpoint. | 06-05-2014 |
20150019840 | Highly Integrated Scalable, Flexible DSP Megamodule Architecture - This invention addresses implements a range of interesting technologies into a single block. Each DSP CPU has a streaming engine. The streaming engines include: a SE to L2 interface that can request 512 bits/cycle from L2; a loose binding between SE and L2 interface, to allow a single stream to peak at 1024 bits/cycle; one-way coherence where the SE sees all earlier writes cached in system, but not writes that occur after stream opens; full protection against single-bit data errors within its internal storage via single-bit parity with semi-automatic restart on parity error. | 01-15-2015 |