Abstract: A computer-implemented method for managing cache memory in a distributed symmetric multiprocessing computer is described. The method may include receiving, at a first central processor (CP) chip, a fetch request from a first chip. The method may further include determining via address compare mechanisms on the first CP chip whether one or more of a second CP chip and a third CP chip is requesting access to a target line. The first chip, the second chip, and the third chip are within the same chip cluster. The method further includes providing access to the target line if both of the second CP chip and the third CP chip have accessed the target line at least one time since the first CP chip has accessed the target line.

Abstract: In an approach for managing data transfer across a bus shared by processors, a request for a first set of data is received from a first processor. A request for a second set of data is received from a second processor. First portions of the first set of data and the second set of data are written to a buffer. Additional portions of each set of data are written to the buffer as portions are received. It is determined that a portion of the first set of data has a higher priority to the bus than a portion of the second set of data based on a priority scheme, wherein the priority scheme is based on return progress of each respective set of data having at least a portion of data in the buffer. The portion of the first set of data is granted access to the bus.

Abstract: Methods, systems, and computer program products for managing broadcasts in a distributed symmetric multiprocessing computer are provided. Aspects include defining a default broadcast rate for a plurality of processors in the distributed symmetric multiprocessing computer, wherein the default broadcast rate is a rate at which a processor broadcasts a request for a resource. The one or more broadcasted requests by a first processor are monitored and related responses are utilized to determine a state of the one or more broadcasted requests. The default broadcast rate is adjusted based at least in part on the state of the one or more broadcasted requests.

Abstract: A computer-implemented method for managing cache memory in a distributed symmetric multiprocessing computer is described. The method may include receiving, at a first central processor (CP) chip, a fetch request from a first chip. The method may further include determining via address compare mechanisms on the first CP chip whether one or more of a second CP chip and a third CP chip is requesting access to a target line. The first chip, the second chip, and the third chip are within the same chip cluster. The method further includes providing access to the target line if both of the second CP chip and the third CP chip have accessed the target line at least one time since the first CP chip has accessed the target line.

Abstract: In an approach for managing data transfer across a bus shared by processors, a request for a first set of data is received from a first processor. A request for a second set of data is received from a second processor. First portions of the first set of data and the second set of data are written to a buffer. Additional portions of each set of data are written to the buffer as portions are received. It is determined that a portion of the first set of data has a higher priority to the bus than a portion of the second set of data based on a priority scheme, wherein the priority scheme is based on return progress of each respective set of data having at least a portion of data in the buffer. The portion of the first set of data is granted access to the bus.

Abstract: In an approach for managing data transfer across a bus shared by processors, a request for a first set of data is received from a first processor. A request for a second set of data is received from a second processor. First portions of the first set of data and the second set of data are written to a buffer. Additional portions of each set of data are written to the buffer as portions are received. It is determined that a portion of the first set of data has a higher priority to the bus than a portion of the second set of data based on a priority scheme, wherein the priority scheme is based on return progress of each respective set of data having at least a portion of data in the buffer. The portion of the first set of data is granted access to the bus.

Abstract: In an approach for taking corrupt portions of cache offline during runtime, a notification of a section of a cache to be taken offline is received, wherein the section includes one or more sets in one or more indexes of the cache. An indication is associated with each set of the one or more sets in a first index of the one or more indexes, wherein the indication marks the respective set as unusable for future operations. Data is purged from the one or more sets in the first index of the cache. Each set of the one or more sets in the first index is marked as invalid.

Abstract: In an approach for taking corrupt portions of cache offline during runtime, a notification of a section of a cache to be taken offline is received, wherein the section includes one or more sets in one or more indexes of the cache. An indication is associated with each set of the one or more sets in a first index of the one or more indexes, wherein the indication marks the respective set as unusable for future operations. Data is purged from the one or more sets in the first index of the cache. Each set of the one or more sets in the first index is marked as invalid.

Abstract: A technique is provided for accumulating failures. A failure of a first row is detected in a group of array macros, the first row having first row address values. A mask has mask bits corresponding to each of the first row address values. The mask bits are initially in active status. A failure of a second row, having second row address values, is detected. When none of the first row address values matches the second row address values, and when mask bits are all in the active status, the array macros are determined to be bad. When at least one of the first row address values matches the second row address values, mask bits that correspond to at least one of the first row address values that match are kept in active status, and mask bits that correspond to non-matching first address values are set to inactive status.

Abstract: An aspect includes receiving a fetch request for a data block at a cache memory system that includes cache memory that is partitioned into a plurality of cache data ways including a cache data way that contains the data block. The data block is fetched and it is determined whether the in-line ECC checking and correcting should be bypassed. The determining is based on a bypass indicator corresponding to the cache data way. Based on determining that in-line ECC checking and correcting should be bypassed, returning the fetched data block to the requestor and performing an ECC process for the fetched data block subsequent to returning the fetched data block to the requestor. Based on determining that in-line ECC checking and correcting should not be bypassed, performing the ECC process for the fetched data block and returning the fetched data block to the requestor subsequent to performing the ECC process.

Abstract: An aspect includes receiving a fetch request for a data block at a cache memory system that includes cache memory that is partitioned into a plurality of cache data ways including a cache data way that contains the data block. The data block is fetched and it is determined whether the in-line ECC checking and correcting should be bypassed. The determining is based on a bypass indicator corresponding to the cache data way. Based on determining that in-line ECC checking and correcting should be bypassed, returning the fetched data block to the requestor and performing an ECC process for the fetched data block subsequent to returning the fetched data block to the requestor. Based on determining that in-line ECC checking and correcting should not be bypassed, performing the ECC process for the fetched data block and returning the fetched data block to the requestor subsequent to performing the ECC process.

Abstract: In an approach for backing up designated data located in a cache, data stored within an index of a cache is identified, wherein the data has an associated designation indicating that the data is applicable to be backed up to a higher level memory. It is determined that the data stored to the cache has been updated. A status associated with the data is adjusted, such that the adjusted status indicates that the data stored to the cache has not been changed. A copy of the data is created. The copy of the data is stored to the higher level memory.

Abstract: Main memory operation in a symmetric multiprocessing computer, the computer comprising one or more processors operatively coupled through a cache controller to at least one cache of main memory, the main memory shared among the processors, the computer further comprising input/output (‘I/O’) resources, including receiving, in the cache controller from an issuing resource, a memory instruction for a memory address, the memory instruction requiring writing data to main memory; locking by the cache controller the memory address against further memory operations for the memory address; advising the issuing resource of completion of the memory instruction before the memory instruction completes in main memory; issuing by the cache controller the memory instruction to main memory; and unlocking the memory address only after completion of the memory instruction in main memory.

Abstract: A technique is provided for accumulating failures. A failure of a first row is detected in a group of array macros, the first row having first row address values. A mask has mask bits corresponding to each of the first row address values. The mask bits are initially in active status. A failure of a second row, having second row address values, is detected. When none of the first row address values matches the second row address values, and when mask bits are all in the active status, the array macros are determined to be bad. When at least one of the first row address values matches the second row address values, mask bits that correspond to at least one of the first row address values that match are kept in active status, and mask bits that correspond to non-matching first address values are set to inactive status.

Abstract: In an approach for taking corrupt portions of cache offline during runtime, a notification of a section of a cache to be taken offline is received, wherein the section includes one or more sets in one or more indexes of the cache. An indication is associated with each set of the one or more sets in a first index of the one or more indexes, wherein the indication marks the respective set as unusable for future operations. Data is purged from the one or more sets in the first index of the cache. Each set of the one or more sets in the first index is marked as invalid.

Abstract: A technique is provided for accumulating failures. A failure of a first row is detected in a group of array macros, the first row having first row address values. A mask has mask bits corresponding to each of the first row address values. The mask bits are initially in active status. A failure of a second row, having second row address values, is detected. When none of the first row address values matches the second row address values, and when mask bits are all in the active status, the array macros are determined to be bad. When at least one of the first row address values matches the second row address values, mask bits that correspond to at least one of the first row address values that match are kept in active status, and mask bits that correspond to non-matching first address values are set to inactive status.

Abstract: In an approach for taking corrupt portions of cache offline during runtime, a notification of a section of a cache to be taken offline is received, wherein the section includes one or more sets in one or more indexes of the cache. An indication is associated with each set of the one or more sets in a first index of the one or more indexes, wherein the indication marks the respective set as unusable for future operations. Data is purged from the one or more sets in the first index of the cache. Each set of the one or more sets in the first index is marked as invalid.

Abstract: A method, system, and/or computer program product for dynamic array masking is provided. Dynamic array masking includes, during execution of computer instructions that access a cache memory, detecting an error condition in a portion of the cache memory. The portion of the cache memory contains an array macro. Dynamic array masking, during the execution of the computer instructions that access a cache memory, further includes dynamically setting mask bits to indicate the error condition in the portion of the cache memory and preventing subsequent writes to the portion of the cache memory in accordance with the dynamically set mask bits. Embodiments also include evicting cache entries from the portion of the cache memory. This evicting can include performing a cache purge operation for the cache entries corresponding to the dynamically set mask bits.

Abstract: A technique is provided for accumulating failures. A failure of a first row is detected in a group of array macros, the first row having first row address values. A mask has mask bits corresponding to each of the first row address values. The mask bits are initially in active status. A failure of a second row, having second row address values, is detected. When none of the first row address values matches the second row address values, and when mask bits are all in the active status, the array macros are determined to be bad. When at least one of the first row address values matches the second row address values, mask bits that correspond to at least one of the first row address values that match are kept in active status, and mask bits that correspond to non-matching first address values are set to inactive status.

Abstract: A technique is provided for accumulating failures. A failure of a first row is detected in a group of array macros, the first row having first row address values. A mask has mask bits corresponding to each of the first row address values. The mask bits are initially in active status. A failure of a second row, having second row address values, is detected. When none of the first row address values matches the second row address values, and when mask bits are all in the active status, the array macros are determined to be bad. When at least one of the first row address values matches the second row address values, mask bits that correspond to at least one of the first row address values that match are kept in active status, and mask bits that correspond to non-matching first address values are set to inactive status.