1. A method of orchestrated shuffling of data in a non-uniform memory access device that includes a plurality of processing nodes, the method comprising:
running an application on a plurality of threads executing on the plurality of processing nodes, wherein running the application includes dividing data on each thread into partitions according to a target thread on which the data is to be processed, and the plurality of processing nodes are connected to each other by interconnects;
identifying, by the threads, data to be shuffled from source threads running on source processing nodes among the processing nodes to target threads running on target processing nodes among the processing nodes;
generating a plan for orchestrating shuffling of the data among a plurality of memory devices associated with the plurality of processing nodes and for simultaneously transmitting data over different interconnects to a plurality of different target processing nodes from a plurality of different source processing nodes, the plan including utilizing a data-shifting table to identify an order in which the data partitions are to be transferred from the source threads of the source processing nodes to the target threads of the target processing nodes;
shuffling the data among the plurality of memory devices associated with the plurality of processing nodes based on the plan by simultaneously transmitting data partitions from the plurality of source threads to the plurality of target threads according to the data-shifting table;
shifting the data-shifting table to associate each source thread with a different target thread; and
transmitting another set of data partitions from the plurality of source threads to the plurality of target threads based on shifting the data-shifting table.
2. The method of claim 1, wherein the data includes operand data and operational state data of the source threads.
3. The method of claim 1, wherein at least two of the processing nodes are connected to separate local memory devices and to each other, such that each processing node is capable of accessing data from a first local memory device via a direct interconnect and is capable of accessing data from a second local memory device via another processing node.
4. The method of claim 1, wherein the plan for orchestrating the shuffling of data corresponds to a first ring including separate segments for each separate data partition and a second ring located inside the first ring including separate segments for each separate processing node, and
shifting the instruction execution table includes rotating the first ring with respect to the second ring.
5. A non-transitory computer-readable medium having stored therein an instruction-execution table that defines an association of a plurality of data partitions with a plurality of processing nodes, the non-transitory computer-readable medium including instructions that, when executed by one or more processors, controls the one or more processors to perform a method of orchestrated data shuffling, the method comprising:
running an application on a plurality of threads executing on the plurality of processing nodes, wherein running the application includes dividing data on each thread into partitions according to a target thread on which the data is to be processed and the plurality of processing nodes are connected to each other by interconnects;
identifying, by the threads, data to be shuffled from source threads running on source processing nodes among the processing nodes to target threads running on target processing nodes among the processing nodes;
generating a plan for orchestrating shuffling of the data among a plurality of memory devices associated with the plurality of processing nodes and for simultaneously transmitting data over different interconnects to a plurality of different target processing nodes from a plurality of different source processing nodes, the plan including utilizing a data-shifting table to identify an order in which the data partitions are to be transferred from the source threads of the source processing nodes to the target threads of the target processing nodes;
shuffling the data among the plurality of memory devices associated with the plurality of processing nodes based on the plan by simultaneously transmitting data partitions from the plurality of source threads to the plurality of target threads according to the data-shifting table;
shifting the data-shifting table to associate each source thread with a different target thread; and
transmitting another set of data partitions from the plurality of source threads to the plurality of target threads based on shifting the data-shifting table.
6. The non-transitory computer-readable medium of claim 5, wherein the data includes operand data and operational state data of the source threads.
7. The non-transitory computer-readable medium of claim 5, wherein at least two of the processing nodes are connected to separate local memory devices and to each other, such that each processing node is capable of accessing data from a first local memory device via a direct interconnect and is capable of accessing data from a second local memory device via another processing node.
8. The non-transitory computer-readable medium of claim 5, wherein the plan for orchestrating the shuffling of data corresponds to a first ring including separate segments for each separate data partition and a second ring located inside the first ring including separate segments for each separate processing node, and
shifting the instruction execution table includes rotating the first ring with respect to the second ring.
9. A non-uniform memory access system, comprising:
a plurality of processing nodes including processing circuitry to execute instructions;
a plurality of local memory modules, at least one local memory module connected directly to at least one first processing node, and the at least one local memory module connected to at least one second processing node only indirectly via the at least one first processing node,
wherein the plurality of processing nodes is configured perform a data-shuffling process, comprising:
running an application on a plurality of threads executing on the plurality of processing nodes, wherein running the application includes dividing data on each thread into partitions according to a target thread on which the data is to be processed and the plurality of processing nodes are connected to each other by interconnects;
identifying, by the threads, data to be shuffled from source threads running on source processing nodes among the processing nodes to target threads running on target processing nodes among the processing nodes;
generating a plan for orchestrating shuffling of the data among a plurality of memory devices associated with the plurality of processing nodes and for simultaneously transmitting data over different interconnects to a plurality of different target processing nodes from a plurality of different source processing nodes, the plan including utilizing a data-shifting table to identify an order in which the data partitions are to be transferred from the source threads of the source processing nodes to the target threads of the target processing nodes;
shuffling the data among the plurality of memory devices associated with the plurality of processing nodes based on the plan by simultaneously transmitting data partitions from the plurality of source threads to the plurality of target threads according to the data-shifting table;
shifting the data-shifting table to associate each source thread with a different target thread; and
transmitting another set of data partitions from the plurality of source threads to the plurality of target threads based on shifting the data-shifting table.
10. The system of claim 9, wherein the data includes operand data and operational state data of the source threads.
11. The system of claim 9, wherein at least two of the processing nodes are connected to separate local memory devices and to each other, such that each processing node is capable of accessing data from a first local memory device via a direct interconnect and is capable of accessing data from a second local memory device via another processing node.
12. The system of claim 9, wherein the plan for orchestrating the shuffling of data corresponds to a first ring including separate segments for each separate data partition and a second ring located inside the first ring including separate segments for each separate processing node, and
shifting the instruction execution table includes rotating the first ring with respect to the second ring.
The claims below are in addition to those above.
All refrences to claim(s) which appear below refer to the numbering after this setence.
1. Rotatable tool for chip removing machining, comprising a basic body rotatable about a geometrical center axis, and a replaceable cutting part which is rigidly connectable to an axially front end of the basic body by a malefemale coupling; the coupling comprising a groove formed in a front end of the basic body, and a male part insertable into the groove and protruding axially rearwardly from the cutting part; the male part comprising a front base portion which is delimited by a pair of opposite, first flank surfaces, as well as a rear wedge portion which is narrower than the base portion and delimited by a pair of opposite, second flank surfaces; a forwardly open slot being formed in the front part of the basic body, which slot communicates with the groove and separates two elastically deflectable legs of the basic body which clamp the male part in the groove; the groove comprising two axially separated front and rear spaces; the front space mouthing at a free front end surface of the basic body and delimited by a pair of first, opposite side surfaces; the rear space being delimited by a second pair of opposite side surfaces; at least one of the two second flank surfaces of said male part being inclined at a first acute angle in relation to the center axis in a radially outward and axially rearward direction; one of the second side surfaces being disposed at the rear space of the groove and being inclined in relation to the center axis at a second acute angle in a direction radially inward and axially forward; the rear space of the groove forming a jaw having a variable minimum width in a radial plane oriented perpendicular to the center axis; the wedge portion of the male part having a maximum width in a radial plane; the legs being elastically deflectable away from one another, wherein the minimum width of the groove upon deflection of the legs away from one another being larger than the maximum width of the wedge portion of the male part; the first and second acute angles being equally large to create surface contact between the flank surfaces and the side surfaces when the legs resiliently spring back against the male part from their deflected state.
2. The tool according to claim 1 wherein the two flank surfaces of the wedge portion and the two side surfaces of the rear space are inclined at the same angle of inclination to the center axis.
3. The tool according to claim 2 wherein the angles of inclination are at least 76 degrees.
4. The tool according to claim 2 wherein said angles of inclination are at most 81 degrees.
5. The tool according to claim 1 further including a centering protrusion on a rear end surface of the wedge portion of the cutting part, the protrusion having a truncated conical shape and a smallest diameter and a largest diameter; the smallest diameter being smaller than the diameter of a front limiting edge of a rotationally symmetrical seating formed in a bottom surface of the groove; the largest diameter being larger than the diameter of the front limiting edge.
6. The tool according to claim 1 wherein said inclined side surface transforms via an edge line into a second side surface.
7. The tool according to claim 6 wherein two inclined side surfaces transform via a respective edge line into a respective second side surface.
8. The tool according to claim 6 wherein the second side surface is inclined in relation to the center axis at a third acute angle which is smaller than said second angle.
9. The tool according to claim 8 wherein said second side surface is inclined in a direction radially outwardly and axially forwardly.