Abstract: A technology can be provided for managing shared memory used by a plurality of compute nodes. An example system can include a shared globally addressable memory to enable access to shared data by the plurality of compute nodes. A memory interface can process memory requests sent to the shared globally addressable memory from the plurality of processors. A memory write module can be included for the memory interface to allocate memory locations in the shared globally addressable memory and write read-only data to the globally addressable memory from a writing compute node. In addition, a read module for the memory interface can map read-only data in the globally addressable shared memory as read-only for subsequent accesses by the plurality of compute nodes.

Abstract: A single application can be executed across multiple execution environments in an efficient manner if at least a relevant portion of the virtual memory assigned to the application was equally accessible by each of the multiple execution environments. A request by a process in one execution environment can, thereby, be directed to an operating system, or other core software, in another execution environment and can be made by a shadow of the requesting process in the same manner as the original request was made by the requesting process itself. Because of the memory invariance between the execution environments, the results of the request will be equally accessible to the original requesting process even though the underlying software that responded to the request may be executing in a different execution environment. A similar thread invariance can be maintained to provide for accurate translation of requests between execution environments.

Abstract: A processor chip may have a built-in hardware lock and deterministic exclusive locking of the hardware lock by execution units executing in parallel on the chip. A set of software locks may be maintained, where the execution units set and release the software locks only by first acquiring a lock of the hardware lock. A first execution unit sets a software lock after acquiring a lock of the hardware lock, and other execution units, even if exclusively locking the hardware lock, are unable to lock the software lock until after the first execution unit has reacquired a lock of the hardware lock and possibly released the software lock while exclusively locking the hardware lock. An execution unit may release a software lock after and while holding a lock of the hardware lock. The hardware lock is released when a software lock has been set or released.

Abstract: A technology can be provided for managing shared memory used by a plurality of compute nodes. An example system can include a shared globally addressable memory to enable access to shared data by the plurality of compute nodes. A memory interface can process memory requests sent to the shared globally addressable memory from the plurality of processors. A memory write module can be included for the memory interface to allocate memory locations in the shared globally addressable memory and write read-only data to the globally addressable memory from a writing compute node. In addition, a read module for the memory interface can map read-only data in the globally addressable shared memory as read-only for subsequent accesses by the plurality of compute nodes.

Abstract: Various technologies and techniques are disclosed that provide fast context switching. One embodiment provides a method for a context switch comprising preloading a host virtual machine context in a first portion of a processor, operating a guest virtual machine in a second portion of the processor, writing parameters of the host virtual machine context to a memory location shared by the host virtual machine and the guest virtual machine, and operating the host virtual machine in the processor. In this manner, a fast context switch may be accomplished by preloading the new context in a virtual processor, thus reducing the delay to switch to the new context.

Abstract: A processor chip may have a built-in hardware lock and deterministic exclusive locking of the hardware lock by execution units executing in parallel on the chip. A set of software locks may be maintained, where the execution units set and release the software locks only by first acquiring a lock of the hardware lock. A first execution unit sets a software lock after acquiring a lock of the hardware lock, and other execution units, even if exclusively locking the hardware lock, are unable to lock the software lock until after the first execution unit has reacquired a lock of the hardware lock and possibly released the software lock while exclusively locking the hardware lock. An execution unit may release a software lock after and while holding a lock of the hardware lock. The hardware lock is released when a software lock has been set or released.

Abstract: A single application can be executed across multiple execution environments in an efficient manner if at least a relevant portion of the virtual memory assigned to the application was equally accessible by each of the multiple execution environments. A request by a process in one execution environment can, thereby, be directed to an operating system, or other core software, in another execution environment and can be made by a shadow of the requesting process in the same manner as the original request was made by the requesting process itself. Because of the memory invariance between the execution environments, the results of the request will be equally accessible to the original requesting process even though the underlying software that responded to the request may be executing in a different execution environment. A similar thread invariance can be maintained to provide for accurate translation of requests between execution environments.

Abstract: A single application can be executed across multiple execution environments in an efficient manner if at least a relevant portion of the virtual memory assigned to the application was equally accessible by each of the multiple execution environments. A request by a process in one execution environment can, thereby, be directed to an operating system, or other core software, in another execution environment and can be made by a shadow of the requesting process in the same manner as the original request was made by the requesting process itself. Because of the memory invariance between the execution environments, the results of the request will be equally accessible to the original requesting process even though the underlying software that responded to the request may be executing in a different execution environment. A similar thread invariance can be maintained to provide for accurate translation of requests between execution environments.

Abstract: A single application can be executed across multiple execution environments in an efficient manner if at least a relevant portion of the virtual memory assigned to the application was equally accessible by each of the multiple execution environments. A request by a process in one execution environment can, thereby, be directed to an operating system, or other core software, in another execution environment and can be made by a shadow of the requesting process in the same manner as the original request was made by the requesting process itself. Because of the memory invariance between the execution environments, the results of the request will be equally accessible to the original requesting process even though the underlying software that responded to the request may be executing in a different execution environment. A similar thread invariance can be maintained to provide for accurate translation of requests between execution environments.

Abstract: Various technologies and techniques are disclosed that provide fast context switching. One embodiment provides a method for a context switch comprising preloading a host virtual machine context in a first portion of a processor, operating a guest virtual machine in a second portion of the processor, writing parameters of the host virtual machine context to a memory location shared by the host virtual machine and the guest virtual machine, and operating the host virtual machine in the processor. In this manner, a fast context switch may be accomplished by preloading the new context in a virtual processor, thus reducing the delay to switch to the new context.