TimeGraph: GPU scheduling for real-time multi-tasking environments
Citations
265 citations
Cites methods from "TimeGraph: GPU scheduling for real-..."
...Proposed techniques have been successfully applied to CPU management [10], [7], [6] and more recently to GPU management [14], [15]....
[...]
256 citations
191 citations
Cites background from "TimeGraph: GPU scheduling for real-..."
...Both the scheduling framework and scheduling policies are implement in hardware to avoid the long latency of issuing commands to the GPU [17]....
[...]
...Issues with GPU sharing, such as priority inversion and no fairness, have already been noticed by operating systems [30, 17, 18, 27] and real-time [16, 6] research communities....
[...]
...Because the latency of issuing a command to the GPU is significant [17], commands are sent to the GPU as soon as possible....
[...]
...GERM [7] and TimeGraph [17] focus on graphics applications and provide a GPU command schedulers integrated in the device driver....
[...]
190 citations
Cites background or methods from "TimeGraph: GPU scheduling for real-..."
...As discussed in previous work [15], it is very hard to analyze GPU commands and recognize the corresponding API calls in the OS....
[...]
...Gdev uses a similar scheme to TimeGraph [15] for GPU scheduling....
[...]
...Although we make use of some previous techniques [14, 15], Gdev...
[...]
...GPU Resource Management: TimeGraph [15] and GERM [2] provide a GPU command-driven scheduler integrated in the device driver....
[...]
...We further plan to integrate the configuration of priority and reserve for each single task into /proc, using the TimeGraph approach [15]....
[...]
160 citations
References
5,397 citations
517 citations
"TimeGraph: GPU scheduling for real-..." refers background or methods in this paper
...The amount of limit is computed by a traditional resource-reservation model based onC andT of each reserve [26]....
[...]
...CPU Scheduling: TimeGraph shares the concept of priority and reservation, which has been well-studied by the real-time systems community [13, 26], but there is a fundamental difference from these traditional studies in that TimeGraph is designed to address an arbitrarilyarriving non-preemptive GPU execution model, whereas the real-time systems community has often considered a periodic preemptive CPU execution model....
[...]
...It should be noted that this enforcement mechanism is very different from traditional CPU reservation mechanisms [20, 26] that use timers or ticks to suspend tasks, since GPU command groups are non-preemptive, and hence we need to perform enforcement at GPU command group boundary....
[...]
...replenishment used in real-time systems [20, 26]....
[...]
438 citations
"TimeGraph: GPU scheduling for real-..." refers background in this paper
...Several bandwidth-preserving approaches [12, 29, 30] for an arbitrarily-arriving model exist, but a non-preemptive model has not been much studied yet....
[...]
359 citations
303 citations