diff options
| author | Tony Gutierrez <anthony.gutierrez@amd.com> | 2016-10-26 22:48:28 -0400 |
|---|---|---|
| committer | Tony Gutierrez <anthony.gutierrez@amd.com> | 2016-10-26 22:48:28 -0400 |
| commit | de72e36619350f9b3e3a3dc8de63b490c4cecf2d (patch) | |
| tree | 8bb5b723e8d3cec80d56bc61ebdc6141b1330b39 /src/gpu-compute/compute_unit.cc | |
| parent | b63eb1302b006682bd227a5e236f7b3b95e9b8e8 (diff) | |
| download | gem5-de72e36619350f9b3e3a3dc8de63b490c4cecf2d.tar.xz | |
gpu-compute: support in-order data delivery in GM pipe
this patch adds an ordered response buffer to the GM pipeline
to ensure in-order data delivery. the buffer is implemented as
a stl ordered map, which sorts the request in program order by
using their sequence ID. when requests return to the GM pipeline
they are marked as done. only the oldest request may be serviced
from the ordered buffer, and only if is marked as done.
the FIFO response buffers are kept and used in OoO delivery mode
Diffstat (limited to 'src/gpu-compute/compute_unit.cc')
| -rw-r--r-- | src/gpu-compute/compute_unit.cc | 12 |
1 files changed, 1 insertions, 11 deletions
diff --git a/src/gpu-compute/compute_unit.cc b/src/gpu-compute/compute_unit.cc index 93cffbe1e..ffa5243d2 100644 --- a/src/gpu-compute/compute_unit.cc +++ b/src/gpu-compute/compute_unit.cc @@ -1033,17 +1033,7 @@ ComputeUnit::DataPort::MemRespEvent::process() if (gpuDynInst->n_reg > MAX_REGS_FOR_NON_VEC_MEM_INST) gpuDynInst->statusVector.clear(); - if (gpuDynInst->isLoad() || gpuDynInst->isAtomic()) { - assert(compute_unit->globalMemoryPipe.isGMLdRespFIFOWrRdy()); - - compute_unit->globalMemoryPipe.getGMLdRespFIFO() - .push(gpuDynInst); - } else { - assert(compute_unit->globalMemoryPipe.isGMStRespFIFOWrRdy()); - - compute_unit->globalMemoryPipe.getGMStRespFIFO() - .push(gpuDynInst); - } + compute_unit->globalMemoryPipe.handleResponse(gpuDynInst); DPRINTF(GPUMem, "CU%d: WF[%d][%d]: packet totally complete\n", compute_unit->cu_id, gpuDynInst->simdId, |