Commit 9e1966b7 by Leyuan Wang Committed by Tianqi Chen

barrier fence added for warp mem (#1174)

parent f7d9d7e8
...@@ -454,6 +454,7 @@ def build(sch, ...@@ -454,6 +454,7 @@ def build(sch,
if current_build_config().detect_global_barrier: if current_build_config().detect_global_barrier:
func = ir_pass.ThreadSync(func, "global") func = ir_pass.ThreadSync(func, "global")
func = ir_pass.ThreadSync(func, "shared") func = ir_pass.ThreadSync(func, "shared")
func = ir_pass.ThreadSync(func, "warp")
warp_size = target.thread_warp_size warp_size = target.thread_warp_size
func = ir_pass.LowerThreadAllreduce(func, warp_size) func = ir_pass.LowerThreadAllreduce(func, warp_size)
fsplits = [s for s in ir_pass.SplitHostDevice(func)] fsplits = [s for s in ir_pass.SplitHostDevice(func)]
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment