Revert "Synchronize using MTLSharedEvents"

[christiangnrd]

Reverts #633

The MTLCaptureManager failures seems to be related. Not much info online about how stopCapture works but it seems like the capture manager's isCapturing property does not change to false until the command buffer status is MTLCommandBufferStatusCompleted, which the new synchronization via events method does not wait for.

Will reland after determining how to deal with this for capturing.

[github-actions]

Your PR requires formatting changes to meet the project's style guidelines.
Please consider running Runic (git runic main) to apply these changes.

Click here to view the suggested changes.

diff --git a/src/state.jl b/src/state.jl
index 3a0512e5..8f2c9399 100644
--- a/src/state.jl
+++ b/src/state.jl
@@ -65,7 +65,7 @@ Create a new MTLCommandBuffer from the global command queue, commit it to the qu
 and simply wait for it to be completed. Since command buffers *should* execute in a
 First-In-First-Out manner, this synchronizes the GPU.
 """
-@autoreleasepool function synchronize(queue::MTLCommandQueue=global_queue(device()))
+@autoreleasepool function synchronize(queue::MTLCommandQueue = global_queue(device()))
     cmdbuf = MTLCommandBuffer(queue)
     commit!(cmdbuf)
     wait_completed(cmdbuf)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 80.61%. Comparing base (1942968) to head (a52e795).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #645      +/-   ##
==========================================
- Coverage   80.63%   80.61%   -0.03%     
==========================================
  Files          61       61              
  Lines        2722     2713       -9     
==========================================
- Hits         2195     2187       -8     
+ Misses        527      526       -1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[github-actions]

Metal Benchmarks

Benchmark suite	Current: a52e795	Previous: 1942968	Ratio
latency/precompile	9944983208 ns	9844653958 ns	1.01
latency/ttfp	4027831958 ns	3972040229 ns	1.01
latency/import	1293257604.5 ns	1275530958.5 ns	1.01
integration/metaldevrt	906749.5 ns	828500 ns	1.09
integration/byval/slices=1	1646000 ns	1536750 ns	1.07
integration/byval/slices=3	19605000 ns	9632625 ns	2.04
integration/byval/reference	1626604 ns	1543583 ns	1.05
integration/byval/slices=2	2759708 ns	2621958.5 ns	1.05
kernel/indexing	525292 ns	567792 ns	0.93
kernel/indexing_checked	557541 ns	569292 ns	0.98
kernel/launch	9333 ns	9208 ns	1.01
array/construct	6042 ns	6625 ns	0.91
array/broadcast	529708 ns	583375 ns	0.91
array/random/randn/Float32	907500 ns	784333 ns	1.16
array/random/randn!/Float32	597958 ns	623250 ns	0.96
array/random/rand!/Int64	552875 ns	547458 ns	1.01
array/random/rand!/Float32	553250 ns	585291 ns	0.95
array/random/rand/Int64	1016375 ns	771250 ns	1.32
array/random/rand/Float32	786395.5 ns	622687 ns	1.26
array/accumulate/Int64/1d	1382250 ns	1277104.5 ns	1.08
array/accumulate/Int64/dims=1	1936458.5 ns	1868333 ns	1.04
array/accumulate/Int64/dims=2	2322604.5 ns	2183625 ns	1.06
array/accumulate/Int64/dims=1L	12289250 ns	11737104 ns	1.05
array/accumulate/Int64/dims=2L	10154916.5 ns	9771416.5 ns	1.04
array/accumulate/Float32/1d	1190125 ns	1142833 ns	1.04
array/accumulate/Float32/dims=1	1672625 ns	1570458 ns	1.07
array/accumulate/Float32/dims=2	2093771 ns	1931625 ns	1.08
array/accumulate/Float32/dims=1L	10500250 ns	9864375 ns	1.06
array/accumulate/Float32/dims=2L	7549250 ns	7308021 ns	1.03
array/reductions/reduce/Int64/1d	1280583 ns	1373353.5 ns	0.93
array/reductions/reduce/Int64/dims=1	1176959 ns	1069291.5 ns	1.10
array/reductions/reduce/Int64/dims=2	1320709 ns	1193292 ns	1.11
array/reductions/reduce/Int64/dims=1L	2223417 ns	2113062.5 ns	1.05
array/reductions/reduce/Int64/dims=2L	3594292 ns	3456458 ns	1.04
array/reductions/reduce/Float32/1d	772083 ns	971625 ns	0.79
array/reductions/reduce/Float32/dims=1	848833 ns	808458 ns	1.05
array/reductions/reduce/Float32/dims=2	760354 ns	768979 ns	0.99
array/reductions/reduce/Float32/dims=1L	1852834 ns	1739041 ns	1.07
array/reductions/reduce/Float32/dims=2L	1904437.5 ns	1772125 ns	1.07
array/reductions/mapreduce/Int64/1d	1302542 ns	1456146 ns	0.89
array/reductions/mapreduce/Int64/dims=1	1179292 ns	1074875 ns	1.10
array/reductions/mapreduce/Int64/dims=2	1302542 ns	1206417 ns	1.08
array/reductions/mapreduce/Int64/dims=1L	2228459 ns	2119292 ns	1.05
array/reductions/mapreduce/Int64/dims=2L	3585125 ns	3444375 ns	1.04
array/reductions/mapreduce/Float32/1d	779687.5 ns	990792 ns	0.79
array/reductions/mapreduce/Float32/dims=1	855833.5 ns	810062.5 ns	1.06
array/reductions/mapreduce/Float32/dims=2	754750 ns	761104 ns	0.99
array/reductions/mapreduce/Float32/dims=1L	1856917 ns	1740812.5 ns	1.07
array/reductions/mapreduce/Float32/dims=2L	1908416 ns	1781292 ns	1.07
array/private/copyto!/gpu_to_gpu	563833 ns	651375 ns	0.87
array/private/copyto!/cpu_to_gpu	705541.5 ns	805542 ns	0.88
array/private/copyto!/gpu_to_cpu	764792 ns	817667 ns	0.94
array/private/iteration/findall/int	1641312.5 ns	1646500 ns	1.00
array/private/iteration/findall/bool	1558625 ns	1444584 ns	1.08
array/private/iteration/findfirst/int	1793292 ns	1754958.5 ns	1.02
array/private/iteration/findfirst/bool	1733750 ns	1703625 ns	1.02
array/private/iteration/scalar	3396333 ns	4772500 ns	0.71
array/private/iteration/logical	2720896 ns	2536917 ns	1.07
array/private/iteration/findmin/1d	1772875 ns	1815666 ns	0.98
array/private/iteration/findmin/2d	1549021 ns	1431750 ns	1.08
array/private/copy	804542 ns	538167 ns	1.49
array/shared/copyto!/gpu_to_gpu	84625 ns	86375 ns	0.98
array/shared/copyto!/cpu_to_gpu	79584 ns	86583 ns	0.92
array/shared/copyto!/gpu_to_cpu	80083 ns	84833 ns	0.94
array/shared/iteration/findall/int	1645292 ns	1609874.5 ns	1.02
array/shared/iteration/findall/bool	1549937.5 ns	1464354 ns	1.06
array/shared/iteration/findfirst/int	1466208 ns	1377750 ns	1.06
array/shared/iteration/findfirst/bool	1411916.5 ns	1319166 ns	1.07
array/shared/iteration/scalar	160125 ns	217500 ns	0.74
array/shared/iteration/logical	2508708.5 ns	2288708.5 ns	1.10
array/shared/iteration/findmin/1d	1438625 ns	1421750 ns	1.01
array/shared/iteration/findmin/2d	1546874.5 ns	1430854.5 ns	1.08
array/shared/copy	210917 ns	248666 ns	0.85
array/permutedims/4d	2597333 ns	2438438 ns	1.07
array/permutedims/2d	1288417 ns	1193250 ns	1.08
array/permutedims/3d	1945750.5 ns	1768458 ns	1.10
metal/synchronization/stream	14875 ns	19916 ns	0.75
metal/synchronization/context	15854.5 ns	20375 ns	0.78

This comment was automatically generated by workflow using github-action-benchmark.