Update cache.jl

[jarbus]

This seemed like a bug

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 73.54%. Comparing base (972f3f0) to head (7328d3f).
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2604      +/-   ##
==========================================
- Coverage   73.63%   73.54%   -0.10%     
==========================================
  Files         157      157              
  Lines       15207    15207              
==========================================
- Hits        11198    11184      -14     
- Misses       4009     4023      +14     

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

[github-actions]

CUDA.jl Benchmarks

Benchmark suite	Current: 7328d3f	Previous: 62564aa	Ratio
latency/precompile	45354733130 ns	45352888677 ns	1.00
latency/ttfp	6382441349.5 ns	6365236644.5 ns	1.00
latency/import	3033778244 ns	3028318014.5 ns	1.00
integration/volumerhs	9568316 ns	9566694 ns	1.00
integration/byval/slices=1	146684 ns	146977 ns	1.00
integration/byval/slices=3	425436 ns	425622 ns	1.00
integration/byval/reference	144736 ns	144997 ns	1.00
integration/byval/slices=2	286563 ns	286278 ns	1.00
integration/cudadevrt	103507 ns	103635 ns	1.00
kernel/indexing	14284 ns	14502 ns	0.98
kernel/indexing_checked	14901 ns	15491 ns	0.96
kernel/occupancy	717.431654676259 ns	685.9934210526316 ns	1.05
kernel/launch	2153.777777777778 ns	2211.5555555555557 ns	0.97
kernel/rand	18127 ns	15495 ns	1.17
array/reverse/1d	19518 ns	19288.5 ns	1.01
array/reverse/2d	23540 ns	25407 ns	0.93
array/reverse/1d_inplace	10128 ns	11091 ns	0.91
array/reverse/2d_inplace	11486 ns	12732 ns	0.90
array/copy	20470 ns	20971 ns	0.98
array/iteration/findall/int	158142 ns	159562 ns	0.99
array/iteration/findall/bool	138436 ns	139396 ns	0.99
array/iteration/findfirst/int	153157 ns	153501 ns	1.00
array/iteration/findfirst/bool	154954 ns	154850 ns	1.00
array/iteration/scalar	78049 ns	75791 ns	1.03
array/iteration/logical	215615.5 ns	216533 ns	1.00
array/iteration/findmin/1d	40922 ns	41336 ns	0.99
array/iteration/findmin/2d	94290 ns	94735 ns	1.00
array/reductions/reduce/1d	35579 ns	42301 ns	0.84
array/reductions/reduce/2d	50761.5 ns	45473.5 ns	1.12
array/reductions/mapreduce/1d	33633 ns	40507 ns	0.83
array/reductions/mapreduce/2d	41890.5 ns	47968 ns	0.87
array/broadcast	21627 ns	21824 ns	0.99
array/copyto!/gpu_to_gpu	11595 ns	13701 ns	0.85
array/copyto!/cpu_to_gpu	213648 ns	213728 ns	1.00
array/copyto!/gpu_to_cpu	245025 ns	245756 ns	1.00
array/accumulate/1d	108333 ns	108365 ns	1.00
array/accumulate/2d	79477 ns	80195 ns	0.99
array/construct	1184.75 ns	1204.7 ns	0.98
array/random/randn/Float32	43906 ns	43838 ns	1.00
array/random/randn!/Float32	26619 ns	26350 ns	1.01
array/random/rand!/Int64	27193 ns	27179 ns	1.00
array/random/rand!/Float32	8810.333333333334 ns	8916.333333333334 ns	0.99
array/random/rand/Int64	29829 ns	30059 ns	0.99
array/random/rand/Float32	12959 ns	12950 ns	1.00
array/permutedims/4d	67098 ns	67259 ns	1.00
array/permutedims/2d	56667 ns	56558.5 ns	1.00
array/permutedims/3d	59183 ns	59379 ns	1.00
array/sorting/1d	2932018 ns	2932110 ns	1.00
array/sorting/by	3500113 ns	3498470 ns	1.00
array/sorting/2d	1084817.5 ns	1085408 ns	1.00
cuda/synchronization/stream/auto	1042.6 ns	1025.5 ns	1.02
cuda/synchronization/stream/nonblocking	6466.8 ns	6545.6 ns	0.99
cuda/synchronization/stream/blocking	855.7872340425532 ns	796.0645161290323 ns	1.08
cuda/synchronization/context/auto	1198.6 ns	1190.9 ns	1.01
cuda/synchronization/context/nonblocking	6659.6 ns	6739.6 ns	0.99
cuda/synchronization/context/blocking	948.0487804878048 ns	903.25 ns	1.05

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

[maleadt]

Good catch, thanks!