This is the second article in a series of articles around Rusts new async/await
feature. The first article about interfaces can be found
here.
In this part of the series we want to a look at a mechanism which behaves very
different in Rust than in all other languages which feature async/await
support. This mechanism is Cancellation.
Out of the box, my SMB performance on macOS 12.3.1 would top out at around 20MB/s in short ~5 second bursts, which was absolutely horrendous, slow to navigate in Finder and slugish to interact with.
Since making these changes, I now get sustained ~80-100MB/s+ and instant Finder navigation which is superb and how things should be out-of-the-box (OOTB)!
May 2023 update: As of Ventura, the SMB issues were just horribly inconsistent and hard to maintain. Something in the combination of Unraid, macOS and SMB just doesn't play nice. I ended up binning NFS/SMB all together and heading to a locally hosted Nextcloud instance for file syncing, then using SFTP/Ansible Git flow for editing files within appdata.
Out of the box, my SMB performance on macOS 12.3.1 would top out at around 20MB/s in short ~5 second bursts, which was absolutely horrendous, slow to navigate in Finder and slugish to interact with.
Since making these changes, I now get sustained ~80-100MB/s+ and instant Finder navigation which is superb and how things should be out-of-the-box (OOTB)!
May 2023 update: As of Ventura, the SMB issues were just horribly inconsistent and hard to maintain. Something in the combination of Unraid, macOS and SMB just doesn't play nice. I ended up binning NFS/SMB all together and heading to a locally hosted Nextcloud instance for file syncing, then using SFTP/Ansible Git flow for editing files within appdata.
| Building for debugging... | |
| &&& Malformed rewrite rule: τ_0_0._Storage.[Sequence:Element].[_AdjacencyListPerVertex:EdgeData].[_AdjacencyListPerEdge:VertexId] => τ_0_0.[DirectedAdjacencyListProtocol:RawId] | |
| &&& symbol.getKind() != Symbol::Kind::Name | |
| Rewrite system: { | |
| - [DirectedAdjacencyListProtocol].[DirectedAdjacencyListProtocol] => [DirectedAdjacencyListProtocol] [permanent] | |
| - [DirectedAdjacencyListProtocol].RawId => [DirectedAdjacencyListProtocol:RawId] [permanent] | |
| - [DirectedAdjacencyListProtocol]._EdgeData => [DirectedAdjacencyListProtocol:_EdgeData] [permanent] | |
| - [DirectedAdjacencyListProtocol]._VertexData => [DirectedAdjacencyListProtocol:_VertexData] [permanent] | |
| - [DirectedAdjacencyListProtocol].VertexCollection => [DirectedAdjacencyListProtocol:VertexCollection] [permanent] |
| // pullback of BatchNorm.callAsFunction(_:) | |
| sil private @$s10TensorFlow9BatchNormV14callAsFunctionyAA0A0VyxGAGFAA0aB13FloatingPointRzlTJpUSpSr : $@convention(thin) <τ_0_0 where τ_0_0 : TensorFlowFloatingPoint> (@guaranteed Tensor<τ_0_0>, @owned _AD__$s10TensorFlow9BatchNormV14callAsFunctionyAA0A0VyxGAGF_bb6__PB__src_0_wrt_1_10TensorFlow0aB13FloatingPointRzl<τ_0_0>) -> @out BatchNorm<τ_0_0>.TangentVector { | |
| // %0 // user: %440 | |
| // %1 // users: %211, %144 | |
| // %2 // user: %70 | |
| bb0(%0 : $*BatchNorm<τ_0_0>.TangentVector, %1 : $Tensor<τ_0_0>, %2 : $_AD__$s10TensorFlow9BatchNormV14callAsFunctionyAA0A0VyxGAGF_bb6__PB__src_0_wrt_1_10TensorFlow0aB13FloatingPointRzl<τ_0_0>): | |
| %3 = metatype $@thin Tensor<τ_0_0>.Type // users: %238, %236, %219, %218, %418, %417, %414, %413, %346, %344, %340, %339, %337, %334, %330, %329, %304, %303, %294, %293, %171, %169, %152, %151, %126, %125, %121, % |
| import dearpygui.dearpygui as dpg | |
| dpg.enable_docking(dock_space=True) | |
| import numpy as np | |
| width = 800 | |
| height = 800 | |
| grid_arr = np.ones((width, height, 4), dtype=np.float32) | |
| rng = np.random.default_rng() |
| // This script helps you automatically fill each page of 1099 form when you have a consolidated 1099 | |
| // Just paste in the console each time you fill a new page | |
| $("#__employer_details_employer_name").val("ABC INC") | |
| $("#__employer_details_employer_address").val("ADDR AVE NE") | |
| $("#__employer_details_employer_city").val("Palo Alto") | |
| $("#__employer_details_employer_state").val("CA") | |
| $("#__employer_details_employer_zip_code").val("943011") | |
| $("#payers_federal_id").val("38-xxxxxxx") | |
| $("#statecode").val("CA") |
| let 🉑 = true; | |
| class 🤣 { | |
| let 💦: Int; | |
| init() { | |
| self.💦 = 1; | |
| } | |
| init(_ 💦: Int) { |
| swift build -Xcc -I/usr/include/tensorflow | |
| Updating https://github.com/saeta/penguin.git | |
| Updating https://github.com/dan-zheng/tensorboardx-s4tf.git | |
| Updating https://github.com/apple/swift-tools-support-core.git | |
| Updating https://github.com/tensorflow/swift-apis | |
| Updating https://github.com/apple/swift-numerics | |
| Updating https://github.com/pvieito/PythonKit.git | |
| Updating https://github.com/google/swift-benchmark.git | |
| Updating https://github.com/yaslab/CSV.swift.git | |
| Updating https://github.com/vojtamolda/Plotly.swift |
| 0.010068689239966268, 0.0016086790724621075, -0.0312696013842796 | |
| 0.7447114712141675, 0.06205963228562672, -0.5573445530921713 | |
| 1.409740980049445, 0.08655957953753346, -1.030334601927387 | |
| 2.233505212825677, 0.15366358844952932, -1.6774906770770837 | |
| 3.008689347558227, 0.28343220532336866, -2.2535724334656333 | |
| 3.898977471592672, 0.33876676042302156, -2.7418953233705428 | |
| 4.764607669726627, 0.47728574474871466, -3.489205029230695 | |
| 5.505793867917641, 0.5567107296492837, -3.9723234574013357 | |
| 6.315244378420935, 0.5955888262764261, -4.561362956443775 | |
| 6.994184690768557, 0.7089548521732609, -5.251374427565788 |
