starred/terminal
1
0
Fork 0
mirror of https://github.com/microsoft/terminal.git synced 2026-09-24 07:54:35 +00:00
Code Issues 19.0k Packages Projects Releases 20 Wiki Activity

[Performance] Replace std::vector with more lightweight structure in RenderData::GetPatternId #14403

New Issue
Open
opened 2026-01-31 04:09:21 +00:00 by claunia · 8 comments
claunia commented 2026-01-31 04:09:21 +00:00
Owner
Copy Link

Originally created by @skyline75489 on GitHub (Jul 6, 2021).

This is about the scrolling test in #10563. A simplified repro would be:

time bash -c 'yes | head -n1000000'

which takes 7-8s with windows terminal and 300ms with gnome terminal.

Test case provided by @lhecker.

These lines show up in the hot paths of the trace of OpenConsole:

https://github.com/microsoft/terminal/blob/main/src/renderer/base/renderer.cpp#L788

image

It appears that simply constructing empty std::vector (empty because this is in conhost/OpenConsole) is quite expensive.

CC @PankajBhojwani

Originally created by @skyline75489 on GitHub (Jul 6, 2021). This is about the scrolling test in #10563. A simplified repro would be: ```bash time bash -c 'yes | head -n1000000' ``` which takes 7-8s with windows terminal and 300ms with gnome terminal. Test case provided by @lhecker. These lines show up in the hot paths of the trace of `OpenConsole`: https://github.com/microsoft/terminal/blob/main/src/renderer/base/renderer.cpp#L788 ![image](https://user-images.githubusercontent.com/4710575/124551976-7fac6600-de65-11eb-892c-982f5e8ab85b.png) It appears that simply constructing empty `std::vector` (empty because this is in conhost/OpenConsole) is quite expensive. CC @PankajBhojwani
claunia added the Help WantedArea-RenderingIssue-TaskProduct-MetaArea-Performance labels 2026-01-31 04:09:21 +00:00
claunia commented 2026-01-31 04:09:22 +00:00
Author
Owner
Copy Link

@skyline75489 commented on GitHub (Jul 6, 2021):

Proposed solution:

Seem to me the number of possible pattern ids returned should be limited. For example if we limit it to max 2 pattern ids, we can use a simple structure:

struct {
    size_t primaryId;
    size_t secondaryId;
} PatternIds
@skyline75489 commented on GitHub (Jul 6, 2021): Proposed solution: Seem to me the number of possible pattern ids returned should be limited. For example if we limit it to max 2 pattern ids, we can use a simple structure: ```c++ struct { size_t primaryId; size_t secondaryId; } PatternIds ```
claunia commented 2026-01-31 04:09:22 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Jul 6, 2021):

Seem to me the number of possible pattern ids returned should be limited.

Why's that? Somebody could conceivably have many such IDs.

Last time we saw a vector being a performance drain we switched it to an unordered_set. There's also boost::small_vector, which we could set up with space for maybe one or two patterns?

@DHowett commented on GitHub (Jul 6, 2021): > Seem to me the number of possible pattern ids returned should be limited. Why's that? Somebody could conceivably have _many_ such IDs. Last time we saw a vector being a performance drain we switched it to an unordered_set. There's also `boost::small_vector`, which we could set up with space for maybe one or two patterns?
claunia commented 2026-01-31 04:09:23 +00:00
Author
Owner
Copy Link

@skyline75489 commented on GitHub (Jul 6, 2021):

Sure. I am ok with multiple pattern ids, as long as the performance drain can be fixed.

获取 Outlook for iOShttps://aka.ms/o0ukef

@skyline75489 commented on GitHub (Jul 6, 2021): Sure. I am ok with multiple pattern ids, as long as the performance drain can be fixed. 获取 Outlook for iOS<https://aka.ms/o0ukef>
claunia commented 2026-01-31 04:09:23 +00:00
Author
Owner
Copy Link

@skyline75489 commented on GitHub (Jul 9, 2021):

I've tried boost::small_vector, it's better but still not good:

image

I'm beginning to wonder if there's a better way to structure the code to reduce the amount of query for GetPatternId.

@skyline75489 commented on GitHub (Jul 9, 2021): I've tried `boost::small_vector`, it's better but still not good: ![image](https://user-images.githubusercontent.com/4710575/125026084-7403ae00-e0b6-11eb-8c65-16638ca308c1.png) I'm beginning to wonder if there's a better way to structure the code to reduce the amount of query for `GetPatternId`.
claunia commented 2026-01-31 04:09:23 +00:00
Author
Owner
Copy Link

@skyline75489 commented on GitHub (Jul 9, 2021):

On my god. I just realized this isn't about std::vector or not. In OpenConsole this call _pData->GetPatternId() is out-of-proc ?! Nope this is wrong.

I'm so confused that even I changed it to return a single integer, that call is still very expensive.

Now I get it. It's because I have Control Flow Guard (CFG) enabled, which adds additonal __guard_dispatch_icall_fptr instructions and thus increase the costs of method call. No matter what I do, as long as it's a _pData->Something(), it will be expensive.

@skyline75489 commented on GitHub (Jul 9, 2021): <del>On my god. I just realized this isn't about std::vector or not. In OpenConsole this call `_pData->GetPatternId()` is out-of-proc ?!</del> Nope this is wrong. <del>I'm so confused that even I changed it to return a single integer, that call is still very expensive.</del> Now I get it. It's because I have Control Flow Guard (CFG) enabled, which adds additonal `__guard_dispatch_icall_fptr` instructions and thus increase the costs of *method call*. No matter what I do, as long as it's a `_pData->Something()`, it will be expensive.
claunia commented 2026-01-31 04:09:24 +00:00
Author
Owner
Copy Link

@skyline75489 commented on GitHub (Jul 9, 2021):

Closed in favor of #10596 .

@skyline75489 commented on GitHub (Jul 9, 2021): Closed in favor of #10596 .
claunia commented 2026-01-31 04:09:24 +00:00
Author
Owner
Copy Link

@DHowett commented on GitHub (Jul 9, 2021):

It is worth tracking this separately, as it impacts both the dx and vt renderers. @Lhecker should write up his thoughts here

@DHowett commented on GitHub (Jul 9, 2021): It is worth tracking this separately, as it impacts both the dx and vt renderers. @Lhecker should write up his thoughts here
claunia commented 2026-01-31 04:09:24 +00:00
Author
Owner
Copy Link

@skyline75489 commented on GitHub (Jul 15, 2021):

Yea @DHowett is right. Even if we ignore the VT renderer part (with future refactoring work, this is possible), DX renderer still suffers from this particular issue, when the load is very heavy.

@skyline75489 commented on GitHub (Jul 15, 2021): Yea @DHowett is right. Even if we ignore the VT renderer part (with future refactoring work, this is possible), DX renderer still suffers from this particular issue, when the load is very heavy.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
⛺ Reserved
A11yCO
A11yMAS
A11ySev1
A11ySev2
A11ySev3
A11yTTValidated
A11yUsable
A11yVoiceAccess
A11yWCAG
Area-Accessibility
Area-AtlasEngine
Area-AzureShell
Area-Build
Area-Build
Area-Chat
Area-CmdPal
Area-CodeHealth
Area-Commandline
Area-CookedRead
Area-DefApp
Area-Extensibility
Area-Fonts
Area-GroupPolicy
Area-i18n
Area-Input
Area-Interaction
Area-Interop
Area-Localization
Area-Output
Area-Performance
Area-Portable
Area-Quality
Area-Remoting
Area-Rendering
Area-Schema
Area-Server
Area-Settings
Area-SettingsUI
Area-ShellExtension
Area-ShellExtension
Area-ShellExtension
Area-Suggestions
Area-Suggestions
Area-TerminalConnection
Area-TerminalControl
Area-Theming
Area-UserInterface
Area-VT
Area-Windowing
Area-WPFControl
AutoMerge
Blocking-Ingestion
Culprit-Centennial
Culprit-WinUI
Disability-All
Disability-Blind
Disability-LowVision
Disability-Mobility
External-Blocked-WinUI3
Fixed
Gathering-Data
good first issue
HCL-E+D
HCL-WindowsTerminal
Help Wanted
Impact-Compatibility
Impact-Compliance
Impact-Correctness
Impact-Visual
In-PR
InclusionBacklog
InclusionBacklog-Windows TerminalWin32
InclusionCommitted-202206
Issue-Bug
Issue-Docs
Issue-Feature
Issue-Feature
Issue-Question
Issue-Samples
Issue-Scenario
Issue-Task
Needs-Attention
Needs-Author-Feedback
Needs-Bisect
Needs-Discussion
Needs-Repro
Needs-Tag-Fix
Needs-Tag-Fix
Needs-Triage
No-Recent-Activity
Priority-0
Priority-1
Priority-2
Priority-3
Product-Cmd.exe
Product-Colortool
Product-Colortool
Product-Colortool
Product-Conhost
Product-Conpty
Product-Meta
Product-Powershell
Product-Terminal
Product-WSL
pull-request
Mirrored from GitHub Pull Request
 Resolution-Answered
Resolution-By-Design
Resolution-Duplicate
Resolution-External
Resolution-Fix-Available
Resolution-Fix-Committed
Resolution-No-Repro
Resolution-Won't-Fix
Severity-Blocking
Severity-Crash
Severity-DataLoss
spam
this-will-be-a-breaking-change
Tracking-External
WindowsTerminal_Win32
Work-Item
zAskModeBug
zInbox-Bug
No Label Area-Performance Area-Rendering Help Wanted Issue-Task Product-Meta
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
claunia
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: starred/terminal#14403
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?