Skip to content
Navigation Menu
Sign in
Appearance settings
Platform
AI CODE CREATION
GitHub Copilot
Write better code with AI
GitHub Copilot app
Direct agents from issue to merge
MCP Registry
Integrate external tools
DEVELOPER WORKFLOWS
Actions
Automate any workflow
Codespaces
Instant dev environments
Issues
Plan and track work
Code Review
Manage code changes
Code Quality
Enforce quality at merge
APPLICATION SECURITY
GitHub Advanced Security
Find and fix vulnerabilities
Code security
Secure your code as you build
Secret protection
Stop leaks before they start
EXPLORE
Why GitHub
Documentation
Blog
Changelog
Marketplace
View all features
Solutions
BY COMPANY SIZE
Enterprises
Small and medium teams
Startups
Nonprofits
BY USE CASE
App Modernization
DevSecOps
DevOps
CI/CD
View all use cases
BY INDUSTRY
Healthcare
Financial services
Manufacturing
Government
View all industries
View all solutions
Resources
EXPLORE BY TOPIC
AI
Software Development
DevOps
Security
View all topics
EXPLORE BY TYPE
Customer stories
Events & webinars
Ebooks & reports
Business insights
GitHub Skills
SUPPORT & SERVICES
Documentation
Customer support
Community forum
Trust center
Partners
View all resources
Open Source
COMMUNITY
GitHub Sponsors
Fund open source developers
PROGRAMS
Security Lab
Maintainer Community
GitHub Stars
Archive Program
REPOSITORIES
Topics
Trending
Collections
Enterprise
ENTERPRISE SOLUTIONS
Enterprise platform
AI-powered developer platform
AVAILABLE ADD-ONS
GitHub Advanced Security
Enterprise-grade security features
Copilot for Business
Enterprise-grade AI features
Premium Support
Enterprise-grade 24/7 support
Pricing
Search
/
Sign in
Sign up
Appearance settings
You signed in with another tab or window.
Reload
to refresh your session.
You signed out in another tab or window.
Reload
to refresh your session.
You switched accounts on another tab or window.
Reload
to refresh your session.
Dismiss alert
{{ message }}
dotsimulate
/
StreamDiffusion
Public
forked from
cumulo-autumn/StreamDiffusion
Notifications
You must be signed in to change notification settings
Fork
8
Star
24
Code
Pull requests
1
Actions
Projects
Security and quality
0
Insights
Additional navigation options
Code
Pull requests
Actions
Projects
Security and quality
Insights
chore(installer): overhaul install scripts — portable paths, TRT verification, pin alignment
- #9
#9
Closed
forkni
wants to merge 43 commits into
SDTD_031_dev
dotsimulate/StreamDiffusion:SDTD_031_dev
from
SDTD_v3-dev-Alex-synced
dotsimulate/StreamDiffusion:SDTD_v3-dev-Alex-synced
Copy head branch name to clipboard
Conversation
Commits
43
(43)
Checks
Files changed
Closed
chore(installer): overhaul install scripts — portable paths, TRT verification, pin alignment
#9
forkni
wants to merge 43 commits into
SDTD_031_dev
dotsimulate/StreamDiffusion:SDTD_031_dev
from
SDTD_v3-dev-Alex-synced
dotsimulate/StreamDiffusion:SDTD_v3-dev-Alex-synced
Copy head branch name to clipboard
Commits
Commits on Mar 29, 2026
Fix SDXL TensorRT engine build failure on Windows
Show description for a6082cb
dotsimulate
committed
a6082cb
View commit details
Copy full SHA for a6082cb
Browse repository at this point
Commits on Mar 31, 2026
feat: clone decode_image output to prevent TRT VAE buffer reuse
Show description for d71085b
INTER-NYC
and
claude
committed
d71085b
View commit details
Copy full SHA for d71085b
Browse repository at this point
perf: Tier 2 inference performance optimizations
Show description for 28491ce
INTER-NYC
and
claude
committed
28491ce
View commit details
Copy full SHA for 28491ce
Browse repository at this point
feat: auto-resolve IP-Adapter model paths based on detected architecture
Show description for fb293ba
INTER-NYC
and
claude
committed
fb293ba
View commit details
Copy full SHA for fb293ba
Browse repository at this point
Commits on Apr 1, 2026
Fix IP-Adapter crash on SD2.1 models (sd-turbo) due to non-existent ip-adapter_sd21.bin
Show description for 7d29e08
INTER-NYC
and
claude
committed
7d29e08
View commit details
Copy full SHA for 7d29e08
Browse repository at this point
Add cuda-python 13.x compatibility fix for cudart import
Show description for 117cbbc
INTER-NYC
and
claude
committed
117cbbc
View commit details
Copy full SHA for 117cbbc
Browse repository at this point
Quick-win CUDA optimizations: pre-allocated buffers + L2 cache persistence
Show description for bd9a2d3
INTER-NYC
and
claude
committed
bd9a2d3
View commit details
Copy full SHA for bd9a2d3
Browse repository at this point
Fix hardcoded float16 autocasts and add fp32 precision for scheduler division
Show description for 85307e8
INTER-NYC
and
claude
committed
85307e8
View commit details
Copy full SHA for 85307e8
Browse repository at this point
Fix L2 cache: second reserve call was resetting reservation to 0
Show description for 1128d97
INTER-NYC
and
claude
committed
1128d97
View commit details
Copy full SHA for 1128d97
Browse repository at this point
fix: report inference FPS separately from output FPS when similar image filter is active
Show description for 7bf2366
INTER-NYC
and
claude
committed
7bf2366
View commit details
Copy full SHA for 7bf2366
Browse repository at this point
VRAM reduction: text encoder offloading + max_batch_size 4→2
Show description for 9c22342
INTER-NYC
and
claude
committed
9c22342
View commit details
Copy full SHA for 9c22342
Browse repository at this point
Revert max_batch_size 4→2: unsafe for cfg_type=full/initialize
Show description for 1efb9ef
INTER-NYC
and
claude
committed
1efb9ef
View commit details
Copy full SHA for 1efb9ef
Browse repository at this point
Commits on Apr 2, 2026
perf: pre-allocate image output buffers, replace .clone() with .copy_() in pipeline hot path
INTER-NYC
committed
697b548
View commit details
Copy full SHA for 697b548
Browse repository at this point
fix: use opencv-contrib-python 4.9.0.80 and add FP8 deps (modelopt, cupy 13.x)
INTER-NYC
committed
818b03d
View commit details
Copy full SHA for 818b03d
Browse repository at this point
fix: pin onnx 1.17.0, onnxruntime-gpu 1.22.0; remove CPU onnxruntime co-install
INTER-NYC
committed
aa21e73
View commit details
Copy full SHA for aa21e73
Browse repository at this point
fix: bump onnx 1.18.0 + onnxruntime-gpu 1.24.3 (modelopt FLOAT4E2M1 + IR 11)
INTER-NYC
committed
2b6c0aa
View commit details
Copy full SHA for 2b6c0aa
Browse repository at this point
fix: patch ByteSize() for >2GB ONNX in modelopt FP8 quantization
INTER-NYC
committed
600c5bf
View commit details
Copy full SHA for 600c5bf
Browse repository at this point
fix: reduce FP8 calibration batches 128→8 (KVO cache OOM, 281GB→17GB)
INTER-NYC
committed
847be93
View commit details
Copy full SHA for 847be93
Browse repository at this point
fix: export UNet ONNX at opset 19 when FP8 enabled to skip modelopt version converter
INTER-NYC
committed
00cf0c7
View commit details
Copy full SHA for 00cf0c7
Browse repository at this point
fix: merge calibration list-of-dicts into stacked dict for modelopt CalibrationDataProvider
INTER-NYC
committed
9e22ea9
View commit details
Copy full SHA for 9e22ea9
Browse repository at this point
fix: add NVIDIA DLLs to PATH and retry without quantize_mha on ORT EP failure
INTER-NYC
committed
519069f
View commit details
Copy full SHA for 519069f
Browse repository at this point
fix: use single calibration batch for modelopt (avoid rank mismatch), cleanup intermediates on retry
INTER-NYC
committed
cfca95b
View commit details
Copy full SHA for cfca95b
Browse repository at this point
Commits on Apr 3, 2026
fix: resolve 4 FP8 quantization bugs for TRT 10.12 cached attention engine build
Show description for ccecf37
INTER-NYC
and
claude
committed
ccecf37
View commit details
Copy full SHA for ccecf37
Browse repository at this point
Commits on Apr 4, 2026
perf: add CUDA/PyTorch env var tuning and cudnn.benchmark
INTER-NYC
committed
0f50188
View commit details
Copy full SHA for 0f50188
Browse repository at this point
fix: prevent FP8 engine build intermediate file bloat on Windows
Show description for 18fc5ed
INTER-NYC
and
claude
committed
18fc5ed
View commit details
Copy full SHA for 18fc5ed
Browse repository at this point
fix(l2-cache): add TRT activation caching path to setup_l2_persistence
Show description for 888d20a
INTER-NYC
and
claude
committed
888d20a
View commit details
Copy full SHA for 888d20a
Browse repository at this point
perf: clean up deprecated TRT 10.x API usage in engine builder and preprocessing
INTER-NYC
committed
72f6409
View commit details
Copy full SHA for 72f6409
Browse repository at this point
fix: clamp TRT persistent_cache_limit to L2_cache_size//2 to avoid exceeding hardware max
INTER-NYC
committed
07093bf
View commit details
Copy full SHA for 07093bf
Browse repository at this point
fix(fp8): remove direct_io_types/simplify, make allocate_buffers FP8-safe
Show description for 5dc8af4
INTER-NYC
and
claude
committed
5dc8af4
View commit details
Copy full SHA for 5dc8af4
Browse repository at this point
Commits on Apr 5, 2026
perf(trt): static spatial shapes + tactic cleanup for engine builder
Show description for 95d34b8
INTER-NYC
and
claude
committed
95d34b8
View commit details
Copy full SHA for 95d34b8
Browse repository at this point
perf: Tier 1 hot-path allocation elimination (Phase A-C)
Show description for fe85327
INTER-NYC
committed
fe85327
View commit details
Copy full SHA for fe85327
Browse repository at this point
perf: skip text encoder reload on identical prompt; seed FPS EMA from first frame
INTER-NYC
committed
cd9b6ec
View commit details
Copy full SHA for cd9b6ec
Browse repository at this point
fix: guard ControlNet TRT engine compilation behind acceleration check
INTER-NYC
committed
1e6b351
View commit details
Copy full SHA for 1e6b351
Browse repository at this point
fix: remove empty_cache() from text encoder offload to prevent prompt-change stutter
INTER-NYC
committed
b75d15b
View commit details
Copy full SHA for b75d15b
Browse repository at this point
perf: keep text encoders on GPU during inference; add force_offload for quantization
INTER-NYC
committed
e548b40
View commit details
Copy full SHA for e548b40
Browse repository at this point
perf(trt): fully static batch profiles to unlock l2tc on UNet
Show description for 2ed2996
INTER-NYC
and
claude
committed
2ed2996
View commit details
Copy full SHA for 2ed2996
Browse repository at this point
feat(trt): add TRT profiling infrastructure gated by STREAMDIFFUSION_PROFILE_TRT
Show description for 0f9d1d6
INTER-NYC
and
claude
committed
0f9d1d6
View commit details
Copy full SHA for 0f9d1d6
Browse repository at this point
perf(trt): reduce builder_optimization_level from 4 to 3 for static shapes
Show description for 791bd26
INTER-NYC
and
claude
committed
791bd26
View commit details
Copy full SHA for 791bd26
Browse repository at this point
revert(trt): restore builder_optimization_level=4; tactic 0x3e9 is a TRT 10.12 bug
Show description for 3a44259
INTER-NYC
and
claude
committed
3a44259
View commit details
Copy full SHA for 3a44259
Browse repository at this point
fix(trt): guard aten::copy behind _use_prealloc to unblock ONNX export with use_cached_attn
Show description for f1fc4bf
INTER-NYC
and
claude
committed
f1fc4bf
View commit details
Copy full SHA for f1fc4bf
Browse repository at this point
fix(controlnet): pass pipeline resolution to TRT engine builder
Show description for 8d75198
INTER-NYC
and
claude
committed
8d75198
View commit details
Copy full SHA for 8d75198
Browse repository at this point
perf(controlnet): enable CUDA graphs for ControlNet TRT engine
Show description for 0454830
INTER-NYC
and
claude
committed
0454830
View commit details
Copy full SHA for 0454830
Browse repository at this point
Commits on Apr 7, 2026
chore(installer): overhaul install scripts — pins, portability, TRT verification
Show description for b0eeab2
INTER-NYC
committed
b0eeab2
View commit details
Copy full SHA for b0eeab2
Browse repository at this point
You can’t perform that action at this time.
