Skip to content

Apply the allocator gc-gate fix to SD21/SDXL #25

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
TroyHernandez wants to merge 2 commits into from
Closed

Apply the allocator gc-gate fix to SD21/SDXL #25

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
415c444
Apply the allocator gc-gate fix to the SD21/SDXL loader
TroyHernandez Jul 8, 2026
c03a703
Bump version to 0.1.0.4
TroyHernandez Jul 8, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion DESCRIPTION
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,6 +1,6 @@
Package: diffuseR
Title: Functional Interface to Diffusion Models in R
Version: 0.1.0.3
Version: 0.1.0.4
Authors@R: c(
person("Troy", "Hernandez", email = "troy@cornball.ai", role = c("aut", "cre"),
comment = c(ORCID = "0009-0005-4248-604X")),
Expand Down
11 changes: 11 additions & 0 deletions R/load_pipeline.R
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -35,6 +35,17 @@ load_pipeline <- function(model_name, m2d, i2i = FALSE, unet_dtype_str,
device_cpu <- m2d$device_cpu
device_cuda <- m2d$device_cuda

if (any(unlist(devices) == "cuda")) {
# Same chronic-allocator-gate fix as the FLUX/Z-Image pipelines
# (see .flux_gc_gates). Measured at 50 steps (native modules,
# decoder on CPU, RTX 5060 Ti 16 GB), warm generation:
# SDXL 1024^2 234 -> 100 s (gc 209 -> 75 s); SD 2.1 768^2
# 101 -> 26 s (gc 88 -> 15 s). Note the reduced gc cadence lets
# the garbage sawtooth ride higher (SD21 peak 5.4 -> 13.3 GB);
# cards much smaller than 16 GB may need different gate sizing.
.flux_gc_gates(footprint_gb = 12)
}

pipeline <- list()
# Load models into the environment
if (i2i) {
Expand Down