turnip_faux123_v1.2.3 vs turnip_faux123_v1.2.0

On Adreno (TM) 830 (AYN Odin 3). The first driver named is compared against the second, which is the baseline.

Stockturnip_faux123_v1.2.0Mesa 26.3.0-devel (git-dd45fd0baa)

Customturnip_faux123_v1.2.3Mesa 26.3.0-devel (git-dd45fd0baa)

Graphics system version (Vulkan)1.4.359 to 1.4.359

Optional features (extensions)172 to 172+0 -0

Features that pass their check (validated)66 to 704 pass only on turnip_faux123_v1.2.3 0 pass only on turnip_faux123_v1.2.0

Speed test scenes (benchmark)4 faster 0 slower14% faster overall (geometric mean)

Games each driver could run (an estimate from its feature list): green likely, yellow maybe, orange limited, red no

StockDX92002+DX102006+DX112009+DX122015+Switch

CustomDX92002+DX102006+DX112009+DX122015+Switch

compute_nbody_2x3% faster
hdr_2160p32% faster
oit_linked_lists4% faster
pipeline_barriers_2160p19% faster

Custom driver against stock, per scene. Bars to the right are faster, to the left slower.

Not on equal terms: the GPU clock differed during 2 of these scenes (for example compute_nbody_2x: stock at 97% of the device's maximum, custom at 90%). Part of the frame-time difference comes from the clock the device chose, not from the driver.

Not compared: 2 scenes with a paced frame rate (pipeline_barriers, subpasses). At least one of the two drivers was held at a fixed frame rate there, so the two times do not measure the same thing.

turnip_faux123_v1.2.0 details turnip_faux123_v1.2.3 details

Feature checks that differ (4)

Feature	turnip_faux123_v1.2.0	turnip_faux123_v1.2.3	Change
perf.fb_local_barrier_large_draws core	untested	pass	gained
perf.fb_local_barrier_overdraw core	untested	pass	gained
stress.scratch_per_batch core	untested	pass	gained
stress.spill_helper_lanes core	untested	pass	gained

Speed test scenes (benchmark)

Scene	turnip_faux123_v1.2.0 ms	turnip_faux123_v1.2.3 ms	Difference	Starts	GPU clock, MHz	Picture vs stock
compute_nbody	2.80 typical 2.70	4.10 typical 3.99	too light, see compute_nbody_2x	3 of 3 vs 3 of 3	1100 vs 660 not comparable	no large difference 0.0 no small difference worst 0.0 64x36
compute_nbody_2x	8.42 typical 8.33	8.19 typical 8.15	3% faster	3 of 3 vs 3 of 3	1067 vs 995 not comparable	no large difference 0.0 no small difference worst 0.0 64x36
hdr_2160p	7.80 typical 7.87	5.90 typical 5.72	32% faster	3 of 3 vs 3 of 3	1058 vs 1100	no large difference 0.0 no small difference worst 0.0 64x36
oit_linked_lists	17.31 typical 17.30	16.67 typical 16.60	4% faster	3 of 3 vs 3 of 3	1100 vs 1100	no large difference 0.0 no small difference worst 0.0 64x36
pipeline_barriers	1.53 typical 1.42	paced at 2.73	too light, see pipeline_barriers_2160p	3 of 3 vs 3 of 3	1100 vs 660 not comparable	no large difference 0.0 no small difference worst 0.0 64x36
pipeline_barriers_2160p	5.97 typical 5.57	5.02 typical 4.97	19% faster	3 of 3 vs 3 of 3	967 vs 1031 not comparable	no large difference 0.0 no small difference worst 0.0 64x36
subpasses	2.46 typical 2.29	paced at 2.74	not compared	3 of 3 vs 3 of 3	1100 vs 1058	not checked
subpasses_2160p	9.57	n/a	only one side	3 of 3 vs 3 of 3	1061 vs 1100	not checked

Time to draw one frame, per scene: the average (mean) frame time of the scene's best round, which is how many frames the driver delivered; lower is faster. "Typical" beside it is the median frame, which can sit well under the average when a driver's frames come in two speeds. A pair of reports made before per-round averages were recorded is compared on the median. A scene whose rounds disagree by more than 15% is left out. A scene marked "paced" was held at that mean frame time by the driver or the display queue, together with at least one other scene of the same run; its time says how the frames were spaced, not how fast the driver rendered, so it is not compared. A scene marked "too light" took under 8 ms at the standard size (1920 by 1080) on both drivers, where the screen's frame queue spaces the frames more than the chip does; its heavier run is the one compared: the row ending in _2160p is the scene drawn at 3840 by 2160, and compute_nbody_2x is that scene with twice the particles. "Faster" and "slower" describe turnip_faux123_v1.2.3 against turnip_faux123_v1.2.0. "Picture vs stock" compares one fixed frame per scene as a coarse colour grid (0 is identical, above 40 is a different picture); it finds a scene that was not drawn. Where both reports carry the finer 64x36 grid, the second figure is how many of its 2,304 cells differ by more than 4 of 255, and "worst" is the largest single cell; this is what finds a small rendering error in a picture that otherwise matches. A row marked 16x9 comes from a report made before the finer grid existed. "Starts" is how many of a scene's process starts reached a measured round. A GPU clock marked "not comparable" differed by more than 5% of the device's maximum between the two timings.

Technical differences (Vulkan)

Extensions only in turnip_faux123_v1.2.3 (0)

None

Extensions only in turnip_faux123_v1.2.0 (0)

None

Core features that differ (0)

None

Limits that differ (0)

Limit	turnip_faux123_v1.2.0	turnip_faux123_v1.2.3

Formats supported by only one driver (0 gained, 0 lost)

Only in turnip_faux123_v1.2.3:

None

Only in turnip_faux123_v1.2.0:

None