Home / Adreno (TM) 840 / Compare

turnip_faux123_v1.2.5 vs mainline-turnip-V32

On Adreno (TM) 840 (HONOR Magic8 pro, iQOO 15, LENOVO TB324ZC, nubia NX809J, OnePlus 15, POCO F8 Ultra, POCO F9 Ultra, realme RMX5210, Samsung Galaxy S26, Samsung Galaxy S26, Samsung Galaxy S26 Ultra, Samsung Galaxy S26 Ultra, Samsung Galaxy S26 Ultra, Samsung Galaxy Z Fold8, Samsung Galaxy Z Fold8 Ultra, Xiaomi 17, Xiaomi 2512BPNDAG). The first driver named is compared against the second, which is the baseline.

Stockmainline-turnip-V32Mesa 26.3.0-devel (git-9c7e022677)

Customturnip_faux123_v1.2.5Mesa 26.3.0-devel (git-dd45fd0baa)

Graphics system version (Vulkan)1.4.363 to 1.4.359

Optional features (extensions)184 to 179+1 -6

Features that pass their check (validated)66 to 693 pass only on turnip_faux123_v1.2.5 0 pass only on mainline-turnip-V32

Speed test scenes (benchmark)3 faster 1 slower3% 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

hdr_2160p40% faster
oit_linked_lists9% faster
pipeline_barriers44% slower
subpasses_2160p29% faster

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

Small areas of the picture differ from the stock driver in 1 scene (pipeline_barriers: 1486 of 2304 cells). The picture matches overall, so this is a local rendering fault, not a different scene.

Did not start every time: compute_nbody (stock 0 of 3, custom 3 of 3).

mainline-turnip-V32 details turnip_faux123_v1.2.5 details

Feature checks that differ (3)

Feature	mainline-turnip-V32	turnip_faux123_v1.2.5	Change
compat.nested_map core	crash	pass	gained
core.compute_udiv core	fail	pass	gained
core.copy_wide_row_length core	fail	pass	gained

Speed test scenes (benchmark)

Scene	mainline-turnip-V32 ms	turnip_faux123_v1.2.5 ms	Difference	Starts	GPU clock, MHz	Picture vs stock
compute_nbody	n/a	7.77 typical 7.47	only one side	0 of 3 vs 3 of 3	? vs ? clock unknown	not checked
compute_nbody_2x	n/a	8.92 typical 8.97	only one side	? vs 3 of 3	? vs ? clock unknown	not checked
hdr_2160p	8.49 typical 8.50	6.06 typical 5.78	40% faster	3 of 3 vs 3 of 3	? vs ? clock unknown	no large difference 0.0 no small difference worst 0.0 64x36
oit_linked_lists	12.84 typical 12.78	11.76 typical 11.74	9% faster	3 of 3 vs 3 of 3	? vs ? clock unknown	no large difference 0.0 no small difference worst 0.0 64x36
pipeline_barriers	1.55 typical 1.44	2.76 typical 2.35	44% slower	3 of 3 vs 3 of 3	? vs ? clock unknown	no large difference 10.6 differs in 1486 small areas worst 62.7 64x36
pipeline_barriers_2160p	5.46	n/a	only one side	3 of 3 vs 3 of 3	? vs ? clock unknown	no large difference 0.0 no small difference worst 0.0 64x36
subpasses	n/a	n/a	too light, see subpasses_2160p	3 of 3 vs 3 of 3	? vs ? clock unknown	not checked
subpasses_2160p	9.60 typical 9.76	7.43 typical 7.41	29% faster	3 of 3 vs 3 of 3	? vs ? clock unknown	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_v1.2.5 against mainline-turnip-V32. "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_v1.2.5 (1)

VK_EXT_shader_subgroup_ballot

Extensions only in mainline-turnip-V32 (6)

VK_EXT_fragment_density_map
VK_EXT_fragment_density_map_offset
VK_KHR_shader_quad_control
VK_QCOM_fragment_density_map_offset
VK_VALVE_buffer_device_address_allocation_alignment
VK_VALVE_fragment_density_map_layered

Core features that differ (0)

None

Limits that differ (2)

Limit	mainline-turnip-V32	turnip_faux123_v1.2.5
maxComputeSharedMemorySize	65536	32768
maxComputeWorkGroupInvocations	2048	1024

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

Only in turnip_faux123_v1.2.5:

None

Only in mainline-turnip-V32:

None