Enable half-precision complex Metal FFT kernels - #7

Closed

PhysicistJohn wants to merge 54 commits into

draft/reduced-complex-storage-bridgefrom

draft/raw-complex-half-fft

Commits on Aug 3, 2026

Commits on Aug 4, 2026

Fix Glorot/He uniform init docstrings to label the uniform bound, not sigma (#3831)
vineethsaivs
authored
ddb8162
Fix custom metal kernel cache collision for same name, different source (#3833)

katlun-lgtm
and
zcbenz
authored
697828f
Fix log_cosh_loss docstring to document the element-wise loss (#3846)
winklemad
authored
c0d916b
Fix step activation docstring to match >= threshold behavior (#3902)
ayaangazali
authored
a2fa537
docs: document the reduced-precision float32 default and MLX_ENABLE_TF32 (#3894)

stoyoda0012-cyber
and
claude
authored
0b5e91f
Fix InstanceNorm Shape docstring to require at least 3 dimensions (#3903)
ayaangazali
authored
f75ee50
Fix filter_and_map docstring argument order for filter_fn and is_leaf_fn (#3906)
ayaangazali
authored
cff8e0f
Export C++20 requirement to CMake consumers (#3971)
PhysicistJohn
authored
2448298
Fix implicit `thread` address space qualifier becoming explicit in metal 4.1 (#3963)
louen
authored
ef5fc0f
Fix Transformer ignoring a custom encoder or decoder with no parameters (#3962)
ayaangazali
authored
a136dc8
docs: remove references to removed --no-verify-script launch flag (#3959)
Recoordinate
authored
e50c0f2
Add eye(0) support (#3952)
aaishwarymishra
authored
255f953
Refactor the JACCL ring and add threads for the multiple rings (#3900)
angeloskath
authored
022ea8c
Fix shapeless matmul with dynamic batch dimensions (#3813)
varshneydevansh
authored
6dfd814
Fix JIT build with old macOS SDK (#3853)
metascroy
authored
b34e332
chore: Fix cholesky_inv arg name (#3950)
ayaangazali
authored
1c98450
docs: Add new_thread_unsafe_stream (#3968)
ayaangazali
authored
9a1e04d
Add gather_qqmm (#3757)
zcbenz
authored
a79332d

Commits on Aug 5, 2026

docs: Fix missing printoptions page (#3985)
ayaangazali
authored
bd813fb
docs: ThreadLocalStream and iinfo (#3986)
ayaangazali
authored
a681f8b
Make "stop" optional in arange (#3982)
aaishwarymishra
authored
0268c81
docs: Add softsign to the nn functions (#3989)
ayaangazali
authored
5391a8e
docs: Fix broken all_sum and Group references (#3996)
ayaangazali
authored
7c293d1
docs: Fix ast.metal_kernel typo (#3997)
ayaangazali
authored
f99e1dc
chore: Fix unresolved mx.array docstring (#3990)
ayaangazali
authored
391e140
python: fix bfloat16 buffer format itemsize mismatch (#3975)
reckylurker
authored
36e891e
Template Metal complex scalar lanes (#3970)
PhysicistJohn
authored
c4111f8
Pad 2D conv input channels to reach the specialized Metal kernel (#3904)
eyupcanakman
authored
8584610
Support head dimension 96 in Metal full attention (#3943)
dhiltgen
authored
ab946f9
Fix mx.remainder floored-mod for float16/bfloat16 on CPU (#3976)

sashko-zakharchuk
and
zcbenz
authored
2c46b95
Fix sorted gather_mm activation row stride (#3960)
metascroy
authored
268ecd6
Fix state corruption when a primitive throws during eval (#3675)

WindChimeRan
and
zcbenz
authored
fe92a05
python: added __complex__ support (#3984)

aaishwarymishra
and
zcbenz
authored
49ee726

Commits on Aug 6, 2026