-
Notifications
You must be signed in to change notification settings - Fork 344
Pull requests: google/XNNPACK
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
f32-raddstoreexpminusmax-avx-rr2 VEX128 version of SSE2 microkernel
#7125
opened Sep 19, 2024 by
copybara-service
bot
Loading…
Add
#line
statements to the generated microkernel amalgams pointing back to the original files to make tracking errors a bit easier.
#7123
opened Sep 18, 2024 by
copybara-service
bot
Loading…
Remove concatenation of microkernel srcs for microkernel libraries
#7122
opened Sep 18, 2024 by
copybara-service
bot
Loading…
Added quint8 and qint8 for mean_nd in xnnpack_delegate.
#7120
opened Sep 18, 2024 by
copybara-service
bot
Loading…
AVXVNNIINT8 remove 128 bias from input zero point
#7110
opened Sep 17, 2024 by
copybara-service
bot
Loading…
Remove output_min/max from binary elementwise ops API
#7106
opened Sep 17, 2024 by
copybara-service
bot
Loading…
Hook up the
FullyConnected
delegate to use qp8
inputs with both signed and unsigned weights, when appropriate.
#7068
opened Sep 10, 2024 by
copybara-service
bot
Loading…
Implement PRELU as a binary operator instead of a one-off operator type
#7034
opened Sep 5, 2024 by
copybara-service
bot
Loading…
Add
xnn_float16_t
for use instead of uint16_t
#7023
opened Sep 4, 2024 by
copybara-service
bot
Loading…
enable AVX_VNNI_INT8 instruction for qd8-f32-qc8w-gemm/igemm
#7019
opened Sep 4, 2024 by
xujuntwt95329
Loading…
Use the new
pthreadpool_parallelize_1d_dynamic
strategy in the unary and binary ops, where appropriate.
#7008
opened Sep 3, 2024 by
copybara-service
bot
Loading…
Previous Next
ProTip!
Type g p on any issue or pull request to go back to the pull request listing page.