-
Notifications
You must be signed in to change notification settings - Fork 612
Issues: iree-org/iree
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.
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
Author
Label
Projects
Milestones
Assignee
Sort
Issues list
Llama 3.1 8B prefill non-decomposed fp16 fails to compile with 0-D vectors Assertion failure
bug 🐞
Something isn't working
#19141
opened Nov 13, 2024 by
aviator19941
Crash in Something isn't working
codegen/rocm
ROCm code generation compiler backend (HIP/HSA)
codegen
Shared code generation infrastructure and dialects
PadDynamicAllocPass
bug 🐞
#19136
opened Nov 13, 2024 by
IanWood1
Request for being able to compile a efficient blockwise dequantization integer matmul
enhancement ➕
New feature or request
#19133
opened Nov 13, 2024 by
ziereis
--iree-flow-export-benchmark-funcs doesnt work when there are const expr in the IR
bug 🐞
Something isn't working
#19132
opened Nov 13, 2024 by
ziereis
compilation fails with --iree-dispatch-creation-enable-aggressive-fusion=true
bug 🐞
Something isn't working
#19131
opened Nov 13, 2024 by
ziereis
Pipeline convergence and default coverage improvement tasks for ROCM backend
#19121
opened Nov 12, 2024 by
nirvedhmeshram
linalg_ext.attention
should fuse the attention mask generation
#19117
opened Nov 12, 2024 by
rsuderman
Overall Model Testing/Benchmarking Plan
infrastructure/benchmark
Relating to benchmarking infrastructure
infrastructure
Relating to build systems, CI, or testing
#19115
opened Nov 12, 2024 by
saienduri
14 tasks
Compilation of GPT model fails due to assertion error
bug 🐞
Something isn't working
#19110
opened Nov 12, 2024 by
pxanthopoulos
tensor.cat
of dynamic length tensors always results in slow memcopys
#19092
opened Nov 9, 2024 by
rsuderman
iree-opt
does not properly include backend (amdgpu
) dependent dialect for standalone (iree-gpu-lower-ops
) pass
bug 🐞
#19081
opened Nov 8, 2024 by
jerryyin
Op uses negative memory exceeding 65536 byte limit
bug 🐞
Something isn't working
#19077
opened Nov 8, 2024 by
KyleHerndon
[RFC] Formally supporting some suite of "-O*" type flags
enhancement ➕
New feature or request
#19072
opened Nov 8, 2024 by
qedawkins
crash: mlir::linalg::computeAllSliceParameters
bug 🐞
Something isn't working
#19060
opened Nov 7, 2024 by
pdhirajkumarprasad
error: One or more operations with large vector sizes (16384 bytes) were found
bug 🐞
Something isn't working
#19058
opened Nov 7, 2024 by
pdhirajkumarprasad
RISCV cross compile completely infeasible
bug 🐞
Something isn't working
support
Request support or ask a question
#19057
opened Nov 7, 2024 by
soant
Refactor code around Python wrapping IREE's C API
infrastructure
Relating to build systems, CI, or testing
quality of life 😊
Nice things are nice; let's have some
compute_{common,local}_version.py
bindings/python
#19050
opened Nov 6, 2024 by
marbre
Llama 3.1 8b fp16 takes ~20 minutes to compile
bug 🐞
Something isn't working
#19049
opened Nov 6, 2024 by
aviator19941
[LLVMGPU] Add a verifier for tile sizes in ROCm code generation compiler backend (HIP/HSA)
enhancement ➕
New feature or request
onboarding/codegen
Tasks suitable for new team member onboarding
lowering_config
codegen/rocm
#19041
opened Nov 6, 2024 by
kuhar
Optimize Relating to the IREE compiler dialects (flow, hal, vm)
good first issue 🌱
Good for newcomers
performance ⚡
Performance/optimization related work across the compiler and runtime
scf.index_switch
/cf.switch
for known unreachable default cases.
compiler/dialects
#19036
opened Nov 5, 2024 by
benvanik
Previous Next
ProTip!
Add no:assignee to see everything that’s not assigned.