-
It sounds like using a particular driver on Windows allows RDNA3 to work. However, on Linux, none of the 3 Vulkan drivers are currently capable of generating images.
Here are the drivers tested and…
-
No messages in these:
https://github.com/google/iree/blob/fd32e47b1695f105d20c06b8b20a29ef65c5e54c/iree/testing/status_matchers.h#L329-L332
Messages _do_ appear in these:
https://github.com/goo…
-
Following discussion [on Discord here](https://discord.com/channels/689900678990135345/689906000043573354/1163620012033060884).
I see this code dominating performance profiles: https://github.com/o…
-
### What happened?
Hi, I'm getting an all-NAN output for the falcon-180B-gptq model for the CPU backend which gives the correct result for the PyTorch. I have generated a smaller repro from the origi…
-
### What happened?
So, for CPU backend, for [llama IR](https://storage.googleapis.com/shark_tank/MiniGPT4/first_llama_int4_1700.mlir) following scenario takes place :-
1. It either doesn't get thr…
-
The following tests fail after integration #9790:
```
Failed Tests (1):
TENSORFLOW_TESTS :: iree_tfl_tests/llvmaot_mobilebert_tf2_quant.run
```
More relavent [logs](https://source.cloud.goo…
-
https://github.com/nod-ai/iree-amd-aie/pull/845 is blocked because at that commit of IREE, `iree-codegen-iree-comprehensive-bufferize` generates `memref`s with dynamic offsets and we get an error [her…
-
The work laid out in https://github.com/openxla/iree/issues/15158 has been completed, and now we are moving forward with e2e testing of the llama2 7B model with the new changes. This issue will be for…
-
### Request description
Currently `iree_vm_wait_invoke` API is a blocking one (caller thread will be blocked until all wait sources become ready). Can we add a non blocking wait accepting a callbac…
-
This issue is meant to track the IREE-side performance burndown of the prefill step of quantized (int4) LLaMA2 inference on CPU.
[download llama2_7b_int4_cpu.vmfb](https://storage.googleapis.com/sh…