[XPU] XPU accelerator support for Intel GPU device #4547
mrwyattii
merged
36 commits into
deepspeedai:master
from
delock:gma/xpu_upstream
initial merge of Intel Extension for DeepSpeed
27cb0ab2
Remove intel_extension_for_deepspeed from xpu accelerator
a98bf469
remove sycl_kernel_path and sycl_kernel_include
28b4871c
better support for external XPU_Accelerator from IDEX
d736af73
Remove FlashAttention op
01b80495
workaround for ipex.has_xpu()
3df5fc86
op builder clean up
ae1364ac
fix format
f0f7a4c9
remove TransformerBuilder
0b25435d
fix source file path in transformer_inference
28f9feb3
remove inference builder
a7862864
Merge branch 'up-master' into gma/xpu_upstream
cde57d29
add new missing accelerator interfaces
5a542385
fix syntax error
27e07d65
add aio in xpu_upstream (#17)
32ecdaee
remove duplicate is_pinned
e3daa59e
deleting private path in aio builder (#20)
6024e059
aio op_builder delete unused method importing (#22)
beab27b1
add sycl KI for InferenceBuilder by using syclomatic (#21)
c9ef1ef0
remove white changes
674907c5
Merge branch 'master' into gma/xpu_upstream
289c88bb
add available_memory in XPU accelerator
f9da8023
Baodi/rebase ds (#29)
854a92ae
Merge branch 'master' into gma/xpu_upstream
f6ae4b53
Merge branch 'master' into gma/xpu_upstream
cf5e4bf8
Merge branch 'master' into gma/xpu_upstream
44bbe4a7
fix format
10a59a97
Cpuinfo fix (#30)
09fe24b1
Merge branch 'master' into gma/xpu_upstream
41a25398
upstream xpu support (#36)
6ab78497
remove unnecessary change in cuda code
6222291f
Merge branch 'master' into gma/xpu_upstream
f2c60a81
add export_env in accordance with PR#4830
5d8f0bd2
xpu_accelerator.py: add graph operations in accordance with PR#4318 (…
2665cfd1
Merge branch 'master' into gma/xpu_upstream
8b9e2c73
mrwyattii
approved these changes
on 2024-01-04
Merge branch 'master' into gma/xpu_upstream
790ee184
Login to write a write a comment.
Login via GitHub