Skip to content
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.

Sign up for GitHub

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

Jump to bottom

Integrate llvm-project @31015240d366e4bf6f114856caa6e9ce90742b7f #17799

Merged
merged 3 commits into from
Jul 3, 2024
Merged

Integrate llvm-project @31015240d366e4bf6f114856caa6e9ce90742b7f #17799

merged 3 commits into from
Jul 3, 2024

Conversation

qedawkins
Copy link
Contributor

@qedawkins qedawkins commented Jul 2, 2024 •
edited
Loading

Carries the same reverts as the previous integrate

@qedawkins qedawkins force-pushed the integrates/llvm-20240702 branch from 6cca61d to de535ff Compare July 2, 2024 21:46
@github-actions GitHub Actions
Copy link

github-actions bot commented Jul 2, 2024 •
edited
Loading

Abbreviated Benchmark Summary

@ commit 6c0dd89ecaf682ae27e71de3983b98e94ecc95fa (vs. base 5f2ebe99c62e1259ec9e50d1aaed520c3f66b5f6)

Data-Tiling Comparison Table

Click to show
Name No-DT (baseline) DT-Only DT-UK
BertLargeTF(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[30-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 783.192 (1.0X) 273.822 (2.9X) 224.775 (3.5X)
DeepLabV3_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[8-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 6.895 (1.0X) 9.200 (0.7X) 8.527 (0.8X)
EfficientNetV2STF(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[15-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 35.617 (1.0X) 36.507 (1.0X) 34.774 (1.0X)
EfficientNet_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[8-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 5.793 (1.0X) 10.988 (0.5X) 5.010 (1.2X)
GPT2_117M_TF_1X1XI32(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[15-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 9.119 (1.0X) 8.524 (1.1X) 8.513 (1.1X)
GPT2_117M_TF_1X4XI32(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[15-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 11.771 (1.0X) 8.993 (1.3X) 8.983 (1.3X)
MiniLML12H384Uncased(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[15-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 11.876 (1.0X) 15.441 (0.8X) 13.701 (0.9X)
MobileBertSquad_fp16(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[15-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 33.615 (1.0X) 65.431 (0.5X) 61.715 (0.5X)
MobileBertSquad_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[15-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 34.505 (1.0X) 65.813 (0.5X) 62.134 (0.6X)
MobileBertSquad_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[15-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 68.448 (1.0X) 134.019 (0.5X) 64.471 (1.1X)
MobileNetV1_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[8-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 4.978 (1.0X) 5.305 (0.9X) 4.570 (1.1X)
MobileNetV2_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[8-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 3.711 (1.0X) 5.299 (0.7X) 4.893 (0.8X)
MobileNetV2_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[8-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 5.839 (1.0X) 9.524 (0.6X) 5.389 (1.1X)
MobileNetV3Small_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 2.871 (1.0X) 3.378 (0.8X) 2.812 (1.0X)
MobileSSD_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[8-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 8.471 (1.0X) 10.972 (0.8X) 9.931 (0.9X)
PersonDetect_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 0.773 (1.0X) 1.380 (0.6X) 0.586 (1.3X)
PoseNet_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[8-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 4.126 (1.0X) 5.858 (0.7X) 5.249 (0.8X)
matmul_256x256x2048_i8_i4_i32_tile_config_default(linalg) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ c2-standard-60[cpu] 7.587 (1.0X) 7.533 (1.0X) 7.583 (1.0X)
matmul_256x256x2048_i8_i8_i32_tile_config_default(linalg) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ c2-standard-60[cpu] 6.624 (1.0X) 13.269 (0.5X) 1.811 (3.7X)
BertForMaskedLMTF(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[30-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 217.905 (1.0X) 135.385 (1.6X) 106.741 (2.0X)
DeepLabV3_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 32.026 (1.0X) 36.126 (0.9X) 29.809 (1.1X)
EfficientNetV2STF(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 275.390 (1.0X) 256.581 (1.1X) 228.591 (1.2X)
EfficientNet_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 26.767 (1.0X) 51.417 (0.5X) 13.058 (2.0X)
GPT2_117M_TF_1X1XI32(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 69.796 (1.0X) 37.261 (1.9X) 37.673 (1.9X)
GPT2_117M_TF_1X4XI32(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 88.116 (1.0X) 40.018 (2.2X) 39.000 (2.3X)
MiniLML12H384Uncased(stablehlo) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 79.497 (1.0X) 78.258 (1.0X) 55.833 (1.4X)
MobileBertSquad_fp16(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 179.479 (1.0X) 250.450 (0.7X) 185.104 (1.0X)
MobileBertSquad_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 181.782 (1.0X) 252.313 (0.7X) 189.537 (1.0X)
MobileBertSquad_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 516.084 (1.0X) 1084.852 (0.5X) 240.789 (2.1X)
MobileNetV1_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 23.509 (1.0X) 22.510 (1.0X) 17.793 (1.3X)
MobileNetV2_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 11.747 (1.0X) 14.797 (0.8X) 11.209 (1.0X)
MobileNetV2_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 21.619 (1.0X) 42.278 (0.5X) 11.863 (1.8X)
MobileNetV3Small_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ c2-standard-60[cpu] 2.770 (1.0X) 3.227 (0.9X) 2.681 (1.0X)
MobileSSD_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 34.705 (1.0X) 39.023 (0.9X) 31.626 (1.1X)
PersonDetect_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ c2-standard-60[cpu] 0.699 (1.0X) 1.293 (0.5X) 0.524 (1.3X)
PoseNet_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 17.755 (1.0X) 23.726 (0.7X) 19.522 (0.9X)
matmul_1x256x2048_i8_i4_i32_tile_config_default(linalg) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ c2-standard-60[cpu] 0.054 (1.0X) 0.054 (1.0X) 0.054 (1.0X)
matmul_1x256x2048_i8_i8_i32_tile_config_default(linalg) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ c2-standard-60[cpu] 0.042 (1.0X) 0.225 (0.2X) 0.021 (1.9X)

Improved Latencies 🎉

Benchmark Name Average Latency (ms) Median Latency (ms) Latency Standard Deviation (ms)
MobileNetV1\_fp32(tflite) [x86\_64-cascadelake-linux\_gnu-llvm\_cpu][experimental-flags,no-dt] local\_task(embedded\_elf)[1-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 23.509 (vs. 24.659, 4.66%↓) 23.998 0.992

No improved or regressed compilation metrics 🏖️

For more information:

Source Workflow Run

@qedawkins qedawkins mentioned this pull request Jul 3, 2024
Merged
@qedawkins qedawkins force-pushed the integrates/llvm-20240702 branch from 4713dbd to a6c3de9 Compare July 3, 2024 14:40
@qedawkins qedawkins changed the title Integrate llvm-project @59f4267c8e0625c6583327be2db1608930f2d796 Integrate llvm-project @31015240d366e4bf6f114856caa6e9ce90742b7f Jul 3, 2024
@qedawkins qedawkins marked this pull request as ready for review July 3, 2024 15:46
@hanhanW
Copy link
Contributor

hanhanW commented Jul 3, 2024

Can you put the local revert commits to the PR description, like d2bca4e? I think it is very useful for tracking our revert commits.

@qedawkins qedawkins enabled auto-merge (squash) July 3, 2024 17:31
@qedawkins qedawkins merged commit dd7d4a1 into main Jul 3, 2024
60 checks passed
@qedawkins qedawkins deleted the integrates/llvm-20240702 branch July 3, 2024 18:44
LLITCHEV pushed a commit to LLITCHEV/iree that referenced this pull request Jul 30, 2024
@qedawkins @LLITCHEV
Integrate llvm-project @31015240d366e4bf6f114856caa6e9ce90742b7f (ire…
9040fe6 
…e-org#17799)

Carries the same reverts as the previous integrate
- revert

llvm/llvm-project@2c06fb8
- revert

llvm/llvm-project@fa06668
- revert

llvm/llvm-project@137a745
Signed-off-by: Lubo Litchev <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@hanhanW hanhanW hanhanW approved these changes

@ScottTodd ScottTodd ScottTodd approved these changes

@benvanik benvanik Awaiting requested review from benvanik benvanik is a code owner

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@qedawkins @hanhanW @ScottTodd