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

Deprecate linalg tile to forall #17435

Closed
wants to merge 2 commits into from
Closed

Deprecate linalg tile to forall #17435

wants to merge 2 commits into from

Conversation

MaheshRavishankar
Copy link
Contributor

@MaheshRavishankar MaheshRavishankar commented May 17, 2024 •
edited
Loading

Signed-off-by: MaheshRavishankar [email protected]

@MaheshRavishankar MaheshRavishankar force-pushed the deprecate_linalg_tile_to_forall branch 4 times, most recently from 5c1a646 to 7dd3660 Compare May 21, 2024 18:28
@MaheshRavishankar MaheshRavishankar force-pushed the deprecate_linalg_tile_to_forall branch from 7dd3660 to 51375cc Compare May 31, 2024 01:41
@MaheshRavishankar MaheshRavishankar added benchmarks:cuda Run default CUDA benchmarks benchmarks:x86_64 Run default x86_64 benchmarks benchmarks:comp-stats Run default compilation statistics benchmarks benchmarks:android-cpu Run default Android CPU benchmarks benchmarks:android-gpu Run default Android GPU benchmarks benchmarks:vulkan-nvidia Run default Vulkan benchmarks on NVIDIA GPU labels May 31, 2024
@MaheshRavishankar MaheshRavishankar force-pushed the deprecate_linalg_tile_to_forall branch 3 times, most recently from e42597e to feb28ae Compare June 2, 2024 04:27
@MaheshRavishankar MaheshRavishankar mentioned this pull request Jun 2, 2024
Merged
@github-actions GitHub Actions
Copy link

github-actions bot commented Jun 2, 2024

Abbreviated Benchmark Summary

@ commit 61547707e14375ac1c9f384e19680670d9239172 (no previous benchmark results to compare)

Data-Tiling Comparison Table

Click to show
Name No-DT (baseline) DT-Only DT-UK
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] 214.136 (1.0X) N/A 107.601 (2.0X)
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] 775.312 (1.0X) N/A 220.774 (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] 7.005 (1.0X) N/A 8.451 (0.8X)
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] 31.986 (1.0X) N/A 29.753 (1.1X)
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] 36.053 (1.0X) N/A 34.405 (1.0X)
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] 277.057 (1.0X) N/A 228.738 (1.2X)
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] 5.829 (1.0X) N/A 5.001 (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] 26.832 (1.0X) N/A 13.042 (2.1X)
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] 9.326 (1.0X) N/A 8.797 (1.1X)
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] 69.775 (1.0X) N/A 40.749 (1.7X)
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] 11.297 (1.0X) N/A 8.586 (1.3X)
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] 88.902 (1.0X) N/A 41.579 (2.1X)
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] 12.369 (1.0X) N/A 12.984 (1.0X)
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] 80.300 (1.0X) N/A 56.571 (1.4X)
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] 34.370 (1.0X) N/A 62.093 (0.6X)
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] 179.602 (1.0X) N/A 185.698 (1.0X)
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] 34.236 (1.0X) N/A 62.405 (0.5X)
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] 180.288 (1.0X) N/A 192.136 (0.9X)
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] 66.684 (1.0X) N/A 62.529 (1.1X)
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] 490.447 (1.0X) N/A 214.178 (2.3X)
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] 4.883 (1.0X) N/A 4.509 (1.1X)
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] 24.654 (1.0X) N/A 17.915 (1.4X)
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] 3.785 (1.0X) N/A 4.897 (0.8X)
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] 11.713 (1.0X) N/A 11.517 (1.0X)
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] 5.899 (1.0X) N/A 5.401 (1.1X)
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] 21.514 (1.0X) N/A 11.930 (1.8X)
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] 2.895 (1.0X) N/A 2.807 (1.0X)
MobileNetV3Small_fp32(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ c2-standard-60[cpu] 2.752 (1.0X) N/A 2.664 (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] 8.531 (1.0X) N/A 9.801 (0.9X)
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] 34.219 (1.0X) N/A 31.280 (1.1X)
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] 0.817 (1.0X) N/A 0.604 (1.4X)
PersonDetect_int8(tflite) [x86_64-cascadelake-linux_gnu-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ c2-standard-60[cpu] 0.750 (1.0X) N/A 0.540 (1.4X)
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] 4.127 (1.0X) N/A 5.134 (0.8X)
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] 17.840 (1.0X) N/A 18.897 (0.9X)
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] 7.565 (1.0X) N/A 7.590 (1.0X)
DeepLabV3_fp32(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ pixel-6-pro[big-cores] 48.384 (1.0X) N/A 43.901 (1.1X)
DeepLabV3_fp32(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 50.206 (1.0X) N/A 44.557 (1.1X)
DeepLabV3_fp32(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[2-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 30.203 (1.0X) N/A 27.938 (1.1X)
GPT2_117M_TF_1X1XI32(stablehlo) [armv8.2-a-generic-linux_android29-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ pixel-6-pro[big-cores] 92.611 (1.0X) N/A 21.451 (4.3X)
GPT2_117M_TF_1X1XI32(stablehlo) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 93.323 (1.0X) N/A 22.311 (4.2X)
GPT2_117M_TF_1X1XI32(stablehlo) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[2-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 51.808 (1.0X) N/A 22.163 (2.3X)
GPT2_117M_TF_1X4XI32(stablehlo) [armv8.2-a-generic-linux_android29-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ pixel-6-pro[big-cores] 140.031 (1.0X) N/A 27.277 (5.1X)
GPT2_117M_TF_1X4XI32(stablehlo) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 123.979 (1.0X) N/A 28.860 (4.3X)
GPT2_117M_TF_1X4XI32(stablehlo) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[2-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 76.312 (1.0X) N/A 26.419 (2.9X)
MobileBertSquad_fp32(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ pixel-6-pro[big-cores] 707.290 (1.0X) N/A 359.475 (2.0X)
MobileBertSquad_fp32(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 703.844 (1.0X) N/A 357.872 (2.0X)
MobileBertSquad_fp32(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[2-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 395.829 (1.0X) N/A 218.675 (1.8X)
MobileBertSquad_int8(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ pixel-6-pro[big-cores] 1118.220 (1.0X) N/A 304.011 (3.7X)
MobileBertSquad_int8(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 1117.021 (1.0X) N/A 304.188 (3.7X)
MobileBertSquad_int8(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[2-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 574.418 (1.0X) N/A 179.684 (3.2X)
Vit_int8(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ pixel-6-pro[big-cores] 2102.807 (1.0X) N/A 306.605 (6.9X)
Vit_int8(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[1-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 2103.805 (1.0X) N/A 305.405 (6.9X)
Vit_int8(tflite) [armv8.2-a-generic-linux_android29-llvm_cpu] local_task(embedded_elf)[2-thread,full-inference,system-scheduling] with default @ pixel-6-pro[big-cores] 1124.972 (1.0X) N/A 183.012 (6.1X)
matmul_256x256x2048_i8_i4_i32_tile_config_default(linalg) [armv8.2-a-generic-linux_android29-llvm_cpu] local_sync(embedded_elf)[full-inference,default-flags] with default @ pixel-6-pro[big-cores] 11.978 (1.0X) N/A 1.307 (9.2X)

Raw Latencies

Benchmark Name Average Latency (ms) Median Latency (ms) Latency Standard Deviation (ms)
BertForMaskedLMTF(stablehlo) [x86\_64-cascadelake-linux\_gnu-llvm\_cpu][experimental-flags,no-dt] local\_task(embedded\_elf)[30-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 214.136 206.618 15.585
BertLargeTF(stablehlo) [x86\_64-cascadelake-linux\_gnu-llvm\_cpu][default-flags,dt-uk] local\_task(embedded\_elf)[30-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 220.774 220.005 4.560
DeepLabV3\_fp32(tflite) [x86\_64-cascadelake-linux\_gnu-llvm\_cpu][default-flags,dt-uk] local\_task(embedded\_elf)[8-thread,full-inference,default-flags] with default @ c2-standard-60[cpu] 8.451 8.447 0.027

[Top 3 out of 135 results showed]

No improved or regressed compilation metrics 🏖️

For more information:

Source Workflow Run

@MaheshRavishankar MaheshRavishankar force-pushed the deprecate_linalg_tile_to_forall branch from feb28ae to 13e4b19 Compare July 30, 2024 23:11
@MaheshRavishankar
Pick up llvm/llvm-project#91878
5a7b3dc 
Signed-off-by: MaheshRavishankar <[email protected]>
@MaheshRavishankar
Fixes for llvm/llvm-project#91878
b40f498 
Signed-off-by: MaheshRavishankar <[email protected]>
@MaheshRavishankar MaheshRavishankar force-pushed the deprecate_linalg_tile_to_forall branch from 13e4b19 to b40f498 Compare July 31, 2024 17:27
@bjacob bjacob mentioned this pull request Aug 1, 2024
Merged
@bjacob
Copy link
Contributor

bjacob commented Aug 1, 2024

Folded into #18078.

@bjacob bjacob closed this Aug 1, 2024
bjacob added a commit that referenced this pull request Aug 1, 2024
@bjacob @MaheshRavishankar
Integrate LLVM at e7630a0d (#18078)
dcae2e7 
Advance to llvm/llvm-project@e7630a0d

No local patches.

Folds @MaheshRavishankar's #17435.

---------

Signed-off-by: Benoit Jacob <[email protected]>
Signed-off-by: Mahesh Ravishankar <[email protected]>
Co-authored-by: Mahesh Ravishankar <[email protected]>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@hanhanW hanhanW Awaiting requested review from hanhanW hanhanW will be requested when the pull request is marked ready for review hanhanW is a code owner

@antiagainst antiagainst Awaiting requested review from antiagainst antiagainst will be requested when the pull request is marked ready for review antiagainst is a code owner

@qedawkins qedawkins Awaiting requested review from qedawkins qedawkins will be requested when the pull request is marked ready for review qedawkins is a code owner

Assignees
No one assigned
Labels
benchmarks:android-cpu Run default Android CPU benchmarks benchmarks:android-gpu Run default Android GPU benchmarks benchmarks:comp-stats Run default compilation statistics benchmarks benchmarks:cuda Run default CUDA benchmarks benchmarks:vulkan-nvidia Run default Vulkan benchmarks on NVIDIA GPU benchmarks:x86_64 Run default x86_64 benchmarks
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@MaheshRavishankar @bjacob