diff --git a/.circleci/config.yml b/.circleci/config.yml
index 580f5fe2e..d04a33959 100644
--- a/.circleci/config.yml
+++ b/.circleci/config.yml
@@ -18,25 +18,34 @@ jobs:
             MATRIX_JSON=$(python .github/workflows/ci_set_matrix.py)
             echo "MATRIX_JSON=$MATRIX_JSON"
 
-            BUILDSYSTEM_TOOLCHAIN=(
-              "cmake aarch64-gcc"
-              "cmake arm-clang"
-              "cmake arm-gcc"
-              "cmake esp-idf"
-              "cmake msp430-gcc"
-              "cmake riscv-gcc"
+            BUILDSYSTEM_LIST=(
+              "cmake"
+              "make"
+            )
+
+            TOOLCHAIN_LIST=(
+              "aarch64-gcc"
+              "arm-clang"
+              "arm-gcc"
+              "esp-idf"
+              "msp430-gcc"
+              "riscv-gcc"
             )
 
             # only build IAR if not forked PR, since IAR token is not shared
             if [ -z $CIRCLE_PR_USERNAME ]; then
-              BUILDSYSTEM_TOOLCHAIN+=("cmake arm-iar")
+              TOOLCHAIN_LIST+=("arm-iar")
             fi
 
             gen_build_entry() {
               local build_system="$1"
               local toolchain="$2"
               local family="$3"
-              local resource_class="$4"
+              local build_args=""
+
+              if [[ "$toolchain" == "arm-iar" || "$build_system" == "make" ]]; then
+                build_args="--one-per-family"
+              fi
 
               if [[ "$toolchain" == "esp-idf" ]]; then
                 echo "      - build-vm:" >> .circleci/config2.yml
@@ -49,17 +58,21 @@ jobs:
               echo "              build-system: ['$build_system']" >> .circleci/config2.yml
               echo "              toolchain: ['$toolchain']" >> .circleci/config2.yml
               echo "              family: $family" >> .circleci/config2.yml
-              echo "              resource_class: ['$resource_class']" >> .circleci/config2.yml
+              echo "              resource_class: ['large']" >> .circleci/config2.yml
+              echo "              build-args: ['$build_args']" >> .circleci/config2.yml
             }
 
-            for e in "${BUILDSYSTEM_TOOLCHAIN[@]}"; do
-              e_arr=($e)
-              build_system="${e_arr[0]}"
-              toolchain="${e_arr[1]}"
-              FAMILY=$(echo $MATRIX_JSON | jq -r ".\"$toolchain\"")
-              echo "FAMILY_${toolchain}=$FAMILY"
+            for build_system in "${BUILDSYSTEM_LIST[@]}"; do
+              for toolchain in "${TOOLCHAIN_LIST[@]}"; do
+                # make does not support these toolchains
+                if [ "$build_system" == "make" ] && { [ "$toolchain" == "arm-clang" ] || [ "$toolchain" == "arm-iar" ] || [ "$toolchain" == "esp-idf" ]; }; then
+                  continue
+                fi
 
-              gen_build_entry "$build_system" "$toolchain" "$FAMILY" "large"
+                FAMILY=$(echo $MATRIX_JSON | jq -r ".\"$toolchain\"")
+                echo "FAMILY_${toolchain}=$FAMILY"
+                gen_build_entry "$build_system" "$toolchain" "$FAMILY"
+              done
             done
 
       - continuation/continue:
@@ -67,9 +80,5 @@ jobs:
 
 workflows:
   set-matrix:
-    # Only build PR here, Push will be built by github action.
-    when:
-      and:
-        - not: << pipeline.git.branch.is_default >>
     jobs:
       - set-matrix
diff --git a/.circleci/config2.yml b/.circleci/config2.yml
index 869597289..77bc4f790 100644
--- a/.circleci/config2.yml
+++ b/.circleci/config2.yml
@@ -66,6 +66,9 @@ commands:
         type: string
       family:
         type: string
+      build-args:
+        type: string
+        default: ""
 
     steps:
       - checkout
@@ -107,7 +110,7 @@ commands:
           no_output_timeout: 20m
           command: |
             if [ << parameters.toolchain >> == esp-idf ]; then
-              docker run --rm -v $PWD:/project -w /project espressif/idf:v5.3.2 python tools/build.py << parameters.family >>
+              docker run --rm -v $PWD:/project -w /project espressif/idf:v5.3.2 python tools/build.py << parameters.build-args >> << parameters.family >>
             else
               # Toolchain option default is gcc
               if [ << parameters.toolchain >> == arm-clang ]; then
@@ -121,7 +124,7 @@ commands:
 
               # circleci docker return $nproc as 36 core, limit parallel to 4 (resource-class = large)
               # Required for IAR, also prevent crashed/killed by docker
-              python tools/build.py -s << parameters.build-system >> $TOOLCHAIN_OPTION -j 4 << parameters.family >>
+              python tools/build.py -s << parameters.build-system >> $TOOLCHAIN_OPTION -j 4 << parameters.build-args >> << parameters.family >>
             fi
 
 jobs:
@@ -137,6 +140,9 @@ jobs:
         type: string
       family:
         type: string
+      build-args:
+        type: string
+        default: ""
 
     docker:
       - image: cimg/base:current
@@ -147,6 +153,7 @@ jobs:
           build-system: << parameters.build-system >>
           toolchain: << parameters.toolchain >>
           family: << parameters.family >>
+          build-args: << parameters.build-args >>
 
   # Build using VM
   build-vm:
@@ -160,6 +167,9 @@ jobs:
         type: string
       family:
         type: string
+      build-args:
+        type: string
+        default: ""
 
     machine:
       image: ubuntu-2404:current
@@ -170,6 +180,7 @@ jobs:
           build-system: << parameters.build-system >>
           toolchain: << parameters.toolchain >>
           family: << parameters.family >>
+          build-args: << parameters.build-args >>
 
 workflows:
   build:
diff --git a/.github/workflows/build.yml b/.github/workflows/build.yml
index 77f2d573f..a1bacbc27 100644
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -56,10 +56,11 @@ jobs:
           echo "hil_matrix=$HIL_MATRIX_JSON"
           echo "hil_matrix=$HIL_MATRIX_JSON" >> $GITHUB_OUTPUT
 
-  # ---------------------------------------
-  # Build CMake: only one-per-family.
-  # Full built is done by CircleCI in PR
-  # ---------------------------------------
+  # ------------------------------------------------------------------------------
+  # CMake build: only one-per-family. Full built is done by CircleCI in PR
+  # Note:
+  # For Make and IAR build: will be done on CircleCI only (one-per-family too)
+  # ------------------------------------------------------------------------------
   cmake:
     needs: set-matrix
     uses: ./.github/workflows/build_util.yml
@@ -70,7 +71,7 @@ jobs:
           - 'aarch64-gcc'
           #- 'arm-clang'
           - 'arm-gcc'
-          - 'esp-idf'
+          #- 'esp-idf'
           - 'msp430-gcc'
           - 'riscv-gcc'
     with:
@@ -137,52 +138,6 @@ jobs:
           header: code-metrics
           path: metrics_compare.md
 
-
-  # ---------------------------------------
-  # Build Make: only build on push with one-per-family
-  # ---------------------------------------
-  make:
-    if: github.event_name == 'push'
-    needs: set-matrix
-    uses: ./.github/workflows/build_util.yml
-    strategy:
-      fail-fast: false
-      matrix:
-        toolchain:
-          - 'aarch64-gcc'
-          #- 'arm-clang'
-          - 'arm-gcc'
-          - 'msp430-gcc'
-          - 'riscv-gcc'
-          - 'rx-gcc'
-    with:
-      build-system: 'make'
-      toolchain: ${{ matrix.toolchain }}
-      build-args: ${{ toJSON(fromJSON(needs.set-matrix.outputs.json)[matrix.toolchain]) }}
-      one-per-family: true
-
-  # ---------------------------------------
-  # Build IAR
-  # Since IAR Token secret is not passed to forked PR, only build non-forked PR with make.
-  # cmake is built by circle-ci. Due to IAR limit capacity, only build oe per family
-  # ---------------------------------------
-  arm-iar:
-    if: false # disable for now since we got reach capacity limit too often
-    #if: github.event_name == 'push' && github.repository_owner == 'hathach'
-    needs: set-matrix
-    uses: ./.github/workflows/build_util.yml
-    secrets: inherit
-    strategy:
-      fail-fast: false
-      matrix:
-        build-system:
-          - 'make'
-    with:
-      build-system: ${{ matrix.build-system }}
-      toolchain: 'arm-iar'
-      build-args: ${{ toJSON(fromJSON(needs.set-matrix.outputs.json)['arm-iar']) }}
-      one-per-family: true
-
   # ---------------------------------------
   # Build Make/CMake on Windows/MacOS
   # ---------------------------------------
@@ -193,10 +148,9 @@ jobs:
       fail-fast: false
       matrix:
         os: [ windows-latest, macos-latest ]
-        build-system: [ 'make', 'cmake' ]
     with:
       os: ${{ matrix.os }}
-      build-system: ${{ matrix.build-system }}
+      build-system: 'cmake-make'
       toolchain: 'arm-gcc-${{ matrix.os }}'
       build-args: '["stm32h7"]'
       one-per-family: true
diff --git a/.github/workflows/build_util.yml b/.github/workflows/build_util.yml
index 36043a1d5..2fc0eead0 100644
--- a/.github/workflows/build_util.yml
+++ b/.github/workflows/build_util.yml
@@ -68,6 +68,9 @@ jobs:
         run: |
           if [ "$TOOLCHAIN" == "esp-idf" ]; then
             docker run --rm -v $PWD:/project -w /project espressif/idf:tinyusb python tools/build.py ${{ matrix.arg }}
+          elif [ "${{ inputs.build-system }}" == "cmake-make" ] || [ "${{ inputs.build-system }}" == "make-cmake" ]; then
+            python tools/build.py -s make ${{ steps.setup-toolchain.outputs.build_option }} ${{ steps.set-one-per-family.outputs.build_option }} ${{ matrix.arg }}
+            python tools/build.py -s cmake ${{ steps.setup-toolchain.outputs.build_option }} ${{ steps.set-one-per-family.outputs.build_option }} ${{ matrix.arg }}
           else
             python tools/build.py -s ${{ inputs.build-system }} ${{ steps.setup-toolchain.outputs.build_option }} ${{ steps.set-one-per-family.outputs.build_option }} ${{ matrix.arg }}
           fi
diff --git a/examples/build_system/make/toolchain/gcc_common.mk b/examples/build_system/make/toolchain/gcc_common.mk
index 0cbb6774d..42fd01183 100644
--- a/examples/build_system/make/toolchain/gcc_common.mk
+++ b/examples/build_system/make/toolchain/gcc_common.mk
@@ -31,6 +31,9 @@ CFLAGS += \
   -Wreturn-type \
   -Wredundant-decls \
 
+CFLAGS_CLANG += \
+  -Wno-error=unknown-warning-option
+
 #  -Wmissing-prototypes \
 # conversion is too strict for most mcu driver, may be disable sign/int/arith-conversion
 #  -Wconversion
diff --git a/hw/bsp/kinetis_k/family.mk b/hw/bsp/kinetis_k/family.mk
index e95cdb717..7a51a77d8 100644
--- a/hw/bsp/kinetis_k/family.mk
+++ b/hw/bsp/kinetis_k/family.mk
@@ -9,11 +9,13 @@ CFLAGS += \
   -DCFG_TUSB_MCU=OPT_MCU_KINETIS_K \
 
 LDFLAGS += \
-  -nostartfiles \
-  --specs=nosys.specs --specs=nano.specs \
   -Wl,--defsym,__stack_size__=0x400 \
   -Wl,--defsym,__heap_size__=0
 
+LDFLAGS_GCC += \
+  -nostartfiles \
+  --specs=nosys.specs --specs=nano.specs \
+
 SRC_C += \
 	src/portable/nxp/khci/dcd_khci.c \
 	src/portable/nxp/khci/hcd_khci.c \
diff --git a/hw/bsp/kinetis_kl/family.mk b/hw/bsp/kinetis_kl/family.mk
index 8d113aecf..aec53d486 100644
--- a/hw/bsp/kinetis_kl/family.mk
+++ b/hw/bsp/kinetis_kl/family.mk
@@ -9,11 +9,13 @@ CFLAGS += \
   -DCFG_TUSB_MCU=OPT_MCU_KINETIS_KL \
 
 LDFLAGS += \
-  -nostartfiles \
-  -specs=nosys.specs -specs=nano.specs \
   -Wl,--defsym,__stack_size__=0x400 \
   -Wl,--defsym,__heap_size__=0
 
+LDFLAGS_GCC += \
+  -nostartfiles \
+  -specs=nosys.specs -specs=nano.specs \
+
 SRC_C += \
 	src/portable/nxp/khci/dcd_khci.c \
 	src/portable/nxp/khci/hcd_khci.c \
diff --git a/tools/build.py b/tools/build.py
index e4909f45f..c4f1558c0 100755
--- a/tools/build.py
+++ b/tools/build.py
@@ -142,16 +142,12 @@ def make_one_example(example, board, make_option):
         r = 2
     else:
         start_time = time.monotonic()
-        # skip -j for circleci
-        if not os.getenv('CIRCLECI'):
-            make_option += ' -j'
-        make_args = ["make", "-C", f"examples/{example}", f"BOARD={board}"]
+        make_args = ["make", "-C", f"examples/{example}", f"BOARD={board}", '-j', str(parallel_jobs)]
         if make_option:
             make_args += shlex.split(make_option)
-        make_args.append("all")
         if clean_build:
             run_cmd(make_args + ["clean"])
-        build_result = run_cmd(make_args)
+        build_result = run_cmd(make_args + ['all'])
         r = 0 if build_result.returncode == 0 else 1
         print_build_result(board, example, r, time.monotonic() - start_time)
 
diff --git a/tools/metrics.py b/tools/metrics.py
index bb84f803e..c3b366e42 100644
--- a/tools/metrics.py
+++ b/tools/metrics.py
@@ -195,17 +195,13 @@ def compare_maps(base_file, new_file, filters=None):
 
 def format_diff(base, new, diff):
     """Format a diff value with percentage."""
-    if base == 0 and new == 0:
-        return "0"
-    if base == 0:
-        return f"{new} (new)"
-    if new == 0:
-        return f"{base} ➡ 0"
     if diff == 0:
-        return f"{base} ➡ {new}"
+        return f"{new}"
+    if base == 0 or new == 0:
+        return f"{base} ➙ {new}"
     pct = (diff / base) * 100
     sign = "+" if diff > 0 else ""
-    return f"{base} ➡ {new} ({sign}{diff}, {sign}{pct:.1f}%)"
+    return f"{base} ➙ {new} ({sign}{diff}, {sign}{pct:.1f}%)"
 
 
 def get_sort_key(sort_order):
@@ -232,10 +228,11 @@ def write_compare_markdown(comparison, path, sort_order='size'):
     sections = comparison["sections"]
 
     md_lines = [
-        "# TinyUSB Code Size Different Report",
+        "# Size Difference Report",
         "",
-        f"**Base:** `{comparison['base_file']}`",
-        f"**New:** `{comparison['new_file']}`",
+        "Because TinyUSB code size varies by port and configuration, the metrics below represent the averaged totals across all example builds."
+        "",
+        "Note: If there is no change, only one value is shown.",
         "",
     ]