diff --git a/.github/workflows/build-run-kernel-snake.yml b/.github/workflows/build-run-kernel-snake.yml
index d3af9d1f..7b509cc6 100644
--- a/.github/workflows/build-run-kernel-snake.yml
+++ b/.github/workflows/build-run-kernel-snake.yml
@@ -20,4 +20,4 @@ jobs:
         working-directory: kernels/${{ matrix.kernel }}
     strategy:
       matrix:
-        kernel: [alloc, simple_copy, transform_copy, gemm, rescale, gemmini, streamer_alu, streamer_matmul]
+        kernel: [alloc, simple_copy, transform_copy, gemm, rescale, gemmini, streamer_alu, streamer_matmul, tiled_add]
diff --git a/.github/workflows/build-run-kernel.yml b/.github/workflows/build-run-kernel.yml
deleted file mode 100644
index 3cb113ef..00000000
--- a/.github/workflows/build-run-kernel.yml
+++ /dev/null
@@ -1,23 +0,0 @@
-name: Build and run kernels
-
-on:
-  push:
-    branches:
-      - main
-  pull_request:
-
-jobs:
-  build-and-run-kernels:
-    runs-on: ubuntu-24.04
-    steps:
-      - uses: actions/checkout@v3
-      - uses: prefix-dev/setup-pixi@v0.8.1
-        with:
-          cache: true
-          cache-write: ${{ github.event_name == 'push' && github.ref_name == 'main' }}
-      - name: Build and run kernels
-        run: pixi run make allrun
-        working-directory: kernels/${{ matrix.kernel }}
-    strategy:
-      matrix:
-        kernel: [tiled_add]
diff --git a/kernels/gemm/gendata.py b/kernels/gemm/gendata.py
index aa530748..063c2bb0 100755
--- a/kernels/gemm/gendata.py
+++ b/kernels/gemm/gendata.py
@@ -39,5 +39,5 @@ def create_data_files():
         "D_golden": D_golden,
     }
 
-    create_header("data.h", sizes, variables)
-    create_data("data.c", variables)
+    create_header("data", sizes, variables)
+    create_data("data", variables)
diff --git a/kernels/rescale/gendata.py b/kernels/rescale/gendata.py
index ee605de3..04ca03a5 100644
--- a/kernels/rescale/gendata.py
+++ b/kernels/rescale/gendata.py
@@ -69,5 +69,5 @@ def create_data_files():
     sizes = {"MODE": 0, "DATA_LEN": data_len}
     variables = {"A": A, "O": O, "G": G}
 
-    create_header("data.h", sizes, variables)
-    create_data("data.c", variables)
+    create_header("data", sizes, variables)
+    create_data("data", variables)
diff --git a/kernels/simple_copy/gendata.py b/kernels/simple_copy/gendata.py
index c97ecd6f..b70f6bcd 100644
--- a/kernels/simple_copy/gendata.py
+++ b/kernels/simple_copy/gendata.py
@@ -8,5 +8,5 @@ def create_data_files():
     A = np.linspace(1, array_size, array_size, dtype=np.int32)
     sizes = {"N": array_size}
     variables = {"A": A}
-    create_header("data.h", sizes, variables)
-    create_data("data.c", variables)
+    create_header("data", sizes, variables)
+    create_data("data", variables)
diff --git a/kernels/streamer_alu/gendata.py b/kernels/streamer_alu/gendata.py
index 590113c5..e83b8c4a 100644
--- a/kernels/streamer_alu/gendata.py
+++ b/kernels/streamer_alu/gendata.py
@@ -29,5 +29,5 @@ def create_data_files():
     sizes = {"MODE": 0, "DATA_LEN": array_size, "LOOP_ITER": loop_iter}
     variables = {"A": A, "B": B, "O": O, "G": G}
 
-    create_header("data.h", sizes, variables)
-    create_data("data.c", variables)
+    create_header("data", sizes, variables)
+    create_data("data", variables)
diff --git a/kernels/tiled_add/Makefile b/kernels/tiled_add/Makefile
deleted file mode 100644
index 0d471d29..00000000
--- a/kernels/tiled_add/Makefile
+++ /dev/null
@@ -1,61 +0,0 @@
-# Courtesy of Federico Ficarelli
-
-.DEFAULT_GOAL := all
-
-include ../../runtime/snax-alu.rules
-include ../../runtime/Makefile.rules
-
-TESTS =
-TESTS += untiled.acc_dialect.x
-TESTS += tiled.acc_dialect.x
-TESTS += tiled_pipelined.acc_dialect.x
-
-CFLAGS += -std=gnu11
-CFLAGS += -Wall -Wextra -g
-
-MLIRPREPROCFLAGS += --allow-unregistered-dialect
-MLIRPREPROC2FLAGS += --allow-unregistered-dialect
-MLIRPREPROC3FLAGS += --allow-unregistered-dialect
-
-ifdef NO_CHECK
- CFLAGS += -DNO_CHECK
-endif
-
-
-ifdef ACCFGOPT
- ACCFGOPTFLAGS="mlir-opt{executable=mlir-opt generic=true arguments=-cse,-canonicalize,-allow-unregistered-dialect,-mlir-print-op-generic}",accfg-dedup,accfg-config-overlap,
-endif
-
-# Override snax-opt rules to avoid linalg-to-library-call pass
-SNAXOPTACCFLAGS = -p insert-accfg-op{accelerator=snax_alu},convert-linalg-to-accfg,${ACCFGOPTFLAGS}convert-accfg-to-csr,snax-copy-to-dma,memref-to-snax,snax-to-func,clear-memory-space,snax-lower-mcycle
-
-# Set default 
-ARRAY_SIZE ?= 2048
-TILE_SIZE ?= 128
-
-%.preprocfinal.sized.mlir: %.preprocfinal.mlir
-	cat $< | sed -E 's/128/${ARRAY_SIZE}/gm;t' | sed -E 's/%tile_size = arith.constant 16 : index/%tile_size = arith.constant ${TILE_SIZE} : index/gm;t' > $@
-
-%.acc_dialect.snax-opt.mlir: %.preprocfinal.sized.mlir
-	$(SNAXOPT) $(SNAXOPTACCFLAGS) -o $@ $<
-
-data.c data.h:
-	$(PYTHON) gendata.py --array_size=${ARRAY_SIZE}
-
-%.x: %.o main.o data.o
-	$(LD) $(LDFLAGS) $^ -o $@
-
-sim_%: %
-	rm -fr ./logs/
-	$(VLTSIM) $<
-
-RUN = $(addprefix run_, $(TESTS))
-$(RUN): run_%: sim_%
-	mv logs $(subst sim_,,$<).logs
-
-all: $(TESTS)
-
-allrun: $(RUN)
-
-clean:
-	rm -fr *.ll12 *.x *.o *.logs/ logs/ data.h data.c
diff --git a/kernels/tiled_add/Snakefile b/kernels/tiled_add/Snakefile
new file mode 100644
index 00000000..50bd0516
--- /dev/null
+++ b/kernels/tiled_add/Snakefile
@@ -0,0 +1,148 @@
+from util.snake.configs import get_snax_alu_config
+
+config = get_snax_alu_config()
+
+# Untiled example needs preprocessing:
+for entry in config["mlirpreprocflags"]:
+    entry.append("--allow-unregistered-dialect")
+
+
+def get_snax_opt_flags(options):
+    flags = []
+    match options:
+        case "deduponly":
+            flags = ["accfg-dedup"]
+        case "overlaponly":
+            flags = ["accfg-config-overlap"]
+        case "accfgboth":
+            flags = ["accfg-dedup", "accfg-config-overlap"]
+    return ",".join(
+        [
+            "insert-accfg-op{accelerator=snax_alu}",
+            "convert-linalg-to-accfg",
+            "mlir-opt{"
+            + "\ ".join(
+                [
+                    "executable=mlir-opt",
+                    "generic=true",
+                    "arguments='"
+                    + ",".join(
+                        [
+                            "-cse",
+                            "-canonicalize",
+                            "-allow-unregistered-dialect",
+                            "-mlir-print-op-generic",
+                        ]
+                    ),
+                ]
+            )
+            + "'}",
+            *flags,
+            "convert-accfg-to-csr",
+            "snax-copy-to-dma",
+            "memref-to-snax",
+            "snax-to-func",
+            "clear-memory-space",
+            "snax-lower-mcycle",
+        ]
+    )
+
+
+module default_rules:
+    snakefile:
+        "../../util/snake/default_rules.smk"
+    config:
+        config
+
+
+use rule * from default_rules as default_*
+
+
+rule size_mlir:
+    input:
+        "{file}.preprocfinal.mlir",
+    output:
+        temp("{file}_{array_size}_{tile_size}.preprocfinal.sized.mlir"),
+    params:
+        sed_script_array_size=lambda wildcards: f"'s/128/{wildcards.array_size}/gm;t'",
+        sed_script_tile_size=lambda wildcards: f"'s/%tile_size = arith.constant 16 : index/%tile_size = arith.constant {wildcards.tile_size} : index/gm;t'",
+    shell:
+        "cat {input} |"
+        "sed -E {params.sed_script_array_size} |"
+        "sed -E {params.sed_script_tile_size}"
+        "> {output} "
+
+
+rule snax_compile_mlir:
+    input:
+        "{file}.preprocfinal.sized.mlir",
+    output:
+        temp("{file}_{compiler_opt}.snax-opt.mlir"),
+    wildcard_constraints:
+        compiler_opt="noaccfgopt|deduponly|overlaponly|accfgboth",
+    params:
+        # hardcoded to no accfg for now
+        snax_opt_flags=lambda wildcards: get_snax_opt_flags(wildcards.compiler_opt),
+    shell:
+        "{config[snax-opt]} -p {params.snax_opt_flags} -o {output} {input}"
+
+
+rule compile_main:
+    input:
+        "main_{array_size}.rendered.c",
+        "data_{array_size}.o",
+    output:
+        temp("main_{array_size}.o"),
+    shell:
+        "{config[cc]} {config[cflags]} -c {input} -o {output}"
+
+
+rule compile_snax_binary:
+    input:
+        "{file}_{array_size}_{tile_size}_{compiler_opt}.o",
+        "main_{array_size}.o",
+        "data_{array_size}.o",
+    output:
+        "{file}_{array_size}_{tile_size}_{compiler_opt}.x",
+    shell:
+        "{config[ld]} {config[ldflags]} {input} -o {output}"
+
+
+rule render_main:
+    input:
+        "main.c",
+    output:
+        temp("main_{array_size}.rendered.c"),
+    shell:
+        'echo "#include \\"data_{wildcards.array_size}.h\\"" | cat - {input} > {output}'
+
+
+ARRAY_SIZES = ["256"]
+TILE_SIZES = ["16"]
+FILES = ["tiled", "tiled_pipelined", "untiled"]
+COMPILER_OPTS = ["accfgboth", "noaccfgopt"]
+
+
+rule all:
+    input:
+        expand(
+            "{file}_{array_size}_{tile_size}_{compiler_opt}.x",
+            file=FILES,
+            array_size=ARRAY_SIZES,
+            tile_size=TILE_SIZES,
+            compiler_opt=COMPILER_OPTS,
+        ),
+    run:
+        for item in input:
+            shell("{config[vltsim]} {item}")
+
+
+from gendata import generate_data
+
+
+rule generate_data:
+    output:
+        "data_{array_size}.c",
+        "data_{array_size}.h",
+    run:
+        generate_data(int(wildcards.array_size))
diff --git a/kernels/tiled_add/genbenchmark.py b/kernels/tiled_add/genbenchmark.py
deleted file mode 100644
index 0bcef421..00000000
--- a/kernels/tiled_add/genbenchmark.py
+++ /dev/null
@@ -1,53 +0,0 @@
-import pathlib
-
-from util.snax_benchmark import SNAXBenchmark
-
-if __name__ == "__main__":
-
-    def run_all(binary: str, folder: str):
-        binary = "tiled.acc_dialect.x"
-        folder_no_opt = folder + "_no_opt"
-        SIZES = ("ARRAY_SIZE=256", "TILE_SIZE=16", "NO_CHECK=1")
-
-        ## not optimised
-        bm = SNAXBenchmark(
-            kernel="tiled_add",
-            binary=binary,
-            src_dir=str(pathlib.Path.cwd()),
-            export_dir=str(pathlib.Path.cwd()),
-            output_dir=str(pathlib.Path.cwd()),
-        )
-        bm.clean()
-        bm.build(build_opts=[*SIZES])
-        bm.run()
-        bm.trace()
-        bm.process_traces(folder_no_opt)
-        bm.copy_binary(folder_no_opt)
-        bm.copy_logs(folder_no_opt)
-
-        ## optimised
-        folder_opt = folder + "_opt"
-        bm = SNAXBenchmark(
-            kernel="tiled_add",
-            binary=binary,
-            src_dir=str(pathlib.Path.cwd()),
-            export_dir=str(pathlib.Path.cwd()),
-            output_dir=str(pathlib.Path.cwd()),
-        )
-        bm.clean()
-        bm.build(build_opts=[*SIZES, "ACCFGOPT=1"])
-        bm.run()
-
-        bm.trace()
-        bm.process_traces(folder_opt)
-        bm.copy_binary(folder_opt)
-        bm.copy_logs(folder_opt)
-
-    binaries = {
-        "run0": "untiled.acc_dialect.x",
-        "run1": "tiled.acc_dialect.x",
-        "run2": "tiled_pipelined.acc_dialect.x",
-    }
-
-    for folder, binary in binaries.items():
-        run_all(binary, folder)
diff --git a/kernels/tiled_add/gendata.py b/kernels/tiled_add/gendata.py
index 9ab8525d..3c64e826 100755
--- a/kernels/tiled_add/gendata.py
+++ b/kernels/tiled_add/gendata.py
@@ -1,21 +1,11 @@
-import argparse
-
 import numpy as np
 
 from util.gendata import create_data, create_header
 
-if __name__ == "__main__":
-    parser = argparse.ArgumentParser(
-        description="Generate data for snax-alu operations."
-    )
-    parser.add_argument(
-        "--array_size", type=int, default=1024, help="Size of the arrays to generate"
-    )
-    args = parser.parse_args()
 
+def generate_data(array_size):
     low_bound = -128
     high_bound = 127
-    array_size = args.array_size
 
     # snax-alu design-time spatial parallelism
     spatial_par = 4
@@ -37,5 +27,5 @@
     sizes = {"MODE": 0, "DATA_LEN": array_size, "LOOP_ITER": loop_iter}
     variables = {"A": A, "B": B, "O": O, "G": G}
 
-    create_header("data.h", sizes, variables)
-    create_data("data.c", variables)
+    create_header(f"data_{array_size}", sizes, variables)
+    create_data(f"data_{array_size}", variables)
diff --git a/kernels/tiled_add/main.c b/kernels/tiled_add/main.c
index e8def9c2..4f363c1c 100644
--- a/kernels/tiled_add/main.c
+++ b/kernels/tiled_add/main.c
@@ -1,4 +1,3 @@
-#include "data.h"
 #include "memref.h"
 #include "snax_rt.h"
 #include "stdint.h"
diff --git a/kernels/transform_copy/gendata.py b/kernels/transform_copy/gendata.py
index e954e1b4..118bc1e6 100644
--- a/kernels/transform_copy/gendata.py
+++ b/kernels/transform_copy/gendata.py
@@ -13,5 +13,5 @@ def create_files(filename: str):
     B = B.flatten()
     sizes = {"N": array_size, "N_sqrt": sqrt(array_size)}
     variables = {"A": A, "B": B}
-    create_header(f"{filename}.h", sizes, variables)
-    create_data(f"{filename}.c", variables)
+    create_header(filename, sizes, variables)
+    create_data(filename, variables)
diff --git a/tests/benchmark/test_snax_benchmark.py b/tests/benchmark/test_snax_benchmark.py
deleted file mode 100644
index 374e4c68..00000000
--- a/tests/benchmark/test_snax_benchmark.py
+++ /dev/null
@@ -1,22 +0,0 @@
-import pathlib
-
-from util.snax_benchmark import SNAXBenchmark
-
-
-def test_snax_benchmark_runner():
-    folder = "test_run"
-    this_file = pathlib.Path(__file__)
-    bm = SNAXBenchmark(
-        kernel="tiled_add",
-        binary="untiled.acc_dialect.x",
-        src_dir=str(this_file.parent / ".." / ".." / "kernels" / "tiled_add" / ""),
-        export_dir=str(this_file.parent),
-        output_dir=str(this_file.parent),
-    )
-    bm.clean()
-    bm.build(build_opts=["ARRAY_SIZE=128", "TILE_SIZE=16", "NO_CHECK=1"])
-    bm.run()
-    bm.trace()
-    bm.process_traces(folder, accelerator="snax_alu")
-    bm.copy_binary(folder)
-    bm.copy_logs(folder)
diff --git a/util/gendata.py b/util/gendata.py
index b1c97c2f..1cb2ee5d 100644
--- a/util/gendata.py
+++ b/util/gendata.py
@@ -7,9 +7,10 @@
 def create_header(
     file_name: str, sizes: dict[str, int], variables: dict[str, npt.NDArray]
 ) -> None:
-    if os.path.dirname(file_name):
-        os.makedirs(os.path.dirname(file_name), exist_ok=True)
-    with open(file_name, "w") as f:
+    header_file = f"{file_name}.h"
+    if os.path.dirname(header_file):
+        os.makedirs(os.path.dirname(header_file), exist_ok=True)
+    with open(header_file, "w") as f:
         includes = ["#include <stdint.h>", "#pragma once", ""]
         includes = "\n".join(includes)
         variables_string = [""]
@@ -25,12 +26,13 @@ def create_header(
 
 
 def create_data(file_name: str, variables: dict[str, npt.NDArray]):
-    includes = ['#include "data.h"', "", ""]
+    includes = [f'#include "{os.path.basename(file_name)}.h"', "", ""]
     includes = "\n".join(includes)
+    c_file = f"{file_name}.c"
     variables = {i: np.reshape(j, j.size) for i, j in variables.items()}
-    if os.path.dirname(file_name):
-        os.makedirs(os.path.dirname(file_name), exist_ok=True)
-    with open(file_name, "w") as f:
+    if os.path.dirname(c_file):
+        os.makedirs(os.path.dirname(c_file), exist_ok=True)
+    with open(c_file, "w") as f:
         f.write(includes)
         for variable_name, variable_value in variables.items():
             f.write(