fix(sycl): stop building the ggml CPU variant matrix with icpx

.docker/llama-cpp-build-target.sh

@@ -4,6 +4,17 @@ set -euo pipefail
 arch=${1:?target architecture is required}
 build_type=${2-}
 
+# SYCL compiles the whole tree with icpx -fsycl, and icpx never finishes
+# ggml-cpu/arch/x86/repack.cpp at -march=sapphirerapids: the job sits on that one
+# translation unit until GitHub kills it at 6h. gcc builds the same file in
+# seconds, so only the SYCL images have to give up the CPU variant matrix.
+case "$build_type" in
+  sycl*)
+    echo llama-cpp-fallback
+    exit 0
+    ;;
+esac
+
 # GPU arm64 base images do not consistently provide the gcc-14 toolchain needed
 # to compile ggml's armv9.2 CPU variants. Keep their portable fallback until the
 # builder images can supply that compiler.

.docker/turboquant-build-target.sh

@@ -4,6 +4,17 @@ set -euo pipefail
 arch=${1:?target architecture is required}
 build_type=${2-}
 
+# SYCL compiles the whole tree with icpx -fsycl, and icpx never finishes
+# ggml-cpu/arch/x86/repack.cpp at -march=sapphirerapids: the job sits on that one
+# translation unit until GitHub kills it at 6h. gcc builds the same file in
+# seconds, so only the SYCL images have to give up the CPU variant matrix.
+case "$build_type" in
+  sycl*)
+    echo turboquant-fallback
+    exit 0
+    ;;
+esac
+
 # GPU arm64 base images do not consistently provide the gcc-14 toolchain needed
 # to compile ggml's armv9.2 CPU variants. Keep their portable fallback until the
 # builder images can supply that compiler.

backend/cpp/llama-cpp/run.sh

@@ -12,10 +12,11 @@ grep -e "flags" /proc/cpuinfo | head -1
 
 BINARY=llama-cpp-fallback
 
-# CPU images and x86 GPU images ship a single llama-cpp-cpu-all built with ggml
+# CPU images and most x86 GPU images ship a single llama-cpp-cpu-all built with ggml
 # CPU_ALL_VARIANTS: ggml's backend registry dlopens the best libggml-cpu-*.so for this
 # host, so no shell-side AVX probing. GPU arm64 images still ship llama-cpp-fallback
-# until their builder toolchains support ggml's complete arm variant matrix.
+# until their builder toolchains support ggml's complete arm variant matrix, and so do
+# the SYCL images, whose icpx compiler hangs on the sapphirerapids variant.
 if [ -e "$CURDIR"/llama-cpp-cpu-all ]; then
 	BINARY=llama-cpp-cpu-all
 fi

backend/cpp/turboquant/run.sh

@@ -12,11 +12,12 @@ grep -e "flags" /proc/cpuinfo | head -1
 
 BINARY=turboquant-fallback
 
-# CPU images and x86 GPU images ship a single turboquant-cpu-all built with ggml
+# CPU images and most x86 GPU images ship a single turboquant-cpu-all built with ggml
 # CPU_ALL_VARIANTS: ggml's
 # backend registry dlopens the best libggml-cpu-*.so for this host, so no shell-side
 # probing. GPU arm64 images still ship turboquant-fallback until their builder toolchains
-# support ggml's complete arm variant matrix.
+# support ggml's complete arm variant matrix, and so do the SYCL images, whose icpx
+# compiler hangs on the sapphirerapids variant.
 if [ -e "$CURDIR"/turboquant-cpu-all ]; then
 	BINARY=turboquant-cpu-all
 fi

scripts/build/llama-cpp-build-target_test.sh

@@ -23,4 +23,10 @@ assert_target amd64 "" llama-cpp-cpu-all
 assert_target arm64 cublas llama-cpp-fallback
 assert_target arm64 "" llama-cpp-cpu-all
 
+# SYCL builds the whole tree with icpx -fsycl, and icpx never finishes
+# ggml-cpu/arch/x86/repack.cpp at -march=sapphirerapids: every sycl job sat on
+# that one translation unit until GitHub killed it at its 6h limit.
+assert_target amd64 sycl_f16 llama-cpp-fallback
+assert_target amd64 sycl_f32 llama-cpp-fallback
+
 echo "PASS: llama.cpp build target preserves CPU variants where supported"

scripts/build/turboquant-build-target_test.sh

@@ -23,4 +23,10 @@ assert_target amd64 "" turboquant-cpu-all
 assert_target arm64 cublas turboquant-fallback
 assert_target arm64 "" turboquant-cpu-all
 
+# SYCL builds the whole tree with icpx -fsycl, and icpx never finishes
+# ggml-cpu/arch/x86/repack.cpp at -march=sapphirerapids: every sycl job sat on
+# that one translation unit until GitHub killed it at its 6h limit.
+assert_target amd64 sycl_f16 turboquant-fallback
+assert_target amd64 sycl_f32 turboquant-fallback
+
 echo "PASS: turboquant build target preserves CPU variants where supported"