diff --git a/.github/workflows/sycl_linux_build_and_test.yml b/.github/workflows/sycl_linux_build_and_test.yml
index 29f4968c00f53..a50531b9ce8e3 100644
--- a/.github/workflows/sycl_linux_build_and_test.yml
+++ b/.github/workflows/sycl_linux_build_and_test.yml
@@ -40,7 +40,7 @@ on:
       build_configure_extra_args:
         type: string
         required: false
-        default: "--hip --hip-amd-arch=gfx906 --cuda"
+        default: "--hip --cuda"
       build_artifact_suffix:
         type: string
         required: true
@@ -86,7 +86,7 @@ jobs:
               \"build_cache_root\":\"/__w/\",
               \"build_cache_suffix\":\"default\",
               \"build_cache_size\":\"2G\",
-              \"build_configure_extra_args\":\"--hip --hip-amd-arch=gfx906 --cuda\",
+              \"build_configure_extra_args\":\"--hip --cuda\",
               \"build_artifact_suffix\":\"default\",
               \"build_upload_artifact\":\"false\",
               \"intel_drivers_image\":\"ghcr.io/intel/llvm/ubuntu2004_intel_drivers:latest\",
diff --git a/buildbot/configure.py b/buildbot/configure.py
index 77f1ff5ef4463..4e02e6046fd62 100644
--- a/buildbot/configure.py
+++ b/buildbot/configure.py
@@ -66,8 +66,6 @@ def do_configure(args):
         if args.hip_platform == 'AMD':
             llvm_targets_to_build += ';AMDGPU'
             libclc_targets_to_build += libclc_amd_target_names
-            if args.hip_amd_arch:
-                sycl_clang_extra_flags += "-Xsycl-target-backend=amdgcn-amd-amdhsa --offload-arch="+args.hip_amd_arch
 
             # The HIP plugin for AMD uses lld for linking
             llvm_enable_projects += ';lld'
@@ -211,7 +209,6 @@ def main():
     parser.add_argument("--cuda", action='store_true', help="switch from OpenCL to CUDA")
     parser.add_argument("--hip", action='store_true', help="switch from OpenCL to HIP")
     parser.add_argument("--hip-platform", type=str, choices=['AMD', 'NVIDIA'], default='AMD', help="choose hardware platform for HIP backend")
-    parser.add_argument("--hip-amd-arch", type=str, help="Sets AMD gpu architecture for llvm lit tests, this is only needed for the HIP backend and AMD platform")
     parser.add_argument("--arm", action='store_true', help="build ARM support rather than x86")
     parser.add_argument("--enable-esimd-emulator", action='store_true', help="build with ESIMD emulation support")
     parser.add_argument("--no-assertions", action='store_true', help="build without assertions")
diff --git a/sycl/doc/GetStartedGuide.md b/sycl/doc/GetStartedGuide.md
index 5e561780db1dd..6bf8aead56624 100644
--- a/sycl/doc/GetStartedGuide.md
+++ b/sycl/doc/GetStartedGuide.md
@@ -502,10 +502,10 @@ skipped.
 If CUDA support has been built, it is tested only if there are CUDA devices
 available.
 
-If testing with HIP for AMD make sure to specify the GPU being used
-by adding `-hip-amd-arch=<target>`to buildbot/configure.py or add 
-`-Xsycl-target-backend=amdgcn-amd-amdhsa --offload-arch=<target>` 
-to the CMake variable `SYCL_CLANG_EXTRA_FLAGS`.
+If testing with HIP for AMD, the lit tests will use `gfx906` as the default
+architecture. It is possible to change it by adding
+`-Xsycl-target-backend=amdgcn-amd-amdhsa --offload-arch=<target>` to the CMake
+variable `SYCL_CLANG_EXTRA_FLAGS`.
 
 #### Run DPC++ E2E test suite
 
diff --git a/sycl/test/lit.cfg.py b/sycl/test/lit.cfg.py
index 57d7f8ffaed55..e90c426f5f7b5 100644
--- a/sycl/test/lit.cfg.py
+++ b/sycl/test/lit.cfg.py
@@ -35,8 +35,6 @@
 # test_exec_root: The root path where tests should be run.
 config.test_exec_root = os.path.join(config.sycl_obj_root, 'test')
 
-llvm_config.use_clang(additional_flags=config.sycl_clang_extra_flags.split(' '))
-
 # Propagate some variables from the host environment.
 llvm_config.with_system_environment(['PATH', 'OCL_ICD_FILENAMES', 'SYCL_DEVICE_ALLOWLIST', 'SYCL_CONFIG_FILE_NAME'])
 
@@ -100,6 +98,8 @@
 lit_config.note("Triple: {}".format(triple))
 config.substitutions.append( ('%sycl_triple',  triple ) )
 
+additional_flags = config.sycl_clang_extra_flags.split(' ')
+
 if config.cuda_be == "ON":
     config.available_features.add('cuda_be')
 
@@ -115,12 +115,13 @@
 if triple == 'amdgcn-amd-amdhsa':
     config.available_features.add('hip_amd')
     # For AMD the specific GPU has to be specified with --offload-arch
-    if not re.match('.*--offload-arch.*', config.sycl_clang_extra_flags):
-        raise Exception("Error: missing --offload-arch flag when trying to "  \
-                        "run lit tests for AMD GPU, please add "              \
-                        "--hip-amd-arch=<target> to buildbot/configure.py or add"  \
-                        "`-Xsycl-target-backend=amdgcn-amd-amdhsa --offload-arch=<target>` to " \
-                        "the CMake variable SYCL_CLANG_EXTRA_FLAGS")
+    if not any([f.startswith('--offload-arch') for f in additional_flags]):
+        # If the offload arch wasn't specified in SYCL_CLANG_EXTRA_FLAGS,
+        # hardcode it to gfx906, this is fine because only compiler tests
+        additional_flags += ['-Xsycl-target-backend=amdgcn-amd-amdhsa',
+                            '--offload-arch=gfx906']
+
+llvm_config.use_clang(additional_flags=additional_flags)
 
 # Set timeout for test = 10 mins
 try: