diff --git a/sycl/test-e2e/BFloat16/bfloat16_builtins.cpp b/sycl/test-e2e/BFloat16/bfloat16_builtins.cpp
index 1115cc618ae30..b96882aa86851 100644
--- a/sycl/test-e2e/BFloat16/bfloat16_builtins.cpp
+++ b/sycl/test-e2e/BFloat16/bfloat16_builtins.cpp
@@ -14,6 +14,8 @@
 
 // Currently the feature isn't supported on FPGA.
 // UNSUPPORTED: accelerator
+// FIXME: enable opaque pointers support on CPU.
+// XFAIL: cpu
 #include "bfloat16_builtins.hpp"
 
 int main() {
diff --git a/sycl/test-e2e/BFloat16/bfloat16_builtins_cuda_generic.cpp b/sycl/test-e2e/BFloat16/bfloat16_builtins_cuda_generic.cpp
index 3e2b6b1aae433..06621c8d001ba 100644
--- a/sycl/test-e2e/BFloat16/bfloat16_builtins_cuda_generic.cpp
+++ b/sycl/test-e2e/BFloat16/bfloat16_builtins_cuda_generic.cpp
@@ -12,6 +12,8 @@
 
 // Currently the feature isn't supported on FPGA.
 // UNSUPPORTED: accelerator
+// FIXME: enable opaque pointers support on CPU.
+// XFAIL: cpu
 #include "bfloat16_builtins.hpp"
 
 int main() {
diff --git a/sycl/test-e2e/BFloat16/bfloat16_type.cpp b/sycl/test-e2e/BFloat16/bfloat16_type.cpp
index 73936eb5dc4e7..0aef086807f73 100644
--- a/sycl/test-e2e/BFloat16/bfloat16_type.cpp
+++ b/sycl/test-e2e/BFloat16/bfloat16_type.cpp
@@ -7,6 +7,7 @@
 // UNSUPPORTED: accelerator
 
 // FIXME: enable opaque pointers support on CPU.
+// XFAIL: cpu
 
 //==----------- bfloat16_type.cpp - SYCL bfloat16 type test ----------------==//
 //
diff --git a/sycl/test-e2e/syclcompat/atomic/atomic_class.cpp b/sycl/test-e2e/syclcompat/atomic/atomic_class.cpp
index 22522ea09eeb4..34ef1ea9719cd 100644
--- a/sycl/test-e2e/syclcompat/atomic/atomic_class.cpp
+++ b/sycl/test-e2e/syclcompat/atomic/atomic_class.cpp
@@ -32,7 +32,7 @@
 
 // UNSUPPORTED: hip || (windows && level_zero)
 
-// RUN: %clangxx -std=c++20 -fsycl -fsycl-targets=%{sycl_triple} %if any-device-is-cuda %{ -Xsycl-target-backend --cuda-gpu-arch=sm_70 %} %s -o %t.out
+// RUN: %clangxx -std=c++20 -fsycl -fsycl-targets=%{sycl_triple} %if any-device-is-cuda %{ -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70 %} %s -o %t.out
 // RUN: %{run} %t.out
 
 #include <sycl/detail/core.hpp>
diff --git a/sycl/test-e2e/syclcompat/atomic/atomic_memory_acq_rel.cpp b/sycl/test-e2e/syclcompat/atomic/atomic_memory_acq_rel.cpp
index 2acaa85022f0b..047d490140760 100644
--- a/sycl/test-e2e/syclcompat/atomic/atomic_memory_acq_rel.cpp
+++ b/sycl/test-e2e/syclcompat/atomic/atomic_memory_acq_rel.cpp
@@ -32,7 +32,7 @@
 
 // UNSUPPORTED: hip
 
-// RUN: %clangxx -fsycl -fsycl-targets=%{sycl_triple} %if any-device-is-cuda %{ -Xsycl-target-backend --cuda-gpu-arch=sm_70 %} %s -o %t.out
+// RUN: %clangxx -fsycl -fsycl-targets=%{sycl_triple} %if any-device-is-cuda %{ -Xsycl-target-backend=nvptx64-nvidia-cuda --cuda-gpu-arch=sm_70 %} %s -o %t.out
 // RUN: %{run} %t.out
 
 #include <iostream>