diff --git a/.github/workflows/benchmark.yml b/.github/workflows/benchmark.yml
index 5ee7729074..d21637ce66 100644
--- a/.github/workflows/benchmark.yml
+++ b/.github/workflows/benchmark.yml
@@ -206,11 +206,11 @@ jobs:
     - name: Test tensor cores
       run: |
         HSA=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
-        KFD=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
+        AMD=1 python3 test/test_linearizer.py TestLinearizer.test_tensor_cores TestLinearizer.test_tensor_cores_padded
     - name: Run Tensor Core GEMM (HSA)
       run: HSA=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul.txt
-    - name: Run Tensor Core GEMM (KFD)
-      run: KFD=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_kfd.txt
+    - name: Run Tensor Core GEMM (AMD)
+      run: AMD=1 HALF=1 DEBUG=2 python3 extra/gemm/simple_matmul.py | tee matmul_kfd.txt
     # TODO: AMD compiler bug causes this to fail
     #- name: Fuzz Padded Tensor Core GEMM
     #  run: HSA=1 M_START=12 M_STOP=20 M_STEP=1 N_START=12 N_STOP=20 N_STEP=1 K_START=28 K_STOP=36 K_STEP=1 HALF=1 TC_OPT=2 DEBUG=2 python3 ./extra/gemm/fuzz_matmul.py
diff --git a/tinygrad/features/search.py b/tinygrad/features/search.py
index 5f0799e322..b7bd6d9f92 100644
--- a/tinygrad/features/search.py
+++ b/tinygrad/features/search.py
@@ -111,7 +111,7 @@ def beam_search(lin:Linearizer, rawbufs:List[Buffer], amt:int, allow_test_size=T
   beam: List[Tuple[Linearizer, float]] = []
   seen_libs = set()
 
-  default_parallel, min_progress_micros = 1 if lin.opts.device in {"CUDA", "HSA", "KFD", "NV"} else 0, getenv("BEAM_MIN_PROGRESS",0.01)
+  default_parallel, min_progress_micros = 1 if lin.opts.device in {"CUDA", "HSA", "AMD", "NV"} else 0, getenv("BEAM_MIN_PROGRESS",0.01)
   if beam_pool is None and getenv("PARALLEL", default_parallel):
     beam_pool = multiprocessing.get_context("spawn").Pool(multiprocessing.cpu_count(), _init_worker, (), getenv("BEAM_MAX_TASKS_PER_CHILD", 16))