gpu-compute: Number of TLBs equal to number of CUs
authorgauravjain14 <gjain6@wisc.edu>
Mon, 7 Oct 2019 05:18:44 +0000 (00:18 -0500)
committerMatt Sinclair <mattdsinclair@gmail.com>
Fri, 11 Dec 2020 05:03:40 +0000 (05:03 +0000)
The n_cu variable in GPUTLBConifig.py did not take
the number of CUs into consideration and instead
calculated the number of TLBs using cu_per_sa,
sa_per_complex, num_gpu_complexes. Thus changing
the number of cus (n_cus) and none of the other flags
resulted in a  segmentation fault since the required
TLBs were not being instantiated

Change-Id: I569a4e6dc7db9b7a81aeede5ac68aacc0f400a5e
Reviewed-on: https://gem5-review.googlesource.com/c/public/gem5/+/32035
Maintainer: Matt Sinclair <mattdsinclair@gmail.com>
Maintainer: Anthony Gutierrez <anthony.gutierrez@amd.com>
Reviewed-by: Alexandru Duțu <alexandru.dutu@amd.com>
Reviewed-by: Anthony Gutierrez <anthony.gutierrez@amd.com>
Reviewed-by: Matt Sinclair <mattdsinclair@gmail.com>
Reviewed-by: Jason Lowe-Power <power.jg@gmail.com>
Tested-by: kokoro <noreply+kokoro@google.com>
configs/common/GPUTLBConfig.py

index 8e2b1e46e9c10c80e8f1c89f14896f82faa7bcf1..c06bda1a25122f2375affa4dc28f9653d8e93875 100644 (file)
@@ -74,8 +74,7 @@ def create_TLB_Coalescer(options, my_level, my_index, tlb_name,
         coalescer_name.append(eval(Coalescer_constructor(my_level)))
 
 def config_tlb_hierarchy(options, system, shader_idx):
-    n_cu = options.cu_per_sa * options.sa_per_complex * \
-           options.num_gpu_complexes
+    n_cu = options.num_compute_units
 
     if options.TLB_config == "perLane":
         num_TLBs = 64 * n_cu