================
@@ -59,28 +62,18 @@ INTEL_GPU("xe-bmg-g21", XE_BMG_G21, 20, 1, 40, Render)
INTEL_GPU("xe-arl-h", XE_ARL_H, 12, 74, 35, Render)
INTEL_GPU("xe-mtl-h", XE_MTL_H, 12, 71, 30, Render)
INTEL_GPU("xe-mtl-u", XE_MTL_U, 12, 70, 30, Render)
-INTEL_GPU("xe-arl-u", XE_ARL_U, 12, 70, 30, Render)
-INTEL_GPU("xe-arl-s", XE_ARL_S, 12, 70, 30, Render)
INTEL_GPU("xe-pvc-vg", XE_PVC_VG, 12, 61, 20, ComputeExact)
INTEL_GPU("xe-pvc", XE_PVC, 12, 60, 20, ComputeExact)
INTEL_GPU("xe-pvc-sdv", XE_PVC_SDV, 12, 60, 20, ComputeExact)
INTEL_GPU("xe-acm-g12", XE_ACM_G12, 12, 57, 15, RenderExact)
-INTEL_GPU("xe-dg2-g12", XE_DG2_G12, 12, 57, 15, RenderExact)
INTEL_GPU("xe-acm-g11", XE_ACM_G11, 12, 56, 15, RenderExact)
-INTEL_GPU("xe-dg2-g11", XE_DG2_G11, 12, 56, 15, RenderExact)
-INTEL_GPU("xe-ats-m75", XE_ATS_M75, 12, 56, 15, RenderExact)
INTEL_GPU("xe-acm-g10", XE_ACM_G10, 12, 55, 15, RenderExact)
-INTEL_GPU("xe-dg2-g10", XE_DG2_G10, 12, 55, 15, RenderExact)
-INTEL_GPU("xe-ats-m150", XE_ATS_M150, 12, 55, 15, RenderExact)
INTEL_GPU("xe-dg1", XE_DG1, 12, 10, 10, Render)
INTEL_GPU("xe-adl-n", XE_ADL_N, 12, 4, 10, Render)
INTEL_GPU("xe-adl-p", XE_ADL_P, 12, 3, 10, Render)
-INTEL_GPU("xe-rpl-p", XE_RPL_P, 12, 3, 10, Render)
INTEL_GPU("xe-adl-s", XE_ADL_S, 12, 2, 10, Render)
-INTEL_GPU("xe-rpl-s", XE_RPL_S, 12, 2, 10, Render)
INTEL_GPU("xe-rkl", XE_RKL, 12, 1, 10, Render)
INTEL_GPU("xe-tgllp", XE_TGLLP, 12, 0, 10, Render)
----------------
gmlueck wrote:
The `ocloc` tool accepts both `tgl` and `tgllp`. That team is adding `xe-`
prefixes to all the device names, so `ocloc` will accepts both `xe-tgl` and
`xe-tgllp` soon, and the `ocloc` help message will show both as valid device
names. Wouldn't we want to accept the same device names in clang?
https://github.com/llvm/llvm-project/pull/229247
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
