================
@@ -3928,6 +3928,11 @@ static StringRef getCanonicalArchString(Compilation &C,
return StringRef();
}
}
+ } else if (Triple.isSPIRV() && ArchStr.starts_with("igca_") &&
----------------
ianayl wrote:
> We should have a definitive way of identifying targets that are appropriate
> for IGCA. If we don't (and the only reason we don't is because we have so far
> refused to make use of the vendor and OS components to identify Intel devices
> for reasons that remain unclear to me),
That would certain help with figuring out if IGCA is even appropriate for the
context, as of right now this check here for "igca" is my only hint as to if
Intel GPUs/ IGCA is appropriate for the context: For PR #227877 I'm using this
same section of code to emit a list of devices
https://github.com/llvm/llvm-project/pull/227877/changes#diff-f8c64e36dfe3828a6f816c4550e78bb0305769ace1be53207e86ac9a3280ac9eR3961
since triples and TargetInfo don't necessarily need to contain any indication
we're working with Intel GPUs unlike what NVIDIA is doing
As for this PR however, could we leave this as a TODO for now and address this
later? Sorry for suggesting yet another TODO, but IIUC this change is blocking
other work 😅
https://github.com/llvm/llvm-project/pull/227468
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
