Thanks Jeff. The vfmax.vf/vfmin.vf failure are still here, I will continue to 
take care of it.
And all vx_vf/vx* related test cases should be Ok now.

Pan

-----Original Message-----
From: Jeffrey Law <[email protected]> 
Sent: Thursday, February 5, 2026 12:16 AM
To: Li, Pan2 <[email protected]>; [email protected]
Cc: [email protected]; [email protected]; [email protected]; 
[email protected]; Chen, Ken <[email protected]>; Liu, Hongtao 
<[email protected]>
Subject: Re: [PATCH v1 0/6] Add vr2gpr and vr2fpr cost value param



On 2/4/2026 4:47 AM, [email protected] wrote:
> From: Pan Li <[email protected]>
>
> The enhanced middle-end record_stmt_cost introduce slp_node and its'
> vectype for cost model, that make the vr to scalar cost is counted
> during backend adjust cost.  Thus, to make test happy, add param
> include both the vr2gpr and vr2fpr cost.
>
> Pan Li (6):
>    RISC-V: Introduce vr2gpr-cost= for customizing the cost when vr2gpr 
> [PR123916]
>    RISC-V: Adjust the vx-[456]* testcase by param=vr2gpr-cost [PR123916]
>    RISC-V: Introduce vr2fpr-cost= for customizing the cost when vr2fpr 
> [PR123916]
>    RISC-V: Adjust the vf-[378]* testcase by param=vr2fpr-cost [PR123916]
>    RISC-V: Add test case for PR123916
>    RISC-V: Adjust testcase asm check for vx-[56]-i[16|8].c [PR123916]
This series is fine.  Thanks for tacking care of it.

jeff

Reply via email to