| Date | Wed, 24 Oct 2012 15:51:00 -0700 | From | "H. Peter Anvin" <> | Subject | Re: [PATCH 02/16] math128, x86_64: Implement {mul,add}_u128 in 64bit asm |
| |
On 10/24/2012 03:47 PM, Juri Lelli wrote: >> >> How could this work since u128 presumably has not yet been defined as a >> structure? After all, isn't it the absence of ARCH_HAS_INT128 which >> makes that happen? > > Sorry, you were not in the Cc list of the previous patch in the patchset, > so you probably missed that. I should have triple checked git send-email > Cc list. Sorry about that. > > I'll add you there. >
Hmm... you realize that at least on some platform the u128 as a union is going to perform worse than the plain __int128, right? As such IMO it would be better if the union was only defined if needed.
-hpa
|