Message ID | 20211203000103.737833-1-adhemerval.zanella@linaro.org |
---|---|
Headers | show |
Series | Improve hypot | expand |
Dear Adhemerval, a few typos in your commit message: > when fast FMA is avaliable should be "available" > Use a newer algotihm should be "algorithm" interger -> integer > - An improvement from 233442 to 1274 results with 1 ulp of > error for Binary96 (x86_64). > - An improvement from 453045 to 1294 results with 1 ulp of > error for Binary96 (x86_64). should the latter be binary128? > If FMA is uses should be "used" adapated -> adapted > If FMA is uses the binary64 shows a slight worse precision: I confirm, I find: binary64 0.948812 -0x0.5a22c27a3893p-1022,0x0.9cfea180c00dap-1022 Best regards, Paul
On 03/12/2021 05:51, Paul Zimmermann wrote: > Dear Adhemerval, > > a few typos in your commit message: > >> when fast FMA is avaliable > should be "available" > >> Use a newer algotihm > should be "algorithm" > > interger -> integer > >> - An improvement from 233442 to 1274 results with 1 ulp of >> error for Binary96 (x86_64). >> - An improvement from 453045 to 1294 results with 1 ulp of >> error for Binary96 (x86_64). > > should the latter be binary128? Yeah, indeed. > >> If FMA is uses > should be "used" > > adapated -> adapted > >> If FMA is uses the binary64 shows a slight worse precision: > > I confirm, I find: > > binary64 0.948812 -0x0.5a22c27a3893p-1022,0x0.9cfea180c00dap-1022 > > Best regards, > Paul > Thanks for checking on it.