--- Day changed Sat Dec 09 2017 01:12 -!- cloudcell__ [~cloudcell@81.23.9.160] has quit [Remote host closed the connection] 01:13 -!- cloudcell__ [~cloudcell@81.23.9.160] has joined #secp256k1 01:41 -!- cloudcell__ [~cloudcell@81.23.9.160] has quit [Read error: Connection reset by peer] 01:41 -!- cloudcell__ [~cloudcell@81.23.9.160] has joined #secp256k1 01:43 -!- cloudcell__ [~cloudcell@81.23.9.160] has quit [Read error: Connection reset by peer] 01:44 -!- cloudcell__ [~cloudcell@81.23.9.160] has joined #secp256k1 01:48 -!- cloudcell__ [~cloudcell@81.23.9.160] has quit [Remote host closed the connection] 01:48 -!- cloudcell__ [~cloudcell@81.23.9.160] has joined #secp256k1 02:06 -!- cloudcell_ [~cloudcell@81.23.9.160] has joined #secp256k1 02:06 -!- cloudcell__ [~cloudcell@81.23.9.160] has quit [Ping timeout: 248 seconds] 02:06 -!- cloudcell_ is now known as Guest57323 02:13 -!- Guest57323 [~cloudcell@81.23.9.160] has quit [Read error: Connection reset by peer] 02:13 -!- Guest57323 [~cloudcell@81.23.9.160] has joined #secp256k1 03:00 -!- SopaXorzTaker [~SopaXorzT@unaffiliated/sopaxorztaker] has quit [Remote host closed the connection] 03:02 -!- SopaXorzTaker [~SopaXorzT@unaffiliated/sopaxorztaker] has joined #secp256k1 04:06 -!- Guest57323 [~cloudcell@81.23.9.160] has quit [Remote host closed the connection] 07:42 -!- instagibbs [~instagibb@pool-100-15-116-35.washdc.fios.verizon.net] has quit [Ping timeout: 268 seconds] 08:34 -!- instagibbs [~instagibb@pool-100-15-116-35.washdc.fios.verizon.net] has joined #secp256k1 12:20 -!- SopaXorzTaker [~SopaXorzT@unaffiliated/sopaxorztaker] has quit [Quit: Leaving] 13:15 < gmaxwell> andytoshi, sipa: here is an interesting realization: The column sums (buckets) in wnaf-pippenger can run perfectly parallel with each other. The computation graph bucketizes things, then colum sums the buckets then becomes sequential when summed into the final accumulator, due to the doubling. 13:16 < gmaxwell> This means that if we created AVX FE add/mul/normalize that ran 4 or 8 field elements at a time, we could have 4/8 way group law, and probably effectively use SIMD for pippenger, for maybe a 2x speedup. 13:17 < gmaxwell> though with a considerable amount of work. 13:46 < sipa> hmm 13:48 < sipa> only 32-bit operations in SSE and AVX, though 13:49 < sipa> and fe_10x26 is about 4x slower than fe_5x52 14:15 -!- eck [~eck@fsf/member/eck] has joined #secp256k1 17:14 < gmaxwell> unrelated, this thread is live again: https://bitcointalk.org/index.php?topic=1875417.msg20697287#msg20697287 22:11 -!- MrBismuth [~ArcMrBism@2602:306:8325:a300:1426:998c:b44b:81fb] has quit [Read error: Connection reset by peer] 22:12 -!- MrBismuth [~ArcMrBism@2602:306:8325:a300:1426:998c:b44b:81fb] has joined #secp256k1 22:34 -!- maaku [~maaku@173.234.25.100] has quit [Quit: ZNC - http://znc.in] 22:41 -!- maaku [~maaku@173.234.25.100] has joined #secp256k1