Matches for from:mjr, 5343 total results Sorted by newest | relevance
Thu Jul 18 10:50:54 UTC 2013 <mjr_> very coool
Thu Jul 18 10:50:38 UTC 2013 <mjr_> very cool
Thu Jul 18 10:49:24 UTC 2013 <mjr_> very cool
Thu Jul 18 10:49:23 UTC 2013 <mjr_> yeah
Thu Jul 18 10:46:22 UTC 2013 <mjr_> just 45
Thu Jul 18 10:46:20 UTC 2013 <mjr_> i don't see 28 on their site
Thu Jul 18 10:46:05 UTC 2013 <mjr_> so wait
Thu Jul 18 10:45:26 UTC 2013 <mjr_> that is quite impressive results
Thu Jul 18 10:45:14 UTC 2013 <mjr_> but you have to maintain state unfortunately, and keep track of where the first half is
Thu Jul 18 10:44:54 UTC 2013 <mjr_> you can split the hash into two parts and only hash the second half
Thu Jul 18 10:44:29 UTC 2013 <mjr_> but if you figure out the timing issues
Thu Jul 18 10:44:12 UTC 2013 <mjr_> which i think on asic is implemented as psram
Thu Jul 18 10:44:02 UTC 2013 <mjr_> blockram
Thu Jul 18 10:44:00 UTC 2013 <mjr_> that is the key to it all
Thu Jul 18 10:43:52 UTC 2013 <mjr_> SoC
Thu Jul 18 10:43:48 UTC 2013 <mjr_> cool
Thu Jul 18 10:42:39 UTC 2013 <mjr_> which is why you can't combine both very easily
Thu Jul 18 10:42:30 UTC 2013 <mjr_> he said that clock mgmt is kinda tricky
Thu Jul 18 10:42:16 UTC 2013 <mjr_> i discussed a lot with yifu, pipelining vs splitting
Thu Jul 18 10:42:05 UTC 2013 <mjr_> i am really curious about how they implemented their code