(強いAI)技術的特異点/(世界加速) 23at FUTURE
(強いAI)技術的特異点/(世界加速) 23 - 暇つぶし2ch467:YAMAGUTIseisei
18/08/06 01:46:49.27 FnAR0u04o BE:73431465-2BP(3)
Listing 1 is Verilog that generates the ``next readys'' for an N-entry parallel scheduler.
Although there are four ready event input types (predicate true, false, operand #0, operand #1),
by ensuring that predicate target events never occur in the same cycle as operand target events, a single target index bit suffices to distinguish false/operand #0 targets from true/operand #1 targets.
(Further decoding is provided by specific {RT/RF/R0/R1} ENs enables.) Therefore for an instruction window with N=32 entries, T0 and T1 are six bits {input#:1; IID:5}.
The target decoders T00, T01, T10, T11 (target-0input-0, etc.) are each one 6-LUT, as is the broadcast select decoder B.
The next active ready state logic folds together the target decoder outputs with current active and decoded ready state.
This requires another seven LUTs (two for INH NXT), for a total of 32*12 = 384 LUTs.
This may be improved by splitting the 32-entry scheduler into two 16-entry banks of even and odd instructions.
Within a bank a 4-bit bank-IID suffices.
Then T0, T1 narrow to five bits so T00, T01, T10, T11 fit in two 5,5-LUTs, and INH NXT in one 6-LUT, or 2*16*(3+6)=288 LUTs in all.


次ページ
最新レス表示
レスジャンプ
類似スレ一覧
スレッドの検索
話題のニュース
おまかせリスト
オプション
しおりを挟む
スレッドに書込
スレッドの一覧
暇つぶし2ch