Deprecated: The each() function is deprecated. This message will be suppressed on further calls in /home/zhenxiangba/zhenxiangba.com/public_html/phproxy-improved-master/index.php on line 456
JPH06103495B2 - Information processing equipment - Google Patents
[go: Go Back, main page]

JPH06103495B2 - Information processing equipment - Google Patents

Information processing equipment

Info

Publication number
JPH06103495B2
JPH06103495B2 JP62030916A JP3091687A JPH06103495B2 JP H06103495 B2 JPH06103495 B2 JP H06103495B2 JP 62030916 A JP62030916 A JP 62030916A JP 3091687 A JP3091687 A JP 3091687A JP H06103495 B2 JPH06103495 B2 JP H06103495B2
Authority
JP
Japan
Prior art keywords
instruction
vector
read
acquisition means
vector register
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Expired - Lifetime
Application number
JP62030916A
Other languages
Japanese (ja)
Other versions
JPS63197273A (en
Inventor
岳 西川
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NEC Corp
Original Assignee
NEC Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by NEC Corp filed Critical NEC Corp
Priority to JP62030916A priority Critical patent/JPH06103495B2/en
Publication of JPS63197273A publication Critical patent/JPS63197273A/en
Publication of JPH06103495B2 publication Critical patent/JPH06103495B2/en
Anticipated expiration legal-status Critical
Expired - Lifetime legal-status Critical Current

Links

Classifications

    • GPHYSICS
    • G06COMPUTING OR CALCULATING; COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F15/00Digital computers in general; Data processing equipment in general
    • G06F15/76Architectures of general purpose stored program computers
    • G06F15/80Architectures of general purpose stored program computers comprising an array of processing units with common control, e.g. single instruction multiple data processors
    • G06F15/8053Vector processors

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Theoretical Computer Science (AREA)
  • Computing Systems (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Complex Calculations (AREA)

Description

【発明の詳細な説明】 〔産業上の利用分野〕 本発明は、任意のタイミングで読み書き可能な複数のベ
クトルレジスタと、該ベクトルレジスタからのデータを
受け演算し、演算結果を命令で指定された前記ベクトル
レジスタに順次書込むベクトル演算器およびデータ転送
パスとを有し、命令で指定されたベクトルレジスタの任
意のベクトル要素から演算を開始することができる機能
を備えた情報処理装置に関する。
DETAILED DESCRIPTION OF THE INVENTION [Industrial field of application] The present invention specifies a plurality of vector registers that can be read and written at arbitrary timings, receives data from the vector registers, performs an operation, and specifies an operation result by an instruction. The present invention relates to an information processing apparatus having a vector arithmetic unit for sequentially writing to the vector register and a data transfer path, and having a function of starting an arithmetic operation from an arbitrary vector element of a vector register designated by an instruction.

〔従来の技術〕[Conventional technology]

従来、この種の情報処理装置では、命令で指定された演
算開始要素から順次ベクトル長レジスタで指定される要
素数だけ演算を行なう場合で、先行する命令のソースレ
ジスタ(被演算データ格納レジスタ)と後続する命令の
デスティネーションレジスタ(演算結果格納レジスタ)
とが一致し、かつ先行命令が演算に必要なデータを読出
す領域と後続命令が演算結果を書込む領域が重なる場
合、後続命令を先行命令と並行して処理すると、先行命
令が演算に必要なデータを読出す前に後続命令が該領域
に演算結果を書込み演算結果の因果関係を逆転する可能
性があるため、命令実行の順序性を保証するため、最小
限第3図(c)で示すように、後続命令の演算結果の書
込みが先行命令の演算データの読出しより先行しないこ
とを保証すれば充分であるが、第3図(b)で示すよう
に先行命令が完全に終了するのを待って対応する後続命
令を開始するよう制御していた。
Conventionally, in this type of information processing apparatus, when the operation is performed by the number of elements sequentially specified by the vector length register from the operation start element specified by the instruction, the source register of the preceding instruction (operated data storage register) Destination register of the instruction that follows (operation result storage register)
And the area where the preceding instruction reads the data required for the operation and the area where the subsequent instruction writes the operation result overlap, the subsequent instruction is processed in parallel with the preceding instruction, and the preceding instruction is required for the operation. Since the subsequent instruction may write the operation result to the area before reading the appropriate data and reverse the causal relation of the operation result, in order to guarantee the order of instruction execution, at least FIG. As shown, it is sufficient to guarantee that the writing of the operation result of the succeeding instruction does not precede the reading of the operation data of the preceding instruction, but as shown in FIG. 3B, the preceding instruction is completely completed. It was controlled to wait for and start the corresponding subsequent instruction.

〔発明が解決しようとする問題点〕[Problems to be solved by the invention]

上述した従来の情報処理装置は、先行命令のソースレジ
スタと後続命令のデスティネーションレジスタとが一致
し、かつ先行命令が演算に必要なデータを読出す領域と
後続命令が演算結果を書込む領域が重なる場合、先行命
令が完全に終了するのを待って対応する後続命令を開始
するように制御しているため、無駄な待ち合わせ時間が
増加し、ベクトル処理効率が著しく低下するという欠点
がある。
In the above-described conventional information processing apparatus, the source register of the preceding instruction and the destination register of the succeeding instruction match, and the area in which the preceding instruction reads the data necessary for the operation and the area in which the succeeding instruction writes the operation result are When they overlap, the control is performed so that the corresponding subsequent instruction is started after waiting for the preceding instruction to be completely completed, which causes a disadvantage that the waiting time is wasted and the vector processing efficiency is significantly reduced.

〔問題点を解決するための手段〕[Means for solving problems]

本発明の情報処理装置は、任意のタイミングで読み書き
される複数のベクトルレジスタと、該ベクトルレジスタ
からのデータを受け演算し、その演算結果を命令で指定
された前記ベクトルレジスタに順次書込むベクトル演算
器およびデータ転送パスとを有し、命令で指定されたベ
クトルレジスタの任意のベクトル要素から演算を開始す
る機能を備えた情報処置装置であって、 先行命令の被演算データを読み出すベクトルレジスタと
後続命令の演算結果を書き込むベクトルレジスタとが同
じである場合において、前記後続命令による前記ベクト
ルレジスタへの書込み要素位置を取得する書込位置取得
手段と、前記後続命令の演算に要する時間を取得する演
算時間取得手段と、前記書込位置取得手段により取得さ
れた書込み要素位置から前記演算時間取得手段により取
得された時間を減算する減算手段と、前記先行命令によ
る前記ベクトルレジスタの読出し開始要素位置を取得す
る読出し位置取得手段と、この読出し位置取得手段によ
り取得された読出し開始要素位置と前記減算手段による
減算結果とを比較する比較手段を含み、この比較手段の
結果に応じて前記後続命令の起動を制御する制御部を有
している。
The information processing apparatus of the present invention is a vector operation in which a plurality of vector registers read and written at arbitrary timings and data from the vector registers are operated and operated, and the operation results are sequentially written into the vector register designated by an instruction. And a data transfer path, and an information processing device having a function of starting an operation from an arbitrary vector element of a vector register designated by an instruction, wherein a vector register for reading out operated data of a preceding instruction and a subsequent When the vector register for writing the operation result of the instruction is the same, the write position acquisition means for acquiring the write element position to the vector register by the subsequent instruction, and the operation for acquiring the time required for the operation of the subsequent instruction The calculation from the time acquisition means and the write element position acquired by the write position acquisition means Subtraction means for subtracting the time acquired by the time acquisition means, read position acquisition means for acquiring the read start element position of the vector register by the preceding instruction, and read start element position acquired by the read position acquisition means The control unit includes a comparison unit that compares the subtraction result of the subtraction unit, and controls activation of the subsequent instruction according to the result of the comparison unit.

〔作用〕 したがって、ベクトル命令起動時の無駄な待ち合わせが
減少し、ベクトル処理効率が向上する。
[Operation] Therefore, useless waiting at the time of activating a vector instruction is reduced, and the vector processing efficiency is improved.

〔実施例〕〔Example〕

次に、本発明の実施例について図面を参照して説明す
る。
Next, embodiments of the present invention will be described with reference to the drawings.

第1図は本発明の情報処理装置の一実施例の要部を示す
ブロック図、第2図は制御部4のうち命令起動に関する
部分のブロック図、第3図(a)は本実施例の効果を示
すタイムチャート図である。
FIG. 1 is a block diagram showing a main part of an embodiment of an information processing apparatus of the present invention, FIG. 2 is a block diagram of a part related to instruction activation in a control unit 4, and FIG. 3 (a) is a block diagram of this embodiment. It is a time chart figure which shows an effect.

本実施例の情報処理装置は、任意のタイミングで読み書
き可能な複数のベクトルレジスタV0,V1,V2,…,Vnと、こ
れらベクトルレジスタV0〜Vnからのデータを受け、演算
し、演算結果を命令で指定されたベクトルレジスタに順
次書込むベクトル演算器2a,2b,2c,…,2dおよびデータ転
送パス1と、制御部4とを有している。
The information processing apparatus according to the present embodiment receives a plurality of vector registers V0, V1, V2, ..., Vn that can be read and written at arbitrary timings, receives data from these vector registers V0 to Vn, performs an operation, and issues an operation result instruction. , 2d for sequentially writing to the vector register designated by 1 and 2, the data transfer path 1, and the control unit 4.

制御部4は、起動をかけようとしている命令を保持する
命令保持レジスタ41と、命令保持レジスタ41に保持され
ている命令情報からその命令のFunction Unit Time(FU
T:被演算データを読出してから最初の演算結果をレジス
タに書込むまでの時間)を解読するデコーダ42と、命令
保持レジスタ41に保持されている命令情報のうち、該命
令の演算結果の格納開始点を示す情報n2と、デコーダ42
の出力fut2との差を算出する減算器43と、先行命令の被
演算データ読出し要素番号を指示するレジスタ44と、該
被演算データ読出し要素番号が減算器43の出力n2−fut2
を越えたことをを検出する比較器45と、比較器45からの
検出信号を受けて演算器2a〜2mの使用状況等の情報から
命令保持レジスタ41に保持されている命令の起動の可否
を判断し、起動可の場合命令起動信号を出力する命令起
動制御部46から構成されている。
The control unit 4 determines the function unit time (FU) of the instruction from the instruction holding register 41 that holds the instruction that is about to be activated and the instruction information that is held in the instruction holding register 41.
T: Decoder 42 for decoding the time from reading the data to be operated to writing the first operation result in the register), and storing the operation result of the instruction among the instruction information held in the instruction holding register 41 Information n 2 indicating the starting point and the decoder 42
Of the output fut 2 of the subtracter 43, the register 44 for instructing the operand data read element number of the preceding instruction, and the operand data read element number is the output n 2 -fut 2 of the subtractor 43.
Comparator 45 for detecting that the instruction has been exceeded and whether or not the instruction held in the instruction holding register 41 can be activated from the information such as the usage status of the arithmetic units 2a to 2m in response to the detection signal from the comparator 45. The instruction activation control unit 46 is configured to determine and activate the instruction activation signal when activation is possible.

次に、本実施例の動作について説明する。Next, the operation of this embodiment will be described.

第3図に示すように (ベクトルレジスタV0とV1とを第n1要素から順次演算
し、ベクトルレジスタV2へ格納する)と、 (ベクトルレジスタV3とV4とを演算しベクトルレジスタ
V0へ第n2要素から順次格納する)という連続する命令列
を実行する場合、先行する命令のベクトルレジスタV0に
対する読出し開始要素n1が読出し要素番号保持レジスタ
44にセットされ、演算の進行に同期してカウントアップ
される。一方、後続の命令はレジスタ41にセットされ、
該命令の演算結果格納開始要素n2および該命令の演算の
FUT=fut2が解読され、減算器43でn2−fut2が得られ、
比較器45において先行命令の被演算データの読出し要素
番号と比較される。先行命令が開始された時点では該先
行命令の被演算読出し要素番号<n2−fut2の関係、すな
わち後続の演算結果が書込まれる領域のデータを先行命
令が未だ読出していない状態であり、従って後続の命令
は起動されない。次に先行命令の実行が進み被演算デー
タの読出し要素番号>n2−fut2の関係になると命令起動
制御部46において後続命令に対し実行起動がかけられ
る。この時刻よりfut2時間後ベクトルレジスタV0の第n2
要素に対し、先行命令のデータ読出しと後続命令による
演算結果書込みが同時に実施され、以降n2+1,n2+2,…
と全く無駄がなく、かつ命令の順序性が保存された状態
で命令が実行されていく。
As shown in FIG. (The vector registers V0 and V1 are sequentially calculated from the n1th element and stored in the vector register V2), (Vector register V3 and V4 are calculated and vector register
Sequentially store from the n2th element to V0), the read start element n1 for the vector register V0 of the preceding instruction is the read element number holding register
It is set to 44 and is counted up in synchronization with the progress of calculation. On the other hand, the subsequent instruction is set in the register 41,
The operation result storage start element n2 of the instruction and the operation result of the instruction
FUT = fut 2 is decoded, subtracter 43 obtains n 2 −fut 2 ,
The comparator 45 compares the read element number of the operand data of the preceding instruction. Preceding instruction operand read element number <of n2-fut 2 relationship said prior instruction at the time it was initiated, i.e. a state subsequent operation result preceding the data area to be written instruction is not yet read, thus Subsequent instructions will not be activated. Run start to subsequent instruction preceding execution of instructions proceeds becomes read element number> of n2-fut 2 relationship operand data and the instruction start control unit 46 then is applied. Fut 2 hours after this time n2 of vector register V0
For the element, the data reading of the preceding instruction and the writing of the operation result by the succeeding instruction are performed at the same time, and thereafter n2 + 1, n2 + 2, ...
The instructions are executed with no waste and the order of the instructions is preserved.

〔発明の効果〕〔The invention's effect〕

以上説明したように本発明は、先行命令が被演算データ
を読出すベクトルレジスタと、後続命令が演算結果を書
込むベクトルレジスタが同じ場合、先行命令による被演
算データの読出しが、未実行の該後続命令の演算結果格
納開始点を通過したことを検出し、検出した時点で該後
続命令の演算結果を開始するよう該命令の起動を制御す
ることにより、ベクトル命令起動時の無駄な待ち合わせ
を減少させ、効率よいベクトル処理を実現できるという
効果がある。
As described above, in the present invention, when the vector register from which the preceding instruction reads the operated data and the vector register in which the succeeding instruction writes the operation result are the same, the reading of the operated data by the preceding instruction is not executed yet. By detecting that the operation result storage start point of the subsequent instruction has been passed and controlling the start of the instruction so that the operation result of the subsequent instruction is started at the time of detection, unnecessary waiting at the time of starting the vector instruction is reduced. Therefore, there is an effect that efficient vector processing can be realized.

【図面の簡単な説明】[Brief description of drawings]

第1図は本発明の情報処理装置の一実施例の要部を示す
ブロック図、第2図は制御部4のうち命令起動に関する
部分のブロック図、第3図は本実施例の効果を従来と比
較して示すタイムチャートである。 V0.V1,…,Vn……ベクトルレジスタ、 2a,2b,…,2m……ベクトル演算器、 3……データ転送パス、 4……制御部、 41……命令保持レジスタ、 42……デコーダ、 43……減算器、 44……レジスタ、 45……比較器、 46……命令起動制御部。
FIG. 1 is a block diagram showing a main part of an embodiment of an information processing apparatus of the present invention, FIG. 2 is a block diagram of a part relating to instruction activation in a control unit 4, and FIG. It is a time chart shown in comparison with. V0.V1, ..., Vn ... Vector register, 2a, 2b, ..., 2m ... Vector operation unit, 3 ... Data transfer path, 4 ... Control unit, 41 ... Instruction holding register, 42 ... Decoder, 43 …… subtractor, 44 …… register, 45 …… comparator, 46 …… instruction activation control section.

Claims (1)

【特許請求の範囲】[Claims] 【請求項1】任意のタイミングで読み書きされる複数の
ベクトルレジスタと、該ベクトルレジスタからのデータ
を受け演算し、その演算結果を命令で指定された前記ベ
クトルレジスタに順次書込むベクトル演算器およびデー
タ転送パスとを有し、命令で指定されたベクトルレジス
タの任意のベクトル要素から演算を開始する機能を備え
た情報処置装置において、 先行命令の被演算データを読み出すベクトルレジスタと
後続命令の演算結果を書き込むベクトルレジスタとが同
じである場合において、前記後続命令による前記ベクト
ルレジスタへの書込み要素位置を取得する書込位置取得
手段と、前記後続命令の演算に要する時間を取得する演
算時間取得手段と、前記書込位置取得手段により取得さ
れた書込み要素位置から前記演算時間取得手段により取
得された時間を減算する減算手段と、前記先行命令によ
る前記ベクトルレジスタの読出し開始要素位置を取得す
る読出し位置取得手段と、この読出し位置取得手段によ
り取得された読出し開始要素位置と前記減算手段による
減算結果とを比較する比較手段を含み、この比較手段の
結果に応じて前記後続命令の起動を制御する制御部を有
することを特徴とする情報処理装置。
1. A plurality of vector registers that are read and written at arbitrary timings, a vector arithmetic unit that receives data from the vector registers, performs an arithmetic operation, and sequentially writes the arithmetic result to the vector register designated by an instruction, and data. In an information processing device having a transfer path and having a function of starting an operation from an arbitrary vector element of a vector register designated by an instruction, a vector register for reading out operated data of a preceding instruction and an operation result of a succeeding instruction are displayed. When the vector register to be written is the same, a write position acquisition means for acquiring a write element position to the vector register by the subsequent instruction, and an operation time acquisition means for acquiring a time required for the operation of the subsequent instruction, From the write element position acquired by the write position acquisition means, the calculation time acquisition means Subtraction means for subtracting the time acquired by the read instruction, read position acquisition means for acquiring the read start element position of the vector register by the preceding instruction, read start element position acquired by the read position acquisition means, and the subtraction means An information processing apparatus, comprising: a comparison unit that compares the subtraction result according to step 1 above, and a control unit that controls the activation of the subsequent instruction according to the result of the comparison unit.
JP62030916A 1987-02-12 1987-02-12 Information processing equipment Expired - Lifetime JPH06103495B2 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
JP62030916A JPH06103495B2 (en) 1987-02-12 1987-02-12 Information processing equipment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
JP62030916A JPH06103495B2 (en) 1987-02-12 1987-02-12 Information processing equipment

Publications (2)

Publication Number Publication Date
JPS63197273A JPS63197273A (en) 1988-08-16
JPH06103495B2 true JPH06103495B2 (en) 1994-12-14

Family

ID=12317025

Family Applications (1)

Application Number Title Priority Date Filing Date
JP62030916A Expired - Lifetime JPH06103495B2 (en) 1987-02-12 1987-02-12 Information processing equipment

Country Status (1)

Country Link
JP (1) JPH06103495B2 (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US9021233B2 (en) * 2011-09-28 2015-04-28 Arm Limited Interleaving data accesses issued in response to vector access instructions

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS57206982A (en) * 1981-06-15 1982-12-18 Fujitsu Ltd Instruction controlling system
JPS58134365A (en) * 1982-02-03 1983-08-10 Hitachi Ltd Vector processor

Also Published As

Publication number Publication date
JPS63197273A (en) 1988-08-16

Similar Documents

Publication Publication Date Title
JPH06139189A (en) Common bus arbitrating mechanism
JP2531760B2 (en) Vector processor
JPH06103495B2 (en) Information processing equipment
JPS5890247A (en) Pipeline controlling system of information processor
JPH0877143A (en) Vector data processor
JPH07244588A (en) Data processing device
JP3084953B2 (en) Information processing device
JP2985244B2 (en) Information processing device
JPS60250438A (en) Information processor
JP2001005689A (en) Trace sampling circuit
JPH06149569A (en) Register number changing device
JPH0675789A (en) Information processor
JP2584528Y2 (en) Information processing device
JP2806690B2 (en) Microprocessor
JPS6134188B2 (en)
JP2693930B2 (en) Vector processing equipment
JPH07210384A (en) Floating point arithmetic unit
JPH01260541A (en) Register saving method
JPH077386B2 (en) Information processing equipment
JPH0424731B2 (en)
JPS61133440A (en) Data processing unit
JPH08286912A (en) Program controller
JPH05303493A (en) Method for switching instruction train and arithmetic processor using the method
JPS61249140A (en) Microprogram controller
JPS63174173A (en) Information processor