1300903 九、發明說明： 【發明所屬之技術領域】 本發明涉及一種對中央處理器（CPU，Central Processing Unit)之電力 消耗進行估算之方法，尤其涉及到一種運用仿真技術估算CPU功率之方法。 【先前技術】1300903 IX. Description of the Invention: [Technical Field] The present invention relates to a method for estimating power consumption of a Central Processing Unit (CPU), and more particularly to a method for estimating CPU power using simulation techniques. [Prior Art]
近年來，由於電腦技術快速發展，電子設備、特別係個人電腦之小型 化及便攜化正不斷改進。在這種趨勢下，如何在提高中央處理器之處理速 度的同時，降低電力消耗成爲一項重要研究課題。在實現低消耗電力化之 過程中，正確快速地估算中央處理器之功率消耗顯得尤爲重要。傳統的功 率測試方法需要數量較多之元件組合，如電阻、電感器等元件，所需測量 數據較多，並且需利用幾個公式進行數據換算，整個測試過程也較爲繁瑣， 尤其涉及到對幾個不同類型之cpu進行功率測試時，每個cpu之測試是各 自獨立的，如此便需更換CPU並反覆測量其相應之電壓、電流、電阻等相 關參數。此外，英代爾公司（Intel)新推出之LGA775平臺（爲一種新的 cpu與主機板之連接規範，介面採用775個觸點，而傳統連接採用478個 ，腳）’其散熱片之覆蓋區域比較大，用傳統的測試方法亦較難進行功率測 爲了提高測試效率及功率估算之準雜，減輕測試人員之 、種二央?T1之功率估异方法’其協助測試人員運用合‘的插座 術絲㈣間崎算出獨麵之巾錢理器之消 耗功率，測夏所需70件少，並且測量步驟簡單。 【發明内容】 供電源；開 包括St較=施=式提供—種中央處理器之功率估算方法，該方法 ι括如下4•接通-電源’其藉由一電源連接器爲主 板中’該中央處理器治具與一電子負載機相連；調整 ，率^表:該姑括-線性内插法公式；安裝中央知雜 電子負載機之輸出電 流^多次職各錄，包括乡次·魏連接魏之輸人電 ^中^處理器治具中之輸出電壓；換算功率轉換效率 換^電^ 出功率與輸人功率之比值；安裝中央處理器至主機板；執行 5 1300903 測里各參數’包括測量電源連接器之輪入電壓、輸入電流及中央處 3核分析換算出中央處理器之功率轉換效率，依據執行該測試程 並測ϊ出之輸入電流與調整電子負載機所多次測量出之輪入電流進行比 =2㈣子貞載機所獲取之複數不鳴人電流帽取與執行測試程式 之輸入電流相鄰近之上下兩個輸入電流，將選取之兩個輸入電流 之功率轉換效率輸人至功率估算表中’換算出中央處理器之 轉換效率；估算出中央處理器之消耗功率及其消耗電流。 相較習知技術，所述中央處理器之功率估算方法 ;:==r(— 【實施方式] 爲便於理解，對下文涉及之術語做簡要說明： ❹ΐϊ啸法··假制目6知數據之變化躲性_，因此可由已知兩 去計算透過這二點之斜線函數，公式如下：（其 為⑥要拉之缝聽值，b點即代表_點，且沾《） ⑽-⑽。 =pG赠程式··為_測試程式，其可_ cpu中近乎全部之單元， 杯你一二之一百之貧源佔用與功率消耗，可令中央處理器子系統中存在之 何一點不穩定因素在短__重新魏⑽之方式表現出來。 數之抑女震CPU治具及該治具與一電子負載機相連時，各個測量參 CPU治具處接=測量之輸人 L :代表 、、六· Γ 、 矜出電壓，Cinl ·代表電源連接器處所測量之輸入電 二功=3=之輸_;PWE1:代表功率轉換效率，其爲 接器臟之代細^ :代表電源連 ^電垒，Vcore2·代表CPU治具處所測量之内核電壓；Cin2·· 代表電源轉H處所繼之輸入電流。 其餘參數之代表涵義·· PWE ••代表CPU之功率轉換效率· P ··代表估 1300903 算得所耗功=in:代表估算得出之咖消耗電流。 之主機巾域㈣之辨料枝錄實施方式 、、㈣m 口測虿結構圖。該CPU治具10爲仿直CPU12進行 ^ ° « 2 ^ ^ 3« 人員可電叙調整和賴雜轉數之職。測試 表、電壓B表等）負I機4之輸出電流c°utl，並利用測量工具（如安培In recent years, due to the rapid development of computer technology, the miniaturization and portability of electronic devices, especially personal computers, are constantly improving. Under this trend, how to improve the processing speed of the central processing unit while reducing power consumption has become an important research topic. It is especially important to estimate the power consumption of the central processor correctly and quickly in the process of achieving low power consumption. Traditional power test methods require a large number of component combinations, such as resistors, inductors, etc., which require more measurement data, and need to use several formulas for data conversion. The whole test process is also cumbersome, especially related to When several different types of CPUs are used for power testing, each cpu test is independent, so the CPU needs to be replaced and its corresponding voltage, current, resistance and other related parameters are measured repeatedly. In addition, Intel's new LGA775 platform (for a new cpu and motherboard connection specification, the interface uses 775 contacts, while the traditional connection uses 478, feet) 'the coverage area of its heat sink Larger, it is more difficult to carry out power measurement with the traditional test method. In order to improve the test efficiency and power estimation, the tester's power metering method is reduced, and the tester uses the socket. The silk (4) Kawasaki calculates the power consumption of the one-sided towel money processor, 70 less for summer measurement, and the measurement procedure is simple. [Invention] The power supply; the method includes a power supply estimation method of the central processing unit, and the method includes the following: 4. The power-on-power supply is provided in the main board by the power connector. The central processor fixture is connected to an electronic load machine; adjustment, rate ^ table: the formula of the guillo-linear interpolation method; the output current of the central knowing miscellaneous electronic load machine ^ multiple records, including the township Wei Connect the output voltage of Weizhi's input power ^^^ processor fixture; convert the power conversion efficiency to the ratio of the power output to the input power; install the central processor to the motherboard; execute 5 1300903 'Including the measurement of the power supply connector's turn-in voltage, input current and the central core 3-core analysis to convert the power conversion efficiency of the central processor, according to the test process and the measured input current and the adjustment of the electronic load machine multiple measurements The round-in current ratio is calculated as the ratio of the two input currents of the two input currents that are obtained by the input current of the tester. The rate conversion efficiency is input to the power estimation table to convert the conversion efficiency of the central processing unit; the power consumption of the central processing unit and its current consumption are estimated. Compared with the prior art, the power estimation method of the central processing unit::==r (- [Embodiment] For ease of understanding, a brief description of the terms involved below is briefly described: ❹ΐϊ 法法··假目目6 know data The change is hiding _, so the slash function through these two points can be calculated by the known two, and the formula is as follows: (It is the value of the slit to be pulled, the point b represents the _ point, and the stain is ") (10) - (10). =pG free program ·· is a test program, which can be almost all the units in the cpu, cups one hundred and one hundred of the poor source and power consumption, can make the central processor subsystem what is unstable The factor is expressed in the form of short __re-wei (10). When the number of the female shock CPU fixture and the fixture are connected with an electronic loader, each measurement is connected to the CPU fixture = the input of the measurement L: representative, , · Γ , 矜 output voltage, Cinl · represents the input power measured at the power connector where the two power = 3 = the loss _; PWE1: represents the power conversion efficiency, which is the dirty generation of the connector ^: represents the power supply ^ The electric barrier, Vcore2· represents the core voltage measured by the CPU fixture; Cin2·· represents the power supply to H The input current is followed by the meaning of the remaining parameters. · PWE •• represents the power conversion efficiency of the CPU · P ··Representative estimate 1300903 Calculated power consumption =in: represents the estimated coffee consumption current. The implementation method of the identification and the (4) m-port measurement structure diagram. The CPU fixture 10 is a straight-forward CPU12 for ^ ° « 2 ^ ^ 3« personnel can be adjusted and the number of rotations. Voltage B table, etc.) Negative I machine 4 output current c °utl, and use measurement tools (such as amps
治具10 ΐί 5 v連接f 3之輸入電壓Vinl、輸入電流&及CPU 参舰ZL a/ _，可參照如第四圖所示之各類數據，其中該電子 負載機4 =輸出電流C〇utl可在〇安培至11〇安培之間進行調整。 之主示’係本發明中央處理器之功率估算方法較佳實施方式 供測量結構圖。電源2藉由電源連接器3爲主機板1提 :、伏特之輸入電源，主機板！中安裝有cpui2。測試人員可在安装。則2 後^仃MaxPower測試程式，並測量出電源連接器3之輸入電壓4、輸 入電流C，及mm之内核電壓ν_2，測量數據可參照第五圖所示。 參閱第三圖所示，係本發明中央處理器之功率估算方法較佳實施方式 之具體實麟糊。首先，轉估絲，絲可爲Exeel格式，並且該 f已經添加線性_法公式，只紐侧數據填人表帽細位，即可換 算出所需數據（步驟S1G)。安裝CPU治具1Q至主機板i之cpu插座14 中（步驟si2)。連接電子負载機4與CPU治具1〇，連接關係可參照第一 圖所示（步驟S14)。藉由調整電子負載機4之輸出電流c—，測量不同輸 出電流Coutl下各個參數對應之數據，包括電源連接器3之輸入電壓v时、 輸入電流cinla CPU治具10之輸出電壓Vc〇rel，此測量隨著輸出電流c。⑽ 的調整多次進行，該輸出電流coutl之區間爲0安培至110安培，測量數據 可參照第四圖所示（步驟S16)。根據步驟S16之各個測量值換算功率轉換 效率PWEi，該功率轉換效率PWEi爲輸出功率與輸入功率之比值，而輸出 功率爲CPU治具1〇處所測量之輸出電壓與電子負載機之輸出電流 1300903 ，之乘積’輸人功率爲電源連接113處所測量之輸人雜vinl與輸入電 - t乘積（步驟Sl8)。除去CPU治具10及電子負載機4，安裝CPU12 少一 1，其連接關係可參照第二圖所示（步驟S20)。安裝CPU後，執 $ x^ower測式（步驟S22)。測量電源連接器3之輸入電壓1、 :入電流‘及CPU12之内核電壓ν_2，測量數據可參照第五圖所示（步 丰^24制里值進行分析換算以得到CPU之功率轉換效率PWE，依據 ζ:=所測里出之電源連接器3之輸入電流Qn2與步驟S16所測量之電 f連接ϋ 3之輸入電流Qni進行比對，選取與輸人電流&柳近之輸入電 中之兩健值’如根據第五圖中9安培之輸人電流Cin2選取第四圖中 • 分職咖安培及9·458安培之輸人電流。，其所對應之功 0、"^ςΡλ¥Εΐ分別爲8275%及81·36%，將所選擇之測量值輸入CPU 功率估算表中，並藉由線性内插法公式求解出cpu之功率轉換效率pw£ (步驟S26)。依據上一步驟得出之㈣功率轉換效率估算⑽之消 $功率P，其中，消耗功率估算值P爲功率轉換效率pWE與步驟s24所測 ，出之輸入電壓Vw及輸入電流&之乘積，可參照第六圖所示公式及麟 (步驟S28 )。 若需要測量其他不同類型CPU之消耗功率，只需將cpu進行替換，缺 後從步驟S22開始執行，將步驟似所獲取之新數據與前一 cpu測量過程 蠢步驟S1^所獲取之數據進行比對分析’並藉由線性内插法公式求解出 _ 替換後cpu消耗功率之估算值，而無需再重複步驟⑽至伽。 所示，係本發财央處理器之功率估算方法較佳實施方式 之域板安裝CPU治具之·雜利_，其中各雜係在第—圖之測量 結構下運用第三圖之實施流程所獲取。例如當調整電子負載機4之輸出電 流c0utl爲10安培時，可測量得出電源連接器3之輸人電虔v如爲㈣伏 特、輸入電流Cinl爲U98安培及CPU治具1〇之輸出電璧爲i 4〇4 伏特，則該功率轉換效率PWEl爲輸出功率與輸入功率之 即 PWEflOO%* / (Vinl*Cinl) =64.69%。 參閱第五圖所示，係本發财央處理器之功率估算方法較佳實施方式 之主機板安裝CPU之測量數據示例圖，其中各數據係在第二圖之測量結構 8Fixture 10 ΐί 5 v Connect f 3 input voltage Vinl, input current & and CPU to ship ZL a / _, refer to the various types of data as shown in the fourth figure, where the electronic loader 4 = output current C 〇utl can be adjusted from ampere to 11 amps. The preferred embodiment of the power estimation method of the central processing unit of the present invention is for measuring the structure. The power supply 2 is provided for the motherboard 1 by the power connector 3: the input power of the volt, the motherboard! There is cpui2 installed in it. Testers can be installed. Then, after the MaxPower test program, and measure the input voltage 4 of the power connector 3, the input current C, and the core voltage ν_2 of mm, the measurement data can be referred to the fifth figure. Referring to the third figure, it is a specific embodiment of the power estimation method of the central processing unit of the present invention. First, the silk is estimated, the wire can be in the Exeel format, and the linear _ method formula has been added to the f, and only the data on the side of the button is filled in the fine position of the cap, and the required data can be calculated (step S1G). Install the CPU fixture 1Q into the cpu socket 14 of the motherboard i (step si2). The electronic loader 4 and the CPU fixture are connected, and the connection relationship can be referred to as shown in the first figure (step S14). By adjusting the output current c_ of the electronic loader 4, the data corresponding to each parameter under different output currents Cout1 is measured, including the input voltage v of the power connector 3, and the output voltage Vc〇rel of the input current cinla CPU fixture 10, This measurement follows the output current c. The adjustment of (10) is performed a plurality of times, and the interval of the output current coutl is 0 amps to 110 amps, and the measurement data can be referred to the fourth figure (step S16). Converting the power conversion efficiency PWEi according to each measured value of step S16, the power conversion efficiency PWEi is the ratio of the output power to the input power, and the output power is the output voltage measured by the CPU fixture 1〇 and the output current of the electronic loader 1300903, The product 'input power' is the product of the input vin1 and the input electric-t measured at the power connection 113 (step S18). When the CPU jig 10 and the electronic loader 4 are removed, the number of mounting CPUs 12 is one less, and the connection relationship can be referred to as shown in the second figure (step S20). After the CPU is installed, the $x^ower test is executed (step S22). Measure the input voltage of the power connector 3: the current into the current and the core voltage ν_2 of the CPU12. The measurement data can be analyzed and converted according to the value shown in the fifth figure (the value of the step is 24) to obtain the power conversion efficiency PWE of the CPU. According to ζ:= the input current Qn2 of the power connector 3 measured in the comparison is compared with the input current Qni of the electric f connection ϋ3 measured in step S16, and is selected in the input current of the input current & The two health values are selected in the fourth picture according to the input current Cin2 of 9 amps in the fifth figure. • The power of the amps and the input current of 9.458 amps. The corresponding work 0, "^ςΡλ¥ Εΐ 8275% and 81.36%, respectively, the selected measurement value is input into the CPU power estimation table, and the power conversion efficiency pw£ of the CPU is solved by the linear interpolation formula (step S26). According to the previous step (4) The power conversion efficiency estimation (10) is eliminated by the power P, wherein the power consumption estimated value P is the product of the power conversion efficiency pWE and the input voltage Vw and the input current & The formula shown in the figure and the lin (step S28). If you need to measure other The power consumption of different types of CPUs is only required to replace the CPU. After the failure, the process starts from step S22, and the new data obtained by the steps is compared with the data obtained by the stupid step S1^ of the previous cpu measurement process. The linear interpolation method formula is used to solve the estimated value of the CPU power consumption after the replacement, without repeating the step (10) to the gamma. As shown in the figure, the power board estimation method of the present invention is the domain plate installation of the preferred embodiment. CPU fixtures and miscellaneous _, each of which is obtained by the implementation process of the third figure under the measurement structure of the first figure. For example, when the output current c0utl of the electronic loader 4 is adjusted to 10 amps, it can be measured. If the input power of the power connector 3 is (four) volts, the input current Cinl is U98 ampere, and the output of the CPU fixture is 〇4 〇 4 volts, the power conversion efficiency PWEl is the output power and input. The power is PWEflOO%* / (Vinl*Cinl) = 64.69%. Referring to the fifth figure, it is an example of the measurement data of the motherboard mounting CPU of the preferred embodiment of the power estimation method of the present invention. Each data is in the second Diagram measurement structure 8
第一圖 流程圖 示例圖 1300903 下運用第三圖之實施流程所獲取。即主機板安裝CPU後，執行馳而 測試程式，測量出魏連接器3之輸入電壓U 1173伏特、輸入電流 ‘爲9安培及CPU12之内核電壓Vc〇re2爲L4伏特。 參閱第六圖所示，係本發明中央處理器之功率估算方法較佳實施方式 之功率估算示欄，其巾各紐據求麟在cpu功率縛騎進行，該表 可爲Excel表格，表中包括線性内插法公式，將第三圖步驟§26所選擇之數 據輸入表中相細位，先換算出CPU《功率轉換效率PW£，再換算出㈣ 消耗功率之估算值P，並可依據齡輕!>鮮丨cpu _電$二估算值 Cin。 ^ 、曰本實財式巾出現之各健據僅爲—組參考值，實際應財還可根據 測量需要，將電子貞錢4之輸出電流之變化範_整至更小續取更爲 精確之數據。 ” 本發明雖啸佳實施_露如上，然其並_嫌林㈣。任何熟 悉此項技藝者，在不脫離本發明之精神和範#可做^與潤飾，?因 此本發明之保護範圍當視後附之申請專利範圍所界定者為準。〃/ 【圖式簡單說明】 裝CPU i係本發明/央㈤處理器之功率估算方法較佳實施方式之主機板安 第二圖係本發明中央處理器之功率估算方法較佳實施方 裝CPU之測量結_。 K主機板女 第三圖係本發明中央處理器之功率估算方法較佳實施方式之具體實施 第四圖係本發明巾央處㈣之神轉方法較佳實施 — 裝CPU治具之測量數據示例目。 W王機扳女 第五圖係本發明中央處理器之功率估算方法較佳實施方 裝CPU之測量輯示例圖。 第六圖係本發明中央處理n之神估算方法較佳實財式之功率估曾 9The first diagram Flowchart Example diagram 1300903 is obtained using the implementation flow of the third diagram. After the CPU is installed on the motherboard, the test program is executed to measure the input voltage U 1173 volts of the Wei connector 3, the input current ‘9 amps, and the core voltage Vc〇re2 of the CPU 12 is L4 volts. Referring to the sixth figure, it is a power estimation column of a preferred embodiment of the power estimation method of the central processing unit of the present invention, and the towel is calculated according to the CPU power, and the table can be an Excel table. Including the linear interpolation formula, the data selected in the §26 step of the third figure is entered into the fine-order phase of the table, and the CPU "power conversion efficiency PW£ is first converted, and then the estimated value of the power consumption is calculated (P), and can be based on Ageing!> Fresh cpu _ electricity $ two estimates Cin. ^, 曰 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实 实Data. Although the invention is implemented in the above-mentioned _ _ _ _ _ _ _ _ _ _ _ _ (4). Anyone familiar with the art, without departing from the spirit and scope of the present invention can be done and retouched, so the scope of protection of the present invention The scope of the patent application is defined as follows. 〃 / [Simplified description of the drawing] CPU i is the power estimation method of the present invention / central (five) processor The preferred embodiment of the motherboard is the second picture of the center of the present invention The power estimation method of the processor is preferably implemented by the CPU. The K-board is the third embodiment of the present invention. The fourth embodiment of the present invention is a fourth embodiment of the present invention. (4) The preferred method of transforming the gods is to carry out the method of measuring the data of the CPU fixture. The fifth figure of the W machine is the example of the power estimation method of the central processor of the present invention. The six figures are the method for estimating the central processing of the central processing n of the present invention.