|Publication number||WO2008025682 A1|
|Publication date||6 Mar 2008|
|Filing date||16 Aug 2007|
|Priority date||1 Sep 2006|
|Also published as||DE602007007491D1, EP2062148A1, EP2062148B1, US7882370, US20080069558|
|Publication number||PCT/2007/58523, PCT/EP/2007/058523, PCT/EP/2007/58523, PCT/EP/7/058523, PCT/EP/7/58523, PCT/EP2007/058523, PCT/EP2007/58523, PCT/EP2007058523, PCT/EP200758523, PCT/EP7/058523, PCT/EP7/58523, PCT/EP7058523, PCT/EP758523, WO 2008/025682 A1, WO 2008025682 A1, WO 2008025682A1, WO-A1-2008025682, WO2008/025682A1, WO2008025682 A1, WO2008025682A1|
|Inventors||Harmander Singh Deogun, Kevin John Nowka, Rahul Rao, Robert Matthew Senger|
|Applicant||International Business Machines Corporation, Ibm United Kingdom Limited|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (2), Classifications (10), Legal Events (4)|
|External Links: Patentscope, Espacenet|
STATIC PULSED BUS CIRCUIT AND METHOD HAVING DYNAMIC POWER SUPPLY RAIL SELECTION
BACKGROUND OF THE INVENTION
The present invention relates generally to electronic interface bus circuits, and more particularly to a pulsed bus circuit and operating method having dynamic power supply rail selection.
Description of the Related Art
Low power electronic systems incorporating large parallel buses are increasingly prevalent, as microprocessor systems are used in notebook computers, personal digital assistants
(PDAs) and other electronic appliances designed for portable battery-operated use. Power consumption is also an increasingly important issue in general, as increasing deployment of large-scale computing systems along with an increase in processing power and consequent rise in power consumption raises the cost of operating those systems to businesses and society at-large.
As circuit operating frequencies and die/circuit sizes increase and operating voltages decrease, bus repeaters are necessary in increasing proportion to interconnect internal circuits in high-density electronic devices. The repeaters are necessary to maintain propagation delay and signal skew at tolerable levels as circuit technologies advance.
However, inclusion of large numbers of bus repeaters raises quiescent bus power consumption of the device including them significantly, due to an increased number of power supply leakage paths provided through the repeaters, even when the repeaters are inactive. The dynamic bus power consumption is also increased, due to the additional drive elements included on the bus. One bus repeater solution that has been implemented to reduce the power consumed by interface buses, is a "static pulsed bus." The static pulsed bus has desirable characteristics in that signal delay and power dissipation due to inter-bus- line coupling capacitance is decreased. Static pulse bus circuits operate by propagating pulses instead of levels, and the pulses are unidirectional for each set of parallel bus segments, reducing the energy used to charge the inter-bus- line coupling capacitance. The presence of a pulse during a period indicates a change in logic state on the particular bus-line and the absence of a pulse indicates no change in logic state. When two parasitically-coupled bus lines transition in the same direction, the effect of the coupling capacitance is zero. When only one bus line transitions, the effect is half of that of the worst-case condition of opposite transitions on the bus lines, which occur in non-pulsed bus designs. Standard buses also have increased current drive requirement in the repeaters in order to overcome the above-described worst-case switching condition, leading to increased leakage through the larger devices.
Therefore, static pulsed bus designs are desirable due to the reduction in both dynamic power consumption due to the reduced effective inter-bit- line capacitance and static leakage current. However, even though static pulsed bus designs lower the power consumption of bus repeater circuits, their power consumption is still significant due to the increasing number of bus repeaters required in emerging electronic devices.
Therefore, it would be desirable to provide a static pulsed bus architecture that further reduces bus power consumption due to leakage and dynamic power consumption.
SUMMARY OF THE INVENTION
The present invention seeks to reduce bus power consumption in a static pulsed bus repeater circuit in a method and apparatus. The method is a method of operation of the apparatus which is a bus interface circuit including a plurality of bus repeaters.
The bus repeaters are organized into alternating groups of repeaters, corresponding to odd and even positions within the cascade of repeaters on each bus line. A first (even) group of bus repeaters has a selectable power supply voltage at one of the power supply rails of the bus repeaters in the first group. The power supply voltage is selected in conformity with the state of the signal input of the bus repeater, so that when a pulse is received by the repeater, the power supply voltage is increased while the (opposite polarity) output pulse is being relayed to the next bus repeater. The power supply voltage can be selected by an analog selector having a select input coupled to the input of the repeater. The second (odd) group of repeaters operates from the lower power supply rail selectable at the first group of repeaters.
The second group of repeaters may also include a selectable power supply voltage on the power supply rail opposite the power supply rail that has selectable voltage in the first group of repeaters. If so, then the first group of repeaters has a second power supply rail (opposite the selectable-voltage power supply rail) that is set to the higher of the voltages selectable at the second group of repeaters.
The foregoing and other features, and advantages of the invention will be apparent from the following, more particular, description of the preferred embodiment of the invention, as illustrated in the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
The present invention will now be described, by way of example only, with reference to the accompanying drawings in which:
Figure 1 is a schematic diagram of a bus in accordance with an embodiment of the invention;
Figure 2 is a time- voltage diagram depicting the relationship of signals in the bus of Figure l;
Figure 3 is a schematic diagram of a bus in accordance with an embodiment of the invention; and
Figure 4 is a time- voltage diagram depicting the relationship of signals in the bus of Figure DESCRIPTION OF ILLUSTRATIVE EMBODIMENT
With reference now to the figures, and in particular with reference to Figure 1 , a schematic diagram of a bus circuit embodying a method and an apparatus in accordance with the present invention is shown. A data signal is conveyed from an input node DATA IN to an output node DATA OUT by a static pulsed bus circuit as shown. A cascade of inverters, represented by inverters II, 12 through IN, IN+I repeat a pulsed bus signal provided by a pulse generator 12 that generates pulses in response to changes in a data signal latched by a clock signal elk from input node DATA IN by a latch 1OA. The polarity of the pulses is opposite from one stage to the next in the illustrated circuit due to the inversion through inverters II,
12 through IN, IN+I - While the illustrative embodiment shows bus repeaters as inverters II, 12 through IN, IN+I, it is contemplated that other circuits can be used as bus repeaters in other embodiments of the present invention. The data signal is reconstructed at the opposite end of the interface by a toggle flip-flop 14 and is latched by a latch 1OB by clock signal elk.
In the depicted embodiment, the first power rail of odd-numbered inverters exemplified by inverters Il and IN is statically provided from power supply VDL, but the power supply voltage across even-numbered inverters 12 and IN+I is dynamically selected by selectors 16A through 16Z and applied to the first power supply rail of inverters 12 and IN+I . The second power supply rail of all of inverters II, 12 through IN, IN+I is connected to a third power supply value, which in the illustrated circuit is ground. When a falling- voltage pulse is present at the input of even stage inverters 12 and IN+I, the power supply voltage provided to the positive supply rails of inverters 12 and IN+I is boosted from the voltage of power supply VDL to the voltage level of power supply VDH- The result is that the delay of inverters 12 and IN+I is reduced in producing the positive pulses at the outputs of inverters 12 and IN+I during the pulse propagation, but the power supply voltage level is quickly restored to a lower level after the pulse has been propagated. In the depicted embodiment, the power supply voltage across odd inverters such as Il and IN is fixed at VDL- VDL is generally chosen as approximately 0.7 VDH, which in the present invention has been shown to yield power savings on the order of 35% reduction in dynamic power consumption over standard static pulsed bus circuit and a 12-15% increase in performance. The interface bus of the present invention can also reduce static power consumption when the bus is inactive by approximately 65%.
The present invention reduces the power consumption by reducing the effect of coupling capacitance between each bit-line segment that connects inverters II, 12 through IN, IN+I with other, parallel bit-line segments carrying other data bits (not shown), as is attained in static pulsed buses in general. However, the depicted circuit further reduces power consumption by maintaining every other stage of the bus (e.g. the outputs of inverters Il and IN) at a reduced voltage level between pulses, reducing the energy needed to switch those inverters to the opposite supply rail when a pulse is received. Further, because inverters Il and IN are supplied with a lowered positive power supply VDL, when a boosted positive pulse is received at the input to inverters Il and IN, PMOS device Pl in inverters Il and IN is driven farther into cut-off, permitting NMOS device Nl to turn on faster, reducing cross-conduction energy and thus reducing overall power consumption, while reducing bus delay. The voltage reduction also provides the added benefit of reducing the magnitude of power supply current spikes from switching the bus repeater stages, both by reducing cross-conduction energy and also by lowering the stored energy associated with the static bus voltage. When the bus is inactive, the reduction in the power supply voltage present across inverters II, 12 through IN, IN+I reduces the power consumption accordingly, and any additional leakage sources present along the bus segments connected to the outputs of the even inverters (e.g., inverters 12,
IN+I), will also have reduced current, due to the lowered static voltage level present on the bus segments.
Referring now to Figure 2, signals within the circuit of Figure 1 are illustrated in a time- voltage diagram. The input to Pulse generator 12 is illustrated as PG IN and is derived from the Data IN signal by latching the Data IN signal on rising edges of clock signal elk. Signal PG Out is the output of pulse generator 12 and is applied to the cascaded inverter chain II- IN+ Ϊ . The signals at the outputs of inverters II, 12 and IN+I and are denoted as Il Out, 12 Out, and IN+I Out, respectively. As can be seen in the diagram, signals Il Out and IN+I Out are falling- voltage pulses having a quiescent value of VDL and a pulse peak value of zero. Signal
12 Out has a quiescent value of zero and a pulse peak value of VDH- Signal T Out is the reconstructed data signal at the output of toggle flip-flop 14. Referring now to Figure 3, a bus circuit in accordance with another embodiment of the present invention. The depicted embodiment is similar to that of the circuit of Figure 1, and therefore only differences between them will be described below. In contrast to the embodiment of Figure 1, each of the bus repeater stages provided by inverters II, 12 through IN, IN+ i has a selectable power supply rail voltage. Additional selectors 18A through 18Z select between a third and fourth power supply voltage, depicted as ground and VSH- In the circuit of Figure 3, the quiescent voltage on the outputs of inverters Il and IN is greater than in the circuit of Figure 1, or in other terms, the magnitude of the power supply voltage is selectably reduced for Il and IN just as the magnitude of the power supply voltage supplied to inverters 12 and IN+I is reduced in both the circuit of Figure 1 and the circuit of Figure 3.
The selection input of selectors 18A through 18Z are connected to the inputs of the corresponding inverters Il and IN SO that when a pulse is received at the inputs of those inverters, the (lower) power supply rail connected to transistor Nl in those inverters is reduced to zero. When the pulse at the input of the inverter terminates, the power supply rail is selectably restored to VSH-
The lower power supply rails of inverters 12 and IN+I are connected to a fourth power supply VSH, rather than ground as in the circuit of Figure 1. The resulting circuit has the similar advantages as in the circuit of Figure 1, with an extension of the benefits described above with respect to the circuit of Figure 1 to all bus repeater stages, and a reduction in static bus voltage at all bus repeater stages to a static voltage of VDL - VSH- For example, transistor Nl of inverters 12 and IN+I will be driven further into cutoff because of a gate pulse peak voltage of zero, with a source voltage VSH, permitting transistor Pl of inverters 12 and IN+I to charge the bus to voltage VDH faster and reducing cross-conduction switching energy in inverters 12 and IN+I . The circuit of Figure 3 will have a static and dynamic power consumption even lower than that of the circuit of Figure 1, with the penalty of added complexity, circuit area and the additional requirement of a fourth power supply output.
Referring now to Figure 4, signals within the circuit of Figure 3 are illustrated in a time- voltage diagram. The input to Pulse generator 12 is illustrated as PG IN and is derived from the Data IN signal by latching the Data IN signal on rising edges of clock signal elk. Signal PG Out is the output of pulse generator 12 and is applied to the cascaded inverter chain II- IN+ L The signals at the outputs of inverters II, 12 and IN+I and are denoted as Il Out, 12 Out, and IN+I Out, respectively. As can be seen in the diagram, signals Il Out and IN+I Out are falling- voltage pulses having a quiescent value of VDL and a pulse peak value of zero. Signal 12 Out has a quiescent value of VSH and a pulse peak value of VDH- Signal T Out is the reconstructed data signal at the output of toggle flip-flop 14.
While the invention has been particularly shown and described with reference to the preferred embodiment thereof, it will be understood by those skilled in the art that the foregoing and other changes in form, and details may be made therein without departing from the spirit and scope of the invention.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US20030038653 *||29 Oct 2002||27 Feb 2003||Mitsubishi Denki Kabushiki Kaisha||Semiconductor integrated circuit device including logic gate that attains reduction of power consumption and high-speed operation|
|US20060023519 *||7 Dec 2004||2 Feb 2006||Seouk-Kyu Choi||Method of operating a semiconductor device and the semiconductor device|
|Cooperative Classification||Y02B60/1235, G06F1/32, G06F13/4045, G06F1/3253, G06F1/3203|
|European Classification||G06F1/32P5P1, G06F13/40D5R, G06F1/32P, G06F1/32|
|14 May 2008||121||Ep: the epo has been informed by wipo that ep was designated in this application|
Ref document number: 07802656
Country of ref document: EP
Kind code of ref document: A1
|27 Feb 2009||WWE||Wipo information: entry into national phase|
Ref document number: 2009526040
Country of ref document: JP
|27 Feb 2009||ENP||Entry into the national phase in:|
Ref document number: 2009526040
Country of ref document: JP
Kind code of ref document: A
|3 Mar 2009||NENP||Non-entry into the national phase in:|
Ref country code: DE