(lipincon) |
|||
(One intermediate revision by the same user not shown) | |||
Line 1: | Line 1: | ||
{{tsmc title|Low-voltage-In-Package-INterCONnect (LIPINCON)}}{{interconnect arch}} | {{tsmc title|Low-voltage-In-Package-INterCONnect (LIPINCON)}}{{interconnect arch}} | ||
'''Low-voltage-In-Package-INterCONnect''' ('''LIPINCON''') is a proprietary system [[interconnect architecture]] that facilitates data transmission across all linked components. | '''Low-voltage-In-Package-INterCONnect''' ('''LIPINCON''') is a proprietary system [[interconnect architecture]] that facilitates data transmission across all linked components. | ||
+ | |||
+ | == Overview == | ||
+ | LIPINCON is an interconnect architecture designed for [[chiplet]] designs with advanced packaging technologies such as {{tsmc|InFO}} and {{tsmc|CoWoS}}. LIPINCON uses a timing compensation mechanism in order to achieve a low-power and small area slave PHY (e.g., memory die or a transceiver die) while excluding the PLL/DLL but retaining good timing margins. | ||
+ | |||
+ | == Architecture == | ||
+ | The architecture is designed to take advantage of an advanced packaging technology such as {{tsmc|CoWoS}} whereby multiple dies are stacked on a silicon interposer communicating by PHYs over short wires. LIPINCON uses two different types of PHYs: '''PHYC''' for an SoC die and '''PHYM''' for a memory, transceiver, or similar kind of dies. The clock source in the PHYM relies on one system clock propagated from PHYC. There is no [[phase-locked loop|PLL]]/[[delay-locked loop|DLL]] function blocks built in PHYM because it assumes that the SoC will generally be fabricated on a more advanced node and is therefore capable of handling this more efficiently than a memory/transceiver that may be made on a more mature node. | ||
+ | |||
+ | == Bibliography == | ||
+ | * TSMC, MS Lin et a., 2013 VLSI Symposium |
Latest revision as of 12:36, 22 June 2019
Interconnect Architectures | |
Concepts | |
General | |
Peripheral | |
Storage Devices | |
Audio Devices | |
Low-voltage-In-Package-INterCONnect (LIPINCON) is a proprietary system interconnect architecture that facilitates data transmission across all linked components.
Overview[edit]
LIPINCON is an interconnect architecture designed for chiplet designs with advanced packaging technologies such as InFO and CoWoS. LIPINCON uses a timing compensation mechanism in order to achieve a low-power and small area slave PHY (e.g., memory die or a transceiver die) while excluding the PLL/DLL but retaining good timing margins.
Architecture[edit]
The architecture is designed to take advantage of an advanced packaging technology such as CoWoS whereby multiple dies are stacked on a silicon interposer communicating by PHYs over short wires. LIPINCON uses two different types of PHYs: PHYC for an SoC die and PHYM for a memory, transceiver, or similar kind of dies. The clock source in the PHYM relies on one system clock propagated from PHYC. There is no PLL/DLL function blocks built in PHYM because it assumes that the SoC will generally be fabricated on a more advanced node and is therefore capable of handling this more efficiently than a memory/transceiver that may be made on a more mature node.
Bibliography[edit]
- TSMC, MS Lin et a., 2013 VLSI Symposium