Electrical computers and digital processing systems: memory – Storage accessing and control – Specific memory composition
Reexamination Certificate
1999-11-16
2002-06-18
Kim, Hong (Department: 2187)
Electrical computers and digital processing systems: memory
Storage accessing and control
Specific memory composition
C711S106000, C711S167000, C711S211000, C365S222000, C365S230010, C365S233100
Reexamination Certificate
active
06408356
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to personal computer systems which utilize SIMMs for memory, and more particularly to a personal computer system and a SIMM for use thereon wherein the SIMM is configured to utilize DRAMs having greater row address space than there are row address signals from the memory controller to address the DRAMs.
BACKGROUND INFORMATION
Personal computer systems have been designed which are adapted to use Single Inline Memory Modules (SIMMs) as on-board memory. The SIMMs utilize random access memory (RAM) which can be either dynamic random access memory (DRAMs) or static random access memory (SRAMs). The computer system utilizes a memory controller in the form of an integrated circuit (IC) chip which controls the various signals which are sent from the central processing unit (CPU) to the memory to either write data to or read data from the memory. The memory controller is designed to accommodate certain configurations of DRAM chips and SIMMs. For example, a common SIMM has been a 72-pin SIMM which utilizes 16 1M×4 DRAM chips to provide 8 megabytes of memory. This configuration allocates a certain number of pins for addressing memory locations and provides two row activation signals (RAS) for activating the DRAMs on either read or write cycles. The DRAMs are functionally arranged in two sets of 8 chips each, with each set having ten row addresses and ten column addresses. Such an address configuration is known as a 10×10. Thus, the need arises for two RAS's, one RAS for one of the sets of 8 chips and one for the other set of 8 chips. Hence, this scheme is known as 10×10 addressing with two sets of addressable chips. To address one set of DRAM chips, one of the RAS signals is driven active, and to address the other set of DRAM chips the other RAS signal is driven active.
As chip technology progresses and larger size chips become cheaper, it becomes economical to replace several chips with a single chip. One particular configuration of chips which has become readily available at attractive prices is a 2M×8 chip, which contains four times the amount of data as a 1M×4. Thus, 4 2M×8 chips can store the same amount of data as 16 1M×4 chips. Thus, these chips are desirable to be utilized on a SIMM for cost reasons. However, it is desirable in providing or utilizing these more cost-effective chips that they be utilized in systems having pre-existing memory controllers which have been designed for the 1M×4 chips. This causes a problem. The 2M×8 SIMMs require 11 row addresses rather than the 10 required by the 1M×4 chips, and thus there is not the available row address without the addition of another pin and significant redesign—and even then the design becomes difficult because all 72 pins have normally been utilized. However, only one RAS signal is needed to activate the read or write functions of the four chips. Thus, one technique to overcome the 11 row limitation is by utilizing the signal on one of the RAS lines to function as an address bit on the 11th row address space.
While this does work, it has certain disadvantages. One of the disadvantages is that utilizing one of the RAS's directly on the 11th address bit will not accommodate a refresh function when refresh is done by both RAS's going low, known as ROR (RAS Only Refresh), rather than the refresh function being performed by a CAS before RAS signal configuration known as CBR (CAS Before RAS). This is because the 11th address bit will always be active when both RAS's go active, thus refreshing only one-half of the chip, i.e., the address portion of the chip utilizing the active bit in the 11 th address space which is the most significant bit. Another problem is that since only a single RAS activation is necessary for the 2M×8 chips, logic such as an AND gate is necessary to perform a row activation function when either RAS input goes low. This additional logic can lead to timing problems, especially with respect to the timing in which activation of the 11th row address takes place, vis-a-vis, the activation of the RAS signal on the chip following the logic circuitry to combine the two RAS signals as a single master RAS signal. OM solution is found in related application Serial No. 08/582,010. However, this solution uses only one of the RAS signals to generate the high order bit, and thus can generate hot spots in memory under certain conditions.
It is therefore an object of the present invention to provide a computer system and SIMM configuration and method of operation which allows DRAMs having a greater number of row address than are supplied by a memory controller to be used by a system using such a memory controller, which eliminates the cause of hot spots during certain operations
SUMMARY OF THE INVENTION
A computer system and method of operation is provided wherein the memory controller of the system generates first and second address signals, typically RAS signals, and address bits for Y rows of addresses in memory; and wherein the memory of the system is configured with Y+1 rows of addresses operable by a single address signal, typically a single RAS signal. The system includes logic, preferably on an ASIC chip that converts one of said RAS signals from the memory controller, together with at least one address signal generated by the CPU and propagated by the system memory controller, to the high order address bit for the memory rows, thus constituting Y+1 rows of address activated space, and generates a master RAS signal when either RAS generated by the memory controller goes active. The logic also provides for a refresh operation of all memory location during a RAS only refresh (ROR) operation.
REFERENCES:
patent: 4656605 (1987-04-01), Clayton
patent: 4882700 (1989-11-01), Mauritz et al.
patent: 5241643 (1993-08-01), Durkin et al.
patent: 5253357 (1993-10-01), Allen et al.
patent: 5260892 (1993-11-01), Testa
patent: 5270964 (1993-12-01), Bechtolsheim et al.
patent: 5272664 (1993-12-01), Alexander et al.
patent: 5278801 (1994-01-01), Dresser et al.
patent: 5357624 (1994-10-01), Lavan
patent: 5371866 (1994-12-01), Cady
patent: 5414670 (1995-05-01), Schaefer
patent: 5465237 (1995-11-01), Chen et al.
patent: 5613094 (1997-03-01), Khan et al.
IBM Technical Disclosure Bulletin, vol. 37, No. 10, Oct. 1994 “Stacking Single Inline Memory Module Card”.
Hogg William N.
International Business Machines - Corporation
Kim Hong
Song Jasmine
LandOfFree
Apparatus and method for modifying signals from a CPU to a... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Apparatus and method for modifying signals from a CPU to a..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Apparatus and method for modifying signals from a CPU to a... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2896424