US4298929A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Integrated multilevel storage hierarchy for a data processing system with improved channel to memory write capability
Assignee: International Business Machines Corp
•
Filing: 1979-01-26
•
Granted: 1981-11-03
•
Expires: 1999-01-26
Abstract
Abstract A data processing system including at least one channel and a multilevel store has the capability of performing a channel write to main memory where the data to be written into main memory crosses a double word boundary in a partial write store. The partial write store is accomplished by a merge operation which takes place in the memory system in a manner such that the main processor, channel and a cache store are freed up for further operation prior to the completion of the write to main memory.
Inventors
- 👤 Anthony J. Capozzi
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-24 18:26
CPC Classifications
G; G06; G06F; G06F12/00; G06F12/02; G06F12/08; G06F12/0802; G06F12/0804; G06F12/04
Claims Analysis
Total Claims: 2
Claim 1
Independent Claim
485 words
In a data processing system including a processing unit, at least one channel connected to said processing unit, a first high speed, low capacity memory cache, a second lower speed, high capacity main memory and an input/output register interconnecting said cache and said processing unit, wherein data is stored in said main memory in the form of full words, each full word comprising a particular number of bytes and having an upper and lower boundary, and wherein data is stored in said cache in the form of cache pages each page comprising a plurality of full words, said channel providing input data for storage in said main memory, said input data being either in the form of full words whose boundaries correspond to boundaries of full words in said main memory or in the form of partial words which have less bytes than said particular number of bytes and the boundaries of which do not coincide with the boundaries of full words in main memory, improved channel to memory write means for storing information in said main memory in response to a partial channel write request wherein the data supplied by said channel and loaded into said I-O register includes a partial word whose boundaries do not coincide with the boundaries of a full word in said main memory, comprising: a swap buffer adapted to receive data from said input/output register and having at least one cache page storage capability; an error correction circuit and logic subassembly connected between said input/output register and said main memory, said subassembly adapted to receive data from said swap buffer; means responsive to a partial channel write request to determine if the data addressed by the channel write request is in said cache (a hit) or if the data addressed is not in said cache (a miss); first means responsive to a hit to generate a full word by retrieving the addressed data from the cache and supplementing the partial word from said partial channel write request with the data from the cache in said I-O register a plurality of bytes at a time; means responsive to said first means to transfer said full word generated by said first means to said main memory; second means responsive to a miss to release said IPU to load said partial word from said I-O register into said swap buffer to retrieve the addressed data from said main memory and to combine the data from said partial channel write stored in said swap buffer with data from main memory; in said error correction circuits and logic subassembly, and means responsive to said second means to store said combined full word to said main memory, whereby said IPU is available for executing instructions unrelated to channel data transfers, while information to complete partial channel write requests is being retrieved and while said data is being transferred from said swap buffer to said memory.
Claim 2
Dependent Claim
485 words
In a data processing system including a processing unit, at least one channel connected to said processing unit, a first high speed, low capacity memory cache, a second lower speed, high capacity main memory and an input/output register interconnecting said cache and said processing unit, wherein data is stored in said main memory in the form of full words, each full word comprising a particular number of bytes and having an upper and lower boundary, and wherein data is stored in said cache in the form of cache pages each page comprising a plurality of full words, said channel providing input data for storage in said main memory, said input data being either in the form of full words whose boundaries correspond to boundaries of full words in said main memory or in the form of partial words which have less bytes than said particular number of bytes and the boundaries of which do not coincide with the boundaries of full words in main memory, improved channel to memory write means for storing information in said main memory in response to a partial channel write request wherein the data supplied by said channel and loaded into said I-O register includes a partial word whose boundaries do not coincide with the boundaries of a full word in said main memory, comprising: a swap buffer adapted to receive data from said input/output register and having at least one cache page storage capability; an error correction circuit and logic subassembly connected between said input/output register and said main memory, said subassembly adapted to receive data from said swap buffer; means responsive to a partial channel write request to determine if the data addressed by the channel write request is in said cache (a hit) or if the data addressed is not in said cache (a miss); first means responsive to a hit to generate a full word by retrieving the addressed data from the cache and supplementing the partial word from said partial channel write request with the data from the cache in said I-O register a plurality of bytes at a time; means responsive to said first means to transfer said full word generated by said first means to said main memory; second means responsive to a miss to release said IPU to load said partial word from said I-O register into said swap buffer to retrieve the addressed data from said main memory and to combine the data from said partial channel write stored in said swap buffer with data from main memory; in said error correction circuits and logic subassembly, and means responsive to said second means to store said combined full word to said main memory, whereby said IPU is available for executing instructions unrelated to channel data transfers, while information to complete partial channel write requests is being retrieved and while said data is being transferred from said swap buffer to said memory.
References Cited (Backward Citations)
US3569938A
Cited
US3588829A
Cited
US3864669A
Cited
US4084258A
Cited
US3866183A
Cited
US3896419A
Cited
US4071910A
Cited
US3967247A
Cited
US4056845A
Cited
US4084234A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1984-11-02 | 1985-05-04 | 1985-11-03 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 1988-11-02 | 1989-05-04 | 1989-11-03 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 1992-11-02 | 1993-05-04 | 1993-11-03 | $7700.00 | Window Closed / Lapsed |