Multiple register memory access instructions, processors, methods, and systems
A processor includes N-bit registers and a decode unit to receive a multiple register memory access instruction. The multiple register memory access instruction is to indicate a memory location and a register. The processor includes a memory access unit coupled with the decode unit and with the N-bit registers. The memory access unit is to perform a multiple register memory access operation in response to the multiple register memory access instruction. The operation is to involve N-bit data, in each of the N-bit registers comprising the indicated register. The operation is also to involve different corresponding N-bit portions of an M×N-bit line of memory corresponding to the indicated memory location. A total number of bits of the N-bit data in the N-bit registers to be involved in the multiple register memory access operation is to amount to at least half of the M×N-bits of the line of memory.
1. A processor comprising:
a plurality of N-bit registers;
a decode unit to decode a multiple register memory access instruction, the multiple register memory access instruction to indicate a memory location and to indicate a register; and
a memory access unit coupled with the decode unit and coupled with the plurality of the N-bit registers, the memory access unit to perform a multiple register memory access operation in response to the multiple register memory access instruction, the multiple register memory access operation to involve N-bit data, in each of the plurality of the N-bit registers that are to comprise the indicated register, and different corresponding N-bit portions of an M×N-bit line of memory, that is to correspond to the indicated memory location, in which a total number of bits of the N-bit data, in the plurality of the N-bit registers to be involved in the multiple register memory access operation, is to amount to at least half of the M×N-bit line of memory.
2. The processor of claim 1 , in which the memory access unit is to perform the operation in which the total number of bits of the N-bit data in the plurality of the N-bit registers to be involved in the multiple register memory access operation is to amount to all of the M×N-bit line of memory.
3. The processor of claim 1 , in which the memory access unit is to perform the operation in which the total number of bits of the N-bit data in the plurality of the N-bit registers to be involved in the multiple register memory access operation is to amount to at least 256-bits.
4. The processor of claim 3 , in which the memory access unit is to perform the operation in which the total number of bits of the N-bit data in the plurality of the N-bit registers to be involved in the multiple register memory access operation is to amount to at least 512-bits.
5. The processor of claim 1 , in which the memory access unit is to perform the operation that is to involve the N-bit data in each of at least three N-bit registers.
6. The processor of claim 5 , in which the memory access unit is to perform the operation that is to involve the N-bit data in each of at least four N-bit registers.
7. The processor of claim 1 , in which the memory access unit is to perform the operation that is to involve 128-bit data, in each of at least four 128-bit registers, and the different corresponding N-bit portions which are 128-bit portions of the line of memory that is to be at least 512-bits.
8. The processor of claim 1 , in which the memory access unit is to perform the operation that is to involve 256-bit data, in each of at least two 256-bit registers, and the different corresponding N-bit portions which are 256-bit portions of the line of memory that is to be at least 512-bits.
9. The processor of claim 1 , in which the processor comprises a reduced instruction set computing (RISC) processor, and in which the multiple register memory access instruction comprises a multiple register load from memory instruction, and in which the memory access unit is to load the different N-bit portions of the M×N-bit line of memory in the plurality of the N-bit registers, in response to the multiple register load from memory instruction, in which the total number of bits of the different N-bit portions to be loaded in the plurality of the N-bit registers from the M×N-bit line of memory is to amount to at least half of the M×N-bit line of memory.
10. The processor of claim 9 , in which the memory access unit is to load different 128-bit portions of the line of memory which is at least 512-bits in each of at least four 128-bit registers.
11. The processor of claim 9 , in which the memory access unit is to load different 256-bit portions of the line of memory which is at least 512-bits in each of at least two 256-bit registers.
12. The processor of claim 1 , in which the processor comprises a reduced instruction set computing (RISC) processor, and in which the multiple register memory access instruction comprises a multiple register write to memory instruction, and in which the memory access unit is to write the N-bit data, from the plurality of the N-bit registers, to the different corresponding N-bit portions of the M×N-bit line of memory, in response to the multiple register write to memory instruction, in which the total number of bits of the N-bit data to be written from the plurality of the N-bit registers to the M×N-bit line of memory is to amount to at least half of the M×N-bit line of memory, in which the at least half of the M×N-bit line of memory is at least 256-bits.
13. The processor of claim 1 , in which the multiple register memory access instruction is to explicitly specify each of the plurality of registers.
14. The processor of claim 1 , in which the multiple register memory access instruction is to specify a number of the plurality of registers.
15. A processor comprising:
a cache to store a plurality of cache lines;
a plurality of general purpose registers;
a plurality of 128-bit packed data registers, including a first destination 128-bit packed data register, and a second destination 128-bit packed data register;
an instruction fetch unit to fetch instructions, including a load from memory instruction;
a decode unit to decode the load from memory instruction, the load from memory instruction indicating a starting memory location in a memory, the starting memory location associated with data to be loaded, and the load from memory instruction having a first field to specify the first destination 128-bit packed data register, and having a second field to specify the second destination 128-bit packed data register; and
a memory access unit coupled to the decode unit, and coupled to the plurality of 128-bit packed data registers, the memory access unit to perform a load from memory operation in response to the decoded load from memory instruction, the load from memory operation to:
load a first 128-bit data from the indicated starting memory location, and store the loaded first 128-bit data in the first destination 128-bit packed data register; and
load a second 128-bit data, which is adjacent to the first 128-bit data, and store the loaded second 128-bit data in the second destination 128-bit packed data register.
16. The processor of claim 15 , further comprising a plurality of write mask registers to predicate result vector writes.
17. The processor of claim 15 , further comprising a 16-wide vector processing unit to execute double-precision float instructions.
18. The processor of claim 15 , wherein the cache is to store 512-bit cache lines.
19. The processor of claim 15 , wherein the cache is to store 1024-bit cache lines.
20. The processor of claim 15 , wherein the cache is to store cache lines that are a multiple of a width of the 128-bit packed data registers.
21. The processor of claim 15 , further comprising:
a branch prediction unit; and
a translation lookaside buffer (TLB).
22. The processor of claim 15 , wherein the processor is a reduced instruction set computing (RISC) processor.
23. A method performed by a processor, the method comprising:
storing a plurality of cache lines in a cache of the processor;
storing data in a plurality of general purpose registers of the processor;
fetching a load from memory instruction with an instruction fetch unit of the processor;
decoding the load from memory instruction with a decode unit of the processor, the load from memory instruction indicating a starting memory location in a memory, the starting memory location associated with data to be loaded, and the load from memory instruction having a first field specifying a first destination 128-bit packed data register of the processor, and having a second field specifying a second destination 128-bit packed data register of the processor; and
performing a load from memory operation, in response to the decoded load from memory instruction, the load from memory operation including:
loading a first 128-bit data from the indicated starting memory location, and storing the loaded first 128-bit data in the first destination 128-bit packed data register; and
loading a second 128-bit data, which is adjacent to the first 128-bit data, and storing the loaded second 128-bit data in the second destination 128-bit packed data register.
24. The method of claim 23 , further comprising predicating result vector writes with a plurality of write mask registers.
25. The method of claim 23 , further comprising executing double-precision float instructions with a 16-wide vector processing unit of the processor.
26. The method of claim 23 , wherein said storing the cache lines in the cache comprises storing 512-bit cache lines in the cache.
27. The method of claim 23 , wherein said storing the cache lines in the cache comprises storing 1024-bit cache lines in the cache.
28. The method of claim 23 , further comprising predicting branches with a branch prediction unit of the processor.
29. An article of manufacture comprising a non-transitory machine-readable storage medium, the non-transitory machine-readable storage medium storing a set of instructions, including a load from memory instruction, the set of instructions if processed by a machine to cause the machine to perform operations comprising to:
store a plurality of cache lines in a cache of a processor;
store data in a plurality of general purpose registers of the processor;
the load from memory instruction if processed by the machine to cause the machine to perform operations comprising to:
decode the load from memory instruction with a decode unit of the processor, the load from memory instruction to indicate a starting memory location in a memory, the starting memory location associated with data to be loaded, and the load from memory instruction having a first field to specify a first destination 128-bit packed data register of the processor, and having a second field to specify a second destination 128-bit packed data register of the processor; and
perform a load from memory operation, in response to the decoded load from memory instruction, the load from memory operation including:
loading a first 128-bit data from the indicated starting memory location, and storing the loaded first 128-bit data in a first destination 128-bit packed data register of the processor; and
loading a second 128-bit data, which is adjacent to the first 128-bit data, and storing the loaded second 128-bit data in a second destination 128-bit packed data register of the processor.
30. The article of manufacture of claim 29 , wherein the set of instructions, if processed by the machine, are to cause the machine to perform operations comprising to predicate result vector writes with a plurality of write mask registers.
31. The article of manufacture of claim 29 , wherein the set of instructions, if processed by the machine, are to cause the machine to perform operations comprising to store 512-bit cache lines in the cache.
32. The article of manufacture of claim 29 , wherein the set of instructions, if processed by the machine, are to cause the machine to perform operations comprising to store 1024-bit cache lines in the cache.