U.S. patent application number 11/485960 was filed with the patent office on 2008-01-17 for latest producer tracking in an out-of-order processor, and applications thereof.
This patent application is currently assigned to MIPS Technologies, Inc.. Invention is credited to Xing Yu Jiang, Kjeld Svendsen.
Application Number | 20080016326 11/485960 |
Document ID | / |
Family ID | 38950611 |
Filed Date | 2008-01-17 |
United States Patent
Application |
20080016326 |
Kind Code |
A1 |
Svendsen; Kjeld ; et
al. |
January 17, 2008 |
Latest producer tracking in an out-of-order processor, and
applications thereof
Abstract
A processor and system for latest producer tracking. In one
embodiment, the processor includes an operand renamer circuit that
includes a register rename map, a producer tracking circuit that
includes a producer tracking map, and a results buffer allocater
circuit that includes a results buffer free list. Control logic
modifies in-register status values stored in the register rename
map based on producer tracking status values stored in the producer
tracking map. The producer tracking status values stored in the
producer tracking map are modified based on buffer identification
values output by the results buffer allocater circuit.
Inventors: |
Svendsen; Kjeld; (San Jose,
CA) ; Jiang; Xing Yu; (Palo Alto, CA) |
Correspondence
Address: |
STERNE, KESSLER, GOLDSTEIN & FOX P.L.L.C.
1100 NEW YORK AVENUE, N.W.
WASHINGTON
DC
20005
US
|
Assignee: |
MIPS Technologies, Inc.
Mountain View
CA
|
Family ID: |
38950611 |
Appl. No.: |
11/485960 |
Filed: |
July 14, 2006 |
Current U.S.
Class: |
712/217 ;
712/E9.029; 712/E9.037; 712/E9.049; 712/E9.053 |
Current CPC
Class: |
G06F 9/3857 20130101;
G06F 9/30141 20130101; G06F 9/30174 20130101; G06F 9/30149
20130101; G06F 9/3017 20130101; G06F 9/3012 20130101; G06F 9/384
20130101 |
Class at
Publication: |
712/217 |
International
Class: |
G06F 9/30 20060101
G06F009/30 |
Claims
1. A processor, comprising: an operand renamer circuit that stores
a plurality of in-register status values, wherein each in-register
status value is associated with a particular architectural state
register; a producer tracking circuit that stores a plurality of
producer tracking status values, wherein each producer tracking
status value is associated with a particular physical register; and
control logic that modifies the in-register status values based on
the producer tracking status values.
2. The processor of claim 1, wherein the operand renamer circuit
includes a register rename map.
3. The processor of claim 2, wherein values stored in the register
rename map are accessed using architectural state register
identification values.
4. The processor of claim 1, wherein the producer tracking circuit
includes a producer tracking map.
5. The processor of claim 4, wherein values stored in the producer
tracking map are accessed using physical register identification
values.
6. The processor of claim 1, wherein each producer tracking status
value consists of a single bit.
7. The processor of claim 1, wherein the operand renamer circuit
sends a signal to the producer tracking circuit to modify a
particular producer tracking status value.
8. The processor of claim 1, wherein the processor further
comprises a results buffer allocater circuit, and the results
buffer allocater circuit sends a signal to the producer tracking
circuit to modify a particular producer tracking status value.
9. A system, comprising: a processor that includes an operand
renamer circuit that stores a plurality of in-register status
values, wherein each in-register status value is associated with a
particular architectural state register, a producer tracking
circuit that stores a plurality of producer tracking status values,
wherein each producer tracking status value is associated with a
particular physical register, and control logic that modifies the
in-register status values based on the producer tracking status
values; and memory coupled to the processor.
10. The system of claim 9, wherein the operand renamer circuit
includes a register rename map.
11. The system of claim 9, wherein the producer tracking circuit
includes a producer tracking map.
12. The system of claim 9, wherein the operand renamer circuit
sends a signal to the producer tracking circuit to modify a
particular producer tracking status value.
13. The system of claim 9, wherein the processor further comprises
a results buffer allocater circuit, and the results buffer
allocater circuit sends a signal to the producer tracking circuit
to modify a particular producer tracking status value.
14. A tangible computer readable storage medium comprising a
processor embodied in software, the processor comprising: an
operand renamer circuit that stores a plurality of in-register
status values, wherein each in-register status value is associated
with a particular architectural state register; a producer tracking
circuit that stores a plurality of producer tracking status values,
wherein each producer tracking status value is associated with a
particular physical register; and control logic that modifies the
in-register status values based on the producer tracking status
values.
15. The tangible computer readable storage medium of claim 14,
wherein the operand renamer circuit includes a register rename
map.
16. The tangible computer readable storage medium of claim 14,
wherein the producer tracking circuit includes a producer tracking
map.
17. The tangible computer readable storage medium of claim 14,
wherein the operand renamer circuit sends a signal to the producer
tracking circuit to modify a particular producer tracking status
value.
18. The tangible computer readable storage medium of claim 14,
wherein the processor further comprises a results buffer allocater
circuit, and the results buffer allocater circuit sends a signal to
the producer tracking circuit to modify a particular producer
tracking status value.
19. The computer readable storage medium of claim 14, wherein the
processor is embodied in hardware description language
software.
20. The computer readable storage medium of claim 19, wherein the
processor is embodied in one of Verilog hardware description
language software and VHDL hardware description language software.
Description
CROSS REFERENCE TO RELATED APPLICATION
[0001] This application is related to commonly owned U.S. patent
application Ser. No. ______, titled "Method For Latest Producer
Tracking In An Out-Of-Order Processor, And Applications Thereof,"
filed on the same day herewith (Attorney Docket No. 1778.2370001),
which is incorporated herein by reference in its entirety.
FIELD OF THE INVENTION
[0002] The present invention relates generally to processors and
more particularly to processors having an out-of-order execution
pipeline.
BACKGROUND OF THE INVENTION
[0003] Reduced Instruction Set Computer (RISC) processors are well
known. RISC processors have instructions that facilitate the use of
a technique known as pipelining. Pipelining enables a processor to
work on different steps of an instruction at the same time and
thereby take advantage of parallelism that exists among the steps
needed to execute an instruction. As a result, a processor can
execute more instructions in a shorter period of time.
Additionally, modern Complex Instruction Set Computer (CISC)
processors often translate their instructions into micro-operations
(i.e., instructions similar to those of a RISC processor) prior to
execution to facilitate pipelining.
[0004] Many pipelined processors, especially those used in the
embedded market, are relatively simple in-order machines. As a
result, they are subject to control, structural, and data hazard
stalls. More complex processors have out-of-order execution
pipelines. These more complex processors, often referred to as
out-of-order processors, schedule execution of instructions around
hazards that would stall an in-order machine.
[0005] Register renaming is a technique used by out-of-order
processors to avoid unnecessary serialization of program operations
imposed by the reuse of logical registers. In a conventional
out-of-order processor, register renaming is implemented using a
custom content-addressable memory (CAM) that holds a register map.
The register map identifies associations formed between physical
registers and logical registers. The CAM register map is searched,
for example, during instruction decode and dispatch operations to
identify physical registers that hold the latest results for source
logical registers specified by an instruction.
[0006] In a conventional out-of-order processor, other register
status information such as, for example, information that indicates
whether register data is available in a register file or off a
bypass is also maintained in a custom CAM. While custom CAMs and
conventional out-of-order processing techniques work for their
intended purposes, they are costly to implement in terms of chip
area, power consumption, and processing speed. As a result,
especially in the embedded market, the number of applications in
which a conventional out-of-order processor may be used is
restricted.
[0007] What are needed are new techniques for implementing
out-of-order processing that overcome the limitations associated
with conventional techniques.
BRIEF SUMMARY OF THE INVENTION
[0008] The present invention provides a processor and system for
latest producer tracking, and applications thereof. In one
embodiment, the processor includes an operand renamer circuit that
includes a register rename map, a producer tracking circuit that
includes a producer tracking map, and a results buffer allocater
circuit that includes a results buffer free list.
[0009] The register rename map associates particular physical
registers of a results buffer with particular logical/architectural
state registers of a register file. The register rename map is
indexed using register identification (RID) values. Each RID value
represents a logical/architectural state register of the register
file. The register rename map stores buffer identification (BID)
values and in-register (INR) status values. Each BID value
represents a physical register of a results buffer. The INR values
are used to determine whether particular data values are available
in a logical/architectural state register of the register file or
in a physical register of the results buffer.
[0010] The producer tracking map stores producer tracking status
values. These status values are used to identify which physical
registers of the results buffer are being used by instructions to
store the latest data prior to the data being transferred to
logical/architectural state registers of the register file. The
producer tracking status values stored in the producer tracking map
are modified in one embodiment by placing BID values produced by
the results buffer allocater circuit on a BID set bus or a BID
clear bus of the producer tracking circuit.
[0011] The results buffer free list stores status values that
identify which physical registers of the results buffer are
available to store a value produced by an instruction. Instructions
that produce values are assigned physical registers in which their
results can be stored until instruction graduation. The function of
the results buffer allocater circuit is to keep track of physical
register availability and to output a BID value representing a
physical register, which can be assigned to an instruction and used
to store the value produced by the instruction.
[0012] Control logic modifies the INR status values stored in the
register rename map based on the producer tracking status values
stored in the producer tracking map.
[0013] Further embodiments, features, and advantages of the present
invention, as well as the structure and operation of the various
embodiments of the present invention, are described in detail below
with reference to the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS/FIGURES
[0014] The accompanying drawings, which are incorporated herein and
form a part of the specification, illustrate the present invention
and, together with the description, further serve to explain the
principles of the invention and to enable a person skilled in the
pertinent art to make and use the invention.
[0015] FIG. 1 is a diagram of a processor according to an
embodiment of the present invention.
[0016] FIG. 2 is a more detailed diagram of the processor of FIG.
1.
[0017] FIG. 3 is a diagram illustrating a relationship between a
producer tracking map, a register rename map, and a results buffer
free list of a processor according to an embodiment of the present
invention.
[0018] FIG. 4 is a diagram that illustrates clearing a status bit
of a producer tracking map according to an embodiment of the
present invention.
[0019] FIG. 5 is a diagram that illustrates setting a status bit of
a producer tracking map and updating a register rename map
according to an embodiment of the present invention.
[0020] FIG. 6 is a diagram that illustrates updating a status bit
of a register rename map according to an embodiment of the present
invention.
[0021] FIG. 7 is a diagram that illustrates operation of a
processor according to an embodiment of the present invention.
[0022] FIG. 8 is a diagram of an example system embodiment of the
present invention.
[0023] The present invention is described with reference to the
accompanying drawings. The drawing in which an element first
appears is typically indicated by the leftmost digit or digits in
the corresponding reference number.
DETAILED DESCRIPTION OF THE INVENTION
[0024] The present invention provides a processor, system, and
method for latest producer tracking in a processor, and
applications thereof. In the detailed description of the invention
that follows, references to "one embodiment", "an embodiment", "an
example embodiment", etc., indicate that the embodiment described
may include a particular feature, structure, or characteristic, but
every embodiment may not necessarily include the particular
feature, structure, or characteristic. Moreover, such phrases are
not necessarily referring to the same embodiment. Further, when a
particular feature, structure, or characteristic is described in
connection with an embodiment, it is submitted that it is within
the knowledge of one skilled in the art to effect such feature,
structure, or characteristic in connection with other embodiments
whether or not explicitly described.
[0025] FIG. 1 illustrates an example processor 100 according to an
embodiment of the present invention. As shown in FIG. 1, processor
100 includes an instruction fetch unit 102, an instruction cache
104, an instruction decode and dispatch unit 106, one or more
instruction execution unit(s) 108, a data cache 110, an instruction
graduation unit 112, a register file 114, and a bus interface unit
116. Processor 100 is preferably capable of implementing
multi-threading, but need not implement multi-threading. As used
herein, multi-threading refers to an ability of an operating system
and a processor to execute different parts of a program, called
threads, simultaneously.
[0026] Instruction fetch unit 102 retrieves instructions from
instruction cache 104 and provides instructions to instruction
decode and dispatch unit 106. Instructions are retrieved in program
order, for example, for one or more program threads. In one
embodiment, instruction fetch unit 102 includes logic for recoding
compressed format instructions to a format that can be decoded and
executed by processor 100. In one embodiment, instruction fetch
unit 102 includes an instruction buffer that enables instruction
fetch unit 102 to hold multiple instructions for multiple program
threads, which are ready for decoding, and to issue more than one
instruction at a time to instruction decode and dispatch unit
106.
[0027] Instruction cache 104 is an on-chip memory array organized
as a direct associative or multi-way set associative cache such as,
for example, a 2-way set associative cache or a 4-way set
associative cache. In one embodiment, instruction cache 104 is
virtually indexed and physically tagged, thereby allowing
virtual-to-physical address translations to occur in parallel with
cache accesses. Instruction cache 104 interfaces with instruction
fetch unit 102.
[0028] Instruction decode and dispatch unit 106 receives one or
more instructions at a time from instruction fetch unit 102 and
decodes them prior to execution. In one embodiment, instruction
decode and dispatch unit 106 receives at least one instruction for
each program thread being implemented during a particular clock
cycle. As described herein, the number of program threads being
implemented at any given point in time is variable. Decoded
instructions are stored in a decoded instruction buffer and issued
to instruction execution unit(s) 108, for example, after it is
determined that selected operands are available. Instructions can
be dispatched from instruction decode and dispatch unit 106 to
instruction execution unit(s) 108 out-of-program-order.
[0029] Instruction execution unit(s) 108 execute instructions
dispatched by instruction decode and dispatch unit 106. In one
embodiment, at least one instruction execution unit 108 implements
a load-store (RISC) architecture with single-cycle arithmetic logic
unit operations (e.g., logical, shift, add, subtract, etc.). Other
instruction execution unit(s) 108 can include, for example, a
floating point unit, a multiple-divide unit and/or other special
purpose co-processing units. In embodiments having multiple
instruction execution units 108, one or more of the units can be
implemented, for example, to operate in parallel. Instruction
execution unit(s) 108 interface with data cache 110, register file
114, and a results buffer (not shown).
[0030] Data cache 110 is an on-chip memory array. Data cache 110 is
preferably virtually indexed and physically tagged. Data cache 110
interfaces with instruction execution unit(s) 108.
[0031] Register file 114 represents a plurality of general purpose
registers (e.g., logical/architectural state register), which are
visible to a programmer. Each general purpose register is a 32-bit
or a 64-bit register, for example, used for logical and/or
mathematical operations and address calculations. In one
embodiment, register file 114 is part of instruction execution
unit(s) 108. Optionally, one or more additional register file sets
(not shown), such as shadow register file sets, can be included to
minimize content switching overhead, for example, during interrupt
and/or exception processing.
[0032] Bus interface unit 116 controls external interface signals
for processor 100. In one embodiment, bus interface unit 116
includes a collapsing write buffer used to merge write-through
transactions and gather writes from uncached stores. Processor 100
can include other features, and thus it is not limited to having
just the specific features described herein.
[0033] FIG. 2 is a more detailed diagram of processor 100. As
illustrated in FIG. 2, processor 100 performs four basic functions:
instruction fetch; instruction decode and dispatch; instruction
execution; and instruction graduation. These four basic functions
are illustrative and not intended to limit the present
invention.
[0034] Instruction fetch (represented in FIG. 1 by instruction
fetch unit 102) begins when a PC selector 202 selects amongst a
variety of program counter values and determines a value that is
used to fetch an instruction from instruction cache 104. In one
embodiment, the program counter value selected is the program
counter value of a new program thread, the next sequential program
counter value for an existing program thread, or a redirect program
counter value associated with a branch instruction or a jump
instruction. After each instruction is fetched, PC selector 202
selects a new value for the next instruction to be fetched.
[0035] During instruction fetch, tags associated with an
instruction to be fetched from instruction cache 104 are checked.
In one embodiment, the tags contain precode bits for each
instruction indicating instruction type. If these precode bits
indicate that an instruction is a control transfer instruction, a
branch history table is accessed and used to determine whether the
control transfer instruction is likely to branch or likely not to
branch.
[0036] In one embodiment, any compressed-format instructions that
are fetched are recoded by an optional instruction recoder 204 into
a format that can be decoded and executed by processor 100. For
example, in one embodiment in which processor 100 implements both
16-bit instructions and 32-bit instructions, any 16-bit
compressed-format instructions are recoded by instruction recoder
204 to form instructions having 32 bits. In another embodiment,
instruction recoder 204 recodes both 16-bit instructions and 32-bit
instructions to a format having more than 32 bits.
[0037] After optional recoding, instructions are written to an
instruction buffer 206. In one embodiment, this stage can be
bypassed and instructions can be dispatched directly to instruction
decoder 208.
[0038] Instruction decode and dispatch (represented in FIG. 1 by
instruction decode and dispatch unit 106) begins, for example, when
one or more instructions are received from instruction buffer 206
and decoded by an instruction decoder 208. In one embodiment,
following resolution of a branch misprediction, the ability to
receive instructions from instruction buffer 206 may be temporarily
halted until selected instructions residing within the instruction
execution portion and/or instruction graduation portion of
processor 100 are purged.
[0039] In parallel with instruction decoding, operands are renamed.
Register rename map(s) located within instruction identification
(ID) generator and operand renamer 210 are updated and used to
determine whether required source operands are available, for
example, in register file 114 and/or a results buffer 218. A
register rename map is a structure that holds the mapping
information between programmer visible architectural state
registers and internal physical registers of processor 100.
Register rename map(s) indicate whether data is available and where
data is available. As will be understood by persons skilled in the
relevant arts given the description herein, register renaming is
used to remove instruction output dependencies and to ensure that
there is a single producer of a given register in processor 100 at
any given time. Source registers are renamed so that data is
obtained from a producer at the earliest opportunity instead of
waiting for the processor's architectural state to be updated. In
parallel with instruction decoding, instruction ID generator and
operand renamer 210 generates and assigns an instruction ID tag to
each instruction. An instruction ID tag assigned to an instruction
is used, for example, to determine the program order of the
instruction relative to other instructions. In one embodiment, each
instruction ID tag is a thread-specific sequentially generated
value that uniquely determines the program order of instructions.
The instruction ID tags can be used to facilitate graduating
instructions in-program-order, which were executed
out-of-program-order.
[0040] Each decoded instruction is assigned a results buffer
identification value or tag by a results buffer allocater 212. The
results buffer identification value determines the location in
results buffer 218 (e.g., a physical register) where instruction
execution unit(s) 108 can write calculated results for an
instruction. In one embodiment, the assignment of results buffer
identification values are accomplished using a free list. The free
list contains as many entries as the number of entries (e.g.,
physical registers) that make up results buffer 218. The free list
can be implemented, for example, using a bitmap. A first bit of the
bitmap can be used to indicate whether the results buffer entry is
either available (e.g., if the bit has a value of one) or
unavailable (e.g., if the bit has a value of zero).
[0041] Assigned results buffer identification values are written
into a graduation buffer 224. In one embodiment, results buffer
completion bits associated with newly renamed instructions are
reset/cleared to indicate incomplete results. As instructions
complete execution, their corresponding results buffer completion
bits are set, thereby enabling the instructions to graduate and
release their associated results buffer identification values. In
one embodiment, control logic (not shown) ensures that one program
thread does not consume more than its share of results buffer
entries.
[0042] Decoded instructions are written to a decoded instruction
buffer 214. An instruction dispatcher 216 selects instructions
residing in decoded instruction buffer 214 for dispatch to
execution unit(s) 108. In embodiments, instructions can be
dispatched for execution out-of-program-order. In one embodiment,
instructions are selected and dispatched, for example, based on
their age (ID tags) assuming that their operands are determined to
be ready.
[0043] Instruction execution unit(s) 108 execute instructions as
they are dispatched. During execution, operand data is obtained as
appropriate from data cache 110, register file 114, and/or results
buffer 218. A multiplexer 215 and/or comparators (not shown) can be
used to select data from results buffer 218 or register file 114. A
result calculated by instruction execution unit(s) 108 for a
particular instruction is written to a location/entry of results
buffer 218 specified by the instruction's associated results buffer
identification value.
[0044] Instruction graduation (represented in FIG. 1 by instruction
graduation unit 112) is controlled by a graduation controller 220.
Graduation controller 220 graduates instructions in accordance with
the results buffer identification values stored in graduation
buffer 224. When an instruction graduates, its associated result is
transferred from results buffer 218 to register file 114. In
conjunction with instruction graduation, graduation controller 220
updates, for example, the free list of results buffer allocater 212
to indicate a change in availability status of the graduating
instruction's assigned results buffer identification value.
[0045] As illustrated in FIG. 3, in embodiments of the present
invention, processor 100 includes a producer tracking circuit 302,
an operand renamer circuit 305, and a results buffer allocater
circuit 307, which are interconnected and used to keep track of
instructions that are the latest producers of values to be stored
in particular registers of register file 114.
[0046] Producer tracking circuit 302 includes a producer tracking
map 303. Producer tracking map 303 stores producer tracking status
values that are used to identify which physical registers of
results buffer 218 are being used by instructions to store the
latest data for particular logical register. As shown in FIG. 3,
the "1" bits stored in producer tracking map 303 for physical
registers B1, B3, and B5 indicate that the instructions writing
their data to physical registers B1, B3, and B5 are the latest
producers of particular data values associated with particular
logical registers.
[0047] In an embodiment, producer tracking map 303 stores "N"
one-bit producer tracking status values, where "N" is the number of
physicals registers of results buffer 218. The "N" one-bit values
are indexed using buffer identification (BID) values associated
with the physical registers of results buffer 218. When a BID value
is placed on an address bus of producer tracking circuit 302, a
producer tracking status value corresponding to the BID value is
output at a read data bus of producer tracking circuit 302. As
shown in FIG. 3, this producer tracking status value is provided to
in-register status value set/clear (INR SET/CLR) logic 304.
Although INR SET/CLR logic 304 is illustrated as being separate
from producer tracking circuit 302 and operand renamer circuit 305,
it is to be understood that INR SET/CLR logic 304 can be
implemented, for example, as a part of producer tracking circuit
302 and/or a part of operand renamer circuit 305. Particular
producer tracking status values stored in producer tracking map 303
can be modified (e.g., set or cleared) by placing a BID value on a
BID set bus or a BID clear bus.
[0048] Operand renamer circuit 305 includes a register rename map
306. Register rename map 306 associates particular physical
registers of results buffer 218 with particular
logical/architectural state registers of register file 114.
[0049] In an embodiment, register rename map 306 is indexed using
register identification (RID) values. Each RID value represents one
of the logical/architectural state registers of register file 114.
As shown in FIG. 3, register rename map 306 stores, for example,
for each RID index value, a buffer identification (BID) value and
an in-register (INR) status value. Each BID value represents a
physical register of results buffer 218. The INR values are used to
determine whether particular data values are available in a
logical/architectural state register of register file 114 or in a
physical register of results buffer 218. Other values such as, for
example, a data availability (AVAIL) status value can also be
stored as part of register rename map 306 and indexed by RID
values. An AVAIL status value can be used, for example, to identify
whether an instruction can be dispatched.
[0050] In an embodiment, operand renamer circuit 305 preferably has
two read address busses, two read data busses, a write address bus,
and a write data bus, as illustrated in FIG. 3. Operand renamer
circuit 305 also has an old BID value read bus that is coupled to
the BID clear bus of producer tracking circuit 302.
[0051] Results buffer allocater circuit 307 includes a results
buffer free list 308. Results buffer free list 308 stores status
values that identify which physical registers of results buffer 218
are available to store a value produced by an instruction. In an
embodiment, results buffer free list 308 stores "N" one-bit status
values, where "N" is the number of physicals registers of results
buffer 218.
[0052] In the instruction decode and dispatch portion of the
pipeline of processor 100, instructions that produce values are
assigned physical registers in which their results can be stored
until instruction graduation. The function of results buffer
allocater circuit 307 is to output a BID value representing a
physical register, which can be assigned to an instruction and used
to store the value produced by the instruction. As shown in FIG. 3,
the BID value output by results buffer allocater circuit 307 is
provided to producer tracking circuit 302 and to operand renamer
circuit 305.
[0053] The operations and interactions of producer tracking circuit
302, operand renamer circuit 305, and results buffer allocater
circuit 307, as they relate to tracking an instruction (e.g., an
ADD instruction) that is the latest producer of a data value
associated with a particular logical register, will now be
described in detail with references to FIGS. 3-6.
[0054] Referring to FIG. 3, in an embodiment of the present
invention, an instruction to be decoded is stored in a processor
pipeline register 310. This occurs in the instruction decode and
dispatch portion of the pipeline of processor 100 (see FIGS. 1 and
2). Each instruction to be decoded may potentially include a first
group of bits 312 that specify a first logical/architectural state
register (Source 1), a second group of bits 314 that specify a
second logical/architectural state register (Source 2), and/or a
third group of bits 316 that specify a third logical/architectural
state register (Destination). These groups of bits, if present, are
provided to operand renamer circuit 305. As an example, consider an
ADD instruction such as "ADD (R3, R1, R2)," which implements
"R3=R1+R2." Such an instruction includes bits that identify the
Source 1 register as register R1, the Source 2 register as register
R2, and the Destination register as register R3.
[0055] Continuing further with the example ADD instruction noted
above, as shown in FIG. 3, the bits 312 of the example ADD
instruction, which represent register R1, are provided to a first
read address bus of operand renamer circuit 305. Bits 312 are used
as an index into register rename map 306. As shown in FIG. 3, the
bits representing register R1 index BID bits stored in register
rename map 306 that represent a physical register B3 of results
buffer 218. As a result of the bits 312 being placed on the first
read address bus, the bits representing physical register B3 are
placed on a first read data bus of operand renamer circuit 305. The
bits representing physical register B3 are then stored as bits 322
in a second processor pipeline register 320 in a subsequent clock
cycle of processor 100.
[0056] The bits 314 of the example ADD instruction, which represent
register R2, are provided to a second read address bus of operand
renamer circuit 305. Bits 314 are also used as an index into
register rename map 306. As shown in FIG. 3, the bits representing
register R2 index BID bits stored in register rename map 306 that
represent a physical register B8 of results buffer 218. As a result
of the bits 314 being placed on the second read address bus, the
bits representing physical register B8 are placed on a second read
data bus of operand renamer circuit 305. The bits representing
physical register B8 are then stored as bits 324 in the second
pipeline register 320 in a subsequent clock cycle of processor
100.
[0057] The bits 316 of the example ADD instruction, which represent
register R3, are provided to a write address bus of operand renamer
circuit 305. Bits 316 are used as an index into register rename map
306. As shown in FIG. 3, the bits representing register R3 index
BID bits stored in register rename map 306 that represent a
physical register B5 of results buffer 218. As a result of the bits
316 being placed on the write address bus, the bits representing
physical register B5 are placed on an old BID bus of operand
renamer circuit 305. This feature of the present invention is
illustrated in more detail in FIG. 4.
[0058] FIG. 4 is a diagram that illustrates the clearing of a
producer tracking status bit of producer tracking map 303 according
to an embodiment of the present invention. As shown in FIG. 4, bits
316 act as an index 402 into register rename map 306 of operand
renamer circuit 305. Index 402 points to a location 404 of register
rename map 306.
[0059] As shown in FIG. 4, location 404 stores bits that represent
physical register B5 of results buffer 218. The association of
physical register B5 with logical register R3, together with the
producer tacking status value stored at location 406 of producer
tracking map 303, indicates that prior to the example ADD
instruction noted above, physical register B5 was the physical
register used by an instruction that was the latest producer of a
particular data value associated with logical register R3. Because
this will no longer be the case (i.e., the example ADD instruction
will be the latest producer), the producer tracking status value
(i.e., the "1" bit stored at location 406 of producer tracking map
303) must be cleared. In an embodiment, clearing the "1" bit in
location 406 is accomplished by placing the bits representing
physical register B5 on the BID clear bus of producer tracking
circuit 302.
[0060] In parallel with clearing the bit at location 406 of
producer tracking map 303, a physical register "B1" is allocated by
results buffer allocater circuit 307 to hold the result of the
example ADD instruction until the example ADD instruction
graduates. As shown in FIG. 3, bits representing physical register
B5 are provided by results buffer allocater circuit 307 to pipeline
buffer 320, operand renamer circuit 305, and producer tracking
circuit 302. As shown in FIG. 3, these bits are stored in pipeline
register 320 as bits 326.
[0061] FIG. 5 is a diagram that illustrates how a BID value output
by results buffer allocater circuit 307 is used to set a producer
tracking status bit of producer tracking map 303 and to update
register rename map 306 according to an embodiment of the present
invention. As noted herein, the function of results buffer
allocater circuit 307 and results buffer free list 308 is to
identify which physical registers of the results buffer are
available to store a value produced by an instruction and to
allocate an available physical register to an instruction that
produces a value, for example, during instruction decode.
[0062] In the embodiment shown in FIG. 5, the example ADD
instruction includes bits 316, which indicate that the resultant
value of the ADD instruction is to be written to
logical/architectural state register R3. Before this happens,
however, the resultant value will first be temporarily stored in a
physical register of results buffer 218. Thus, results buffer
allocater circuit 307 must identify a physical register that is
available and communicate this information to operand renamer
circuit 305.
[0063] As illustrated in FIG. 5, results buffer free list 308 show
a "1 bit" associated with physical register B0. In an embodiment,
the "1 bit" indicates that physical register B0 is currently
assigned to an instruction that has not yet graduated, and thus
physical register B0 is unavailable. Results buffer free list 308
show a "0 bit" associated with physical register B1. The "0 bit"
indicates physical register B1 is available, and thus results
buffer allocater circuit 307 outputs bits (e.g., a BID value) that
are used to represent/identify physical register B1. The "0 bit"
associated with physical register B1 is then set to "1" to indicate
that physical register B1 is no longer available.
[0064] The BID value output by results buffer allocater circuit 307
is communicated to producer tracking circuit 302. In an embodiment,
the BID value is placed on a BID set bus, which causes a bit stored
at location 502 in producer tracking map 303 to be set to a value
of one. A value of one indicates that the instruction associated
with physical register B1 is the latest producer of a data value
(e.g., the resultant value of the ADD instruction that will be
written to register R3 of register file 114 upon graduation of the
ADD instruction). A value of zero stored in producer tracking map
303, which is indexed for example by a BID value representing
physical register B0, indicates that any instruction associated
with physical register B0 is not the latest producer of a value. In
an embodiment, during a cold reset of processor 100, all of the
producer tracking status values of producer tracking map 303 are
reset to zero.
[0065] The BID value output by results buffer allocater circuit 307
is also communicated to operand renamer circuit 305. As shown in
FIG. 5, in an embodiment, the BID value is placed on a write data
bus and written to location 404 of register rename map 305.
Location 404 is selected for storing the BID value as a result of
placing bits 316 on the write address bus of operand renamer
circuit 305. The INR bit stored at location 406 of register rename
map 306 is reset to zero to indicate that the resultant value of
the ADD instruction is not yet available in register R3 of register
file 114. As explained below, the zero bit stored at location 406
may or may not be set to one when the ADD instruction
graduates.
[0066] FIG. 6 is a diagram that illustrates the updating of an INR
status bit of register rename map 306 upon graduation of an
instruction according to an embodiment of the present invention. In
an out-of-order processor, instructions may be executed
out-of-program-order, but all instructions graduate and update the
architectural state of the processor in-program-order. This is
accomplished, for example, by transferring data associated with
instructions that are graduating in-program-order from the physical
registers of results buffer 218 to the logical/architectural state
registers of register file 114 as the instructions graduate.
[0067] When an instruction graduates, both the physical register
and the logical/architectural state register associated with a
producer instruction are known. Otherwise, the value produced by
the instruction could not be transferred from the physical register
of the results buffer to the logical/architectural state register
of the register file. This information is shown in FIG. 6 as BID
value 602 and RID value 604. During graduation of an instruction,
BID value 602 is placed on a BID read bus of producer tracking
circuit 302, and RID value 604 is placed on an RID read bus of
operand renamer circuit 305.
[0068] As shown in FIG. 6, when BID value 602 is placed on the BID
read bus of producer tracking circuit 302, a producer tracking
status value stored in producer tracking map 303 (e.g., at location
502) associated with BID value 602 is provided to IR SET/CLR logic
304. If the producer tracking value provided to INR SET/CLR logic
304 indicates that the graduating instruction is the latest
producer of a data value, INR SET/CLR logic 304 modifies the INR
status value (e.g., at location 406) in register rename map 306
indexed by RID value 604 (e.g., index 402) to indicate that the
data value is now available in register file 114 (e.g., in register
R3). If the producer tracking value provided to BIR SET/CLR logic
304, however, indicates that the instruction associated with the
graduating instruction is not the latest producer of a data value,
the INR status value is not modified to indicate that the data
value is available in register file 114. This feature of the
present invention is described further below with reference to FIG.
7.
[0069] FIG. 7 is a diagram that further illustrates operation of
processor 100 according to an embodiment of the present invention.
FIG. 7 shows how values stored in register rename map 306 and
producer tracking map 303 are modified based on two ADD
instructions over six time periods. The time periods each represent
multiple processor clock cycles, as can be understood by comparing
for example FIG. 2 and FIG. 7, in order to simplify the
description.
[0070] As shown in FIG. 7, at a time period To, a first ADD
instruction (ADD-1) is fetched from instruction cache 104.
Instruction ADD-1 implements the function "R3=R1+R2". As shown in
register rename map 306, at time period T.sub.0, operand R1 is
available register R1 of register file 114 (i.e., the INR status
bit equals one). Operand R2 is also available in register R2 of
register file 114 (i.e., the INR status bit equals one). The next
physical register available to store the resultant value produced
by instruction ADD-1 is assumed to be physical register B1. None of
the physical registers illustrated in producer tracking map 303 are
being used by an instruction that is the latest producer of a data
value (i.e., all of the shown producer tracking status values are
zero).
[0071] At a time period T1, instruction ADD-1 is undergoing
decoding and operand renaming. As shown in the register rename map
of FIG. 7, for time period T.sub.1, no operand renaming is required
for source 1 and source 2 of the instruction.
[0072] During time period T.sub.1, results buffer allocater circuit
307 allocates physical register B1 to store the resultant value of
instruction ADD-1 until such time as instruction ADD-1 graduates.
How this is accomplished is shown, for example, in FIG. 5. The BID
value associated with physical register B1 replaces the BID value
associated with physical register B12 in register rename map 306.
The appropriate INR status value is also reset to zero. In
addition, the producer tracking status value for physical register
B1 in producer tracking map 303 is set to one to indicate the
instruction writing to physical register B1 is the latest producer
of the value that will be written to register R3.
[0073] At a time period T.sub.2, instruction ADD-1 is executed and
generates a resultant value that is stored in physical register B1.
Also during time period T.sub.2, a second ADD instruction (ADD-2)
is fetched from instruction cache 104. Instruction ADD-2 implements
the function "R3=R1+R4". As shown in register rename map 306,
operand R1 and operand R2 are available in register R1 and register
R2, respectfully, of register file 114. The next physical register
available to store the resultant value produced by instruction
ADD-2 is assumed to be physical register B4.
[0074] During a time period T.sub.3, instruction ADD-2 is
undergoing decoding and operand renaming. Results buffer allocater
circuit 307 allocates physical register B4 to store the resultant
value of instruction ADD-2 until such time as instruction ADD-2
graduates. As a result, the BID value associated with physical
register B4 replaces the BID value associated with physical
register B1 in register rename map 306. The appropriate INR status
value is again reset to zero. Furthermore, the producer tracking
status value for physical register B1 in producer tracking map 303
is set to zero, and the producer tracking status value for physical
register B4 is set to one to indicate that the instruction writing
to physical register B4 (i.e., ADD-2) is now the latest producer of
the value that will be written to register R3.
[0075] As shown in FIG. 7, instruction ADD-1 graduates during time
period T.sub.3. During this period of time, the data value stored
in physical register B1 is written to logical/architectural
register R3. Also, in an embodiment of the present invention, as
illustrated for example in FIG. 6, the BID value "B1" is placed on
the BID read bus of producer tracking circuit 302. This causes the
producer tracking status value "0" that is stored in producer
tracking map 303 and associated with BID value "B1" to be provided
to INR SET/CLR logic 304. Because the producer tracking status
value "0" provided to INR SET/CLR logic 304 indicates that the
graduating instruction (ADD-1) is not the latest producer of the
data value that will be written to register R3, INR SET/CLR logic
304 does not modify the INR status value for register R3 in
register rename map 306 to indicate that the data value is
available in register file 114. If this were to occur, subsequent
instructions would use the value stored in register R3 rather than
the latest value stored in physical register B4 by instruction
ADD-2.
[0076] During a time period T.sub.4, instruction ADD-2 is executed
and generates a resultant value that is stored in physical register
B4.
[0077] During a time period T.sub.5, instruction ADD-2 graduates.
During this period of time, the data value stored in physical
register B4 is written to logical/architectural register R3. This
is accomplished as shown, for example, in FIG. 6. As described
herein, the BID value "B4" is placed on the BID read bus of
producer tracking circuit 302. This causes the producer tracking
status value "1" that is stored in producer tracking map 303 and
associated with BID value "B4" to be provided to INR SET/CLR logic
304. Because the producer tracking status value "1" provided to INR
SET/CLR logic 304 indicates that the graduating instruction (ADD-2)
is the latest producer of the data value that will be written to
register R3, INR SET/CLR logic 304 modifies the INR status value
for register R3 in register rename map 306 to indicate that the
data value is now available in register file 114. In addition, the
producer tracking status value for physical register B4 in producer
tracking map 303 is set to zero.
[0078] FIG. 8 is a diagram of an example system 800 according to an
embodiment of the present invention. System 800 includes a
processor 802, a memory 804, an input/output (I/O) controller 806,
a clock 808, and custom hardware 810. In an embodiment, system 800
is an application specific integrated circuit (ASIC) or a system on
a chip (SOC).
[0079] Processor 802 is any processor that includes features of the
present invention described herein and/or implements a method
embodiment of the present invention. In one embodiment, processor
802 includes an instruction fetch unit, an instruction cache, an
instruction decode and dispatch unit, one or more instruction
execution unit(s), a data cache, an instruction graduation unit, a
register file, and a bus interface unit similar to processor 100
described above.
[0080] Memory 804 can be any memory capable of storing instructions
and/or data. Memory 804 can include, for example, random access
memory and/or read-only memory.
[0081] Input/output (I/O) controller 806 is used to enable
components of system 800 to receive and/or send information to
peripheral devices. I/O controller 806 can include, for example, an
analog-to-digital converter and/or a digital-to-analog
converter.
[0082] Clock 808 is used to determine when sequential subsystems of
system 800 change state. For example, each time a clock signal of
clock 808 ticks, state registers of system 800 capture signals
generated by combinatorial logic. In an embodiment, the clock
signal of clock 808 can be varied. The clock signal can also be
divided, for example, before it is provided to selected components
of system 800.
[0083] Custom hardware 810 is any hardware added to system 800 to
tailor system 800 to a specific application. Custom hardware 810
can include, for example, hardware needed to decode audio and/or
video signals, accelerate graphics operations, and/or implement a
smart sensor. Persons skilled in the relevant arts will understand
how to implement custom hardware 810 to tailor system 800 to a
specific application.
[0084] While various embodiments of the present invention have been
described above, it should be understood that they have been
presented by way of example, and not limitation. It will be
apparent to persons skilled in the relevant computer arts that
various changes in form and detail can be made therein without
departing from the spirit and scope of the invention. Furthermore,
it should be appreciated that the detailed description of the
present invention provided herein, and not the summary and abstract
sections, is intended to be used to interpret the claims. The
summary and abstract sections may set forth one or more but not all
exemplary embodiments of the present invention as contemplated by
the inventors.
[0085] For example, in addition to implementations using hardware
(e.g., within or coupled to a Central Processing Unit ("CPU"),
microprocessor, microcontroller, digital signal processor,
processor core, System on Chip ("SOC"), or any other programmable
or electronic device), implementations may also be embodied in
software (e.g., computer readable code, program code, instructions
and/or data disposed in any form, such as source, object or machine
language) disposed, for example, in a computer usable (e.g.,
readable) medium configured to store the software. Such software
can enable, for example, the function, fabrication, modeling,
simulation, description, and/or testing of the apparatus and
methods described herein. For example, this can be accomplished
through the use of general programming languages (e.g., C, C++),
GDSII databases, hardware description languages (HDL) including
Verilog HDL, VHDL, SystemC Register Transfer Level (RTL) and so on,
or other available programs, databases, and/or circuit (i.e.,
schematic) capture tools. Such software can be disposed in any
known computer usable medium including semiconductor, magnetic
disk, optical disk (e.g., CD-ROM, DVD-ROM, etc.) and as a computer
data signal embodied in a computer usable (e.g., readable)
transmission medium (e.g., carrier wave or any other medium
including digital, optical, or analog-based medium). As such, the
software can be transmitted over communication networks including
the Internet and intranets.
[0086] It is understood that the apparatus and method embodiments
described herein may be included in a semiconductor intellectual
property core, such as a microprocessor core (e.g., embodied in
HDL) and transformed to hardware in the production of integrated
circuits. Additionally, the apparatus and methods described herein
may be embodied as a combination of hardware and software. Thus,
the present invention should not be limited by any of the
above-described exemplary embodiments, but should be defined only
in accordance with the following claims and their equivalence.
* * * * *