Introduction
13
Programming the MIPS32® 74K™ Core Family, Revision 02.14
in
and you will find a link to the register description (a hotlink if you’re reading on-screen, and a reference
including page number if you’re reading paper).
Register diagrams in this book are found in the list of figures. Register fields may show a background color, coded to
distinguish different types of fields:
Numeric values below the field diagram show the post-reset value for a field which is reset to a known value.
1.3 74K™ core features
All 74K family cores conform to Release 2 of the MIPS32 architecture. You may have the following options:
•
I- and D-Caches: 4-way set associative; I-cache may be 0 Kbytes, 16Kbytes, 32Kbytes or 64Kbytes in size. D-
cache may be 0 Kbytes, 16Kbytes, 32Kbytes or 64Kbytes in size. 32Kbyte caches are likely to be the most pop-
ular; 64Kbyte caches will involve some cost in frequency in most processes. The D-cache may even be entirely
omitted, when the system is fitted with high-speed memory on the cache interface (scratchpad RAM or SPRAM:
see
Section 3.6 “Scratchpad memory/SPRAM”
•
The caches are virtually indexed but physically tagged (the D-cache also keeps a virtual tag which is used to save
a little time, but the final hit/miss decision is always checked with the physical tag). Optionally (but usually) the
32K and 64K
2
D-cache configurations can be made free of cache aliases — see
Section 3.4.9, "Cache aliases"
which explains some software-visible effects. The option is selected when the “cache wrapper” was defined for
the 74K core in your design and shows up as the
Config7[AR]
bit. L2 (secondary) cache: you can configure your
74K core with MIPS Technologies’ L2 cache between 128Kbyte and 1Mbyte in size. Full details are in
, but pro-
gramming information is in
•
Fast multiplier: 1-per-clock repeat rate for 32
×
32 multiply and multiply/accumulate.
•
DSP ASE: this instruction set extension adds a lot of new computational instructions with a fixed-point math unit
crafted to speed up popular signal-processing algorithms, which form a large part of the computational load for
voice and imaging applications. Some of these functions do two math operations at once on two 16-bit values
held in one 32-bit register. 74K family cores support Revision 2 of the DSP ASE.
There’s a guide to the DSP ASE in
Chapter 7, “The MIPS32® DSP ASE” on page 87
and the full manual is
.
•
Floating point unit (FPU): if fitted, this is a 64-bit unit (with 64-bit load/store operations), which most often runs
at half or two-thirds the clock rate of the integer unit (you can build the system to run the FPU at the same clock
rate as the integer core, but it will then limit the speed of the whole CPU).
•
The “CorExtend®” instruction set extension: is available on all 74K CPUs.
interface which makes it relatively straightforward to add logic to implement new computational (register-to-reg-
ister) instructions in your CPU, using predefined instruction encodings. It’s matched by a set of software tools
read-write
read-only
reserved,
always zero
unused
software-only
write has
unusual effect.
2.
Note that a 4-way set associative cache of 16Kbyte or less (assuming a 4Kbyte minimum page size) can’t suffer
from aliases
.
Summary of Contents for MIPS32 74K Series
Page 1: ...Document Number MD00541 Revision 02 14 March 30 2011 Programming the MIPS32 74K Core Family...
Page 10: ...Programming the MIPS32 74K Core Family Revision 02 14 10...
Page 54: ...3 8 The TLB and translation Programming the MIPS32 74K Core Family Revision 02 14 54...
Page 83: ......
Page 101: ...The MIPS32 DSP ASE 101 Programming the MIPS32 74K Core Family Revision 02 14...
Page 134: ...8 4 Performance counters Programming the MIPS32 74K Core Family Revision 02 14 134...