DPFRO — Data Cache Prefetch for Read Once

Instruction Word

Slot
Inst
6
3
6
2
6
1
6
0
5
9
5
8
5
7
5
6
5
5
5
4
5
3
5
2
5
1
5
0
4
9
4
8
4
7
4
6
4
5
4
4
4
3
4
2
4
1
4
0
3
9
3
8
3
7
3
6
3
5
3
4
3
3
3
2
3
1
3
0
2
9
2
8
2
7
2
6
2
5
2
4
2
3
2
2
2
1
2
0
1
9
1
8
1
7
1
6
1
5
1
4
1
3
1
2
1
1
1
0
9876543210
Format x24 - 24 bit(s) 0
DPFRO 0111 00100010
s 3210
imm8 76543210

Assembler Syntax

DPFRO as, 0..1020

C Syntax

#include <xtensa/tie/xt_datacache.h>

extern void XT_DPFRO(const int * s, immediate i);

Description

(please consult the Xtensa ® Instruction Set Architecture Reference Manual for any cross references and additional information)

DPFRO performs a data cache prefetch for read once. The purpose of DPFRO is to improve performance, but not to affect state defined by the ISA. Therefore, some Xtensa ISA implementations may choose to implement this instruction as a simple "no-operation" instruction. In general, the performance improvement from using this instruction is implementation-dependent. Refer to a specific Xtensa Microprocessor Data Book for more details.

In some Xtensa ISA implementations, DPFRO checks whether the line containing the specified address is present in the data cache, and if not, it begins the transfer of the line from memory. Four data prefetch instructions provide different "hints" about how the data is likely to be used in the future. DPFRO indicates that the data is only likely to be read once before it is replaced by another line in the cache. In some implementations, this hint might be used to select a specific cache way or to select a streaming buffer instead of the cache.

DPFRO forms a virtual address by adding the contents of address register as and an 8-bit zero-extended constant value encoded in the instruction word shifted left by two. Therefore, the offset can specify multiples of four from zero to 1020. If the Region Translation Option or the MMU Option is enabled, the virtual address is translated to the physical address. If not, the physical address is identical to the virtual address. No exception is raised on either translation or memory reference. Instead of raising an exception, the prefetch is dropped and the instruction becomes a nop.

Because the organization of caches is implementation-specific, the operation section below specifies only a call to the implementation's dprefetch function.

Operation

vAddr ← AR[s] + (022||imm8||02)
(pAddr, attributes, cause) ← ltranslate(vAddr, CRING)
if not invalid(attributes) then
	dprefetch(vAddr, pAddr, 0, 1, 0)
endif

Exceptions

EveryInstR Group (see EveryInstR Group:)

Protos that use DPFRO

proto DPFRO { in const int32 * s, in immediate i }{}{
DPFRO s, i + 0;
}