TC3xx架构详解：TriCore TM微控制器DSP的32位统一处理器核心

需积分: 2 88 浏览量更新于2024-06-17 1 收藏 12.76MB PDF 举报

"Infineon的AURIX TC3xx架构用户手册，v1.6.2版本，是关于该公司的TriCore TM单核处理器核心的详细设计文档。TriCore是一种专为实时嵌入式系统优化的32位微控制器-DSP（数字信号处理器）架构，它具有统一的指令集，旨在提供高效和实时性能。此手册分为两卷，第一卷侧重于核心架构和系统交互的详细介绍，而第二卷，即本手册，主要涵盖了完整的TriCore指令集，包括可选的内存管理单元（MMU）和浮点运算单元（FPU）扩展。手册的目标读者包括系统开发者、程序员以及硬件和软件工程师。作者强调，这份文档仅描述了架构本身，而不是特定实现，因为实际的产品可能包含不属于核心架构的额外特性或资源。在使用基于TriCore的实际产品时，应始终参考相应的支持文档，以获取所有产品特性的完整信息。手册还提到了TriCore架构的不同版本，说明了随着技术发展，该架构经历了多个迭代和升级。这意味着在学习和开发过程中，开发者需要了解他们所使用的具体版本特性和兼容性信息。阅读这份手册对于理解TriCore TC3xx架构的核心原理、指令集细节以及如何在实时嵌入式系统设计中有效利用其功能至关重要。通过结合第一卷的系统互动知识和第二卷的指令集扩展，开发者能够构建出高效且精确的解决方案。"

TriCore

TC1.6.2 core architecture manual - Instruction set

32-bit Unified Processor Core

Instruction Set Information

Table 10

Cache Functions

Function

Definition

cache_index_wb

if ((CACHE[index,elem].valid==1) AND (CACHE[index,elem].modified==1))

then {

pa = CACHE[index,elem].physical_address;

M[pa,cacheline] = CACHE[index,elem].data;

CACHE[index,elem].modified = 0;

}

cache_index_wi

if (CACHE[index,elem].valid==1) then {

if (CACHE[index,elem].modified==1) then {

pa = CACHE[index,elem].physical_address;

M[pa,cacheline] = CACHE[index,elem].data;

}

CACHE[index,elem].modified = 0;

CACHE[index,elem].valid = 0;

}

User Manual (Volume 2)

1-12

V1.2.2

2020-01-15

TriCore

TC1.6.2 core architecture manual - Instruction set

32-bit Unified Processor Core

Instruction Set Information

1.3.3

Floating Point Operation Syntax

The following table defines the floating point operation syntax.

Table 11

Floating Point Operation Syntax

Syntax

Definition

ADD_NAN

7FC00001

DIV_NAN

7FC00008

HP_MAX_VALUE

65504

HP_MIN_NORMAL

HP_NEG_INFINITY

FC00

HP_POS_INFINITY

7C00

MUL_NAN

7FC00002

NEG_INFINITY

FF800000

POS_INFINITY

7F800000

SQRT_NAN

7FC00004

add(x,y)

Adds the real value ‘x’ to the real value ‘y’ and returns an infinitely accurate

real result.

approx_inv_sqrt(x)

Takes the real argument x and returns the approximate inverse square root (x

0.5

) to at least 6.75 bits of precision.

denorm_to_zero(x)

If the IEEE-754-2008 32-bit single precision floating point format value ‘x’ is a

denormal, return the appropriately signed infinitely accurate real value 0.

Otherwise return ‘x’ as an infinitely accurate real value; i.e.

if ((x < 0) AND (x > -2

-126

)) then return -0.0;

else if ((x > 0) AND (x < 2

-126)

) then return +0.0;

else return f_real(x);

divide(x,y)

Divides the real value ‘x’ by the real value ‘y’ and returns an infinitely accurate

real result.

fp_abs(x)

Returns the infinitely accurate absolute value of the real value ‘x’; i.e. (x < 0.0)

? (0.0 - x) : x;

f_real_16bit(x)

Returns the IEEE-754-2008 16-bit half precision floating point format value ‘x’

as an infinitely accurate real value.

f_real(x)

Returns the IEEE-754-2008 32-bit single precision floating point format value

‘x’ as an infinitely accurate real value.

ieee754_round_16bit(f, m)

Rounds the real value ‘f’ to another real value which is exactly representable

by a 16-bit half precision format number using the type of rounding specified

by ‘m’, compliant with IEEE-754-2008. Note that the result may be

representable only as a denormal half precision value.

ieee754_16bit_format(f)

Returns the real value ‘f’ in the 16-bit IEEE-754-2008 half precision floating

point format. ‘f’ is converted to the correct IEEE-754-2008 result on overflow

or underflow. The return value may be a 16-bit half precision denormal value.

ieee754_16bit_overflow(f)

Takes a real value ‘f’ and returns the boolean result of the expression:

( | f | > HP_MAX_VALUE)

ieee754_16bit_underflow(f)

Takes a real value ‘f’ and returns the boolean result of the expression:

( | f | < HP_MIN_NORMAL & | f | > 0)

User Manual (Volume 2)

1-13

V1.2.2

2020-01-15

TriCore

TC1.6.2 core architecture manual - Instruction set

32-bit Unified Processor Core

Instruction Set Information

Table 11

Floating Point Operation Syntax

Syntax

Definition

ieee754_32bit_format(x)

Returns the real value ‘x’ in the standard 32-bit single precision IEEE-754-2008

floating point format. ‘x’ is converted to the correct IEEE-754-2008 result on

overflow or underflow.

ieee754_eq(x,y)

Returns TRUE if ‘x’ is equal to ‘y’ according to the IEEE-754-2008 rules for 32-

bit single precision floating point numbers otherwise returns FALSE.

ieee754_gt(x,y)

Returns TRUE if ‘x’ is greater than ‘y’ according to the IEEE-754-2008 rules for

32-bit single precision floating point numbers otherwise returns FALSE.

ieee754_lt(x,y)

Returns TRUE if ‘x’ is less than ‘y’ according to the IEEE-754-2008 rules for 32-

bit single precision floating point numbers otherwise returns FALSE.

ieee754_round(x,y)

Rounds the real value ‘x’ using the type of rounding specified by ‘y’ compliant

with IEEE-754-2008.

i_real(x)

Returns a infinitely accurate real number of equal value to the 32-bit signed

integer value ‘x’.

is_denorm(x)

Takes the IEEE-754-2008 32-bit single precision floating point format value ‘x’

and returns the boolean result of the expression: (x[30:23] == 0) AND (x[22:0]

!= 0);

is_inf(x)

Takes the IEEE-754-2008 32-bit single precision floating point format value ‘x’

and returns the boolean result of the expression: (is_neg_inf(x) OR

is_pos_inf(x));

is_inf_16bit(x)

Takes the IEEE-754-2008 16-bit half precision floating point format value ‘x’

and returns the boolean result of the expression:

(x[15:0] == HP_POS_INFINITY) OR (x[15:0] == HP_NEG_INFINITY)

is_nan(x)

Takes the IEEE-754-2008 32-bit single precision floating point format value ‘x’

and returns the boolean result of the expression: (is_s_nan(x) OR

is_q_nan(x));

is_nan_16bit(x)

Takes the IEEE-754-2008 16-bit half precision floating point format value ‘x’

and returns the boolean result of the expression:

(is_s_nan_16bit(x) OR is_q_nan_16bit(x))

is_neg_inf(x)

Takes the IEEE-754-2008 32-bit single precision floating point format value ‘x’

and returns the boolean result of the expression:

(x[31:0] == NEG_INFINITY);

is_pos_inf(x)

Takes the IEEE-754-2008 32-bit single precision floating point format value ‘x’

and returns the boolean result of the expression:

(x[31:0] == POS_INFINITY);

is_q_nan(x)

Takes the IEEE-754-2008 32-bit single precision floating point format value ‘x’

and returns the boolean result of the expression:

(x[30:23] == 8’b11111111) AND (x[22] == 1’b1);

is q_nan_16bit(x)

Takes the IEEE-754-2008 16-bit half precision floating point format value ‘x’

and returns the boolean result of the expression:

(x[14:10] == 11111

) AND (x[9]==1)

is_s_nan(x)

Takes the IEEE-754-2008 32-bit single precision floating point format value ‘x’

and returns the boolean result of the expression:

(x[30:23] == 8'b11111111) AND (x[22] == 1’b0) AND (x[21:0] != 0);

V1.2.2

User Manual (Volume 2)

1-14

2020-01-15

剩余481页未读，继续阅读

wacious

粉丝: 2
资源: 4

TC3xx架构详解：TriCore TM微控制器DSP的32位统一处理器核心

Infineon-AURIX_TC37x-UserManual-v01_00-EN

Infineon-AURIX_TC39x_UserManual-v01_0..pdf

Infineon-AURIX_TC3xx_Part2-UserManual-v1.5.0-2:2.pdf

Infineon-Ethernet-1-KIT-TC397-TFT-Training-v01-00-EN.pdf

Infineon-TLE9879-EvalKit-UserManual-v01-03-EN.pdf

Infineon-TC21x-TC22x-TC23x-UM-v01_01-EN.pdf

Infineon-TriBoardManual-TC3X7-UM-v01_02-EN.pdf

Infineon-DAVE-Quick-Start-DT-v01-00-EN

Infineon-TC23xAC_DS-DS-v01_00-EN.pdf

Infineon-BTS5180-2EKA-DS-v01_00-EN-1133136.pdf

最新资源