3.7. Floating-Point Functions IP Parameters
These tables list the Floating-Point Functions Intel® FPGA IP parameters.
Parameter | Values | Descriptions |
---|---|---|
Function | ||
Family |
|
Allows you to choose which functions to be displayed in the Function Name Parameter list. The default value is All. |
Name |
|
Allows you to choose your desired function.
Note: This parameter only displays the options you have selected from the Family Parameter.
|
Use Select Signal | On or off | Select this option to generate a Select signal. Use the Select signal to choose the option to use both addition and subtraction functions or either one of the functions. |
Represent angle as multiple of Pi | On or off | Select this option to represent angles as multiple of Pi.
Note: Not available for Arctan2 function.
|
Inputs are within range -2pi to +2pi | On or off | Select this option to disable range reduction.
Note: Only available for Sin and Cos function.
|
Floating Point Data | ||
Format | Single Double Custom |
Allows you to choose the floating point format of the data values. The default value is single. |
Exponent | 5 to 11 | Allows you to specify the width of the exponent. This parameter is only available when the Format parameter is set to custom. The default value is 8. |
Mantissa | 10 to 52 | Allows you to specify the width of the mantissa. This parameter is only available when the Format parameter is set to custom. The default value is 23. |
Input Vector Dimension | Integer | Provide the desired the number of inputs to compute the vector dimension. |
Input Format | Single Double Custom |
Allows you to choose the floating point format of the input data values. The default value is single.
Note: Only available for Floating to Floating Point function.
|
Input Exponent | Integer | Allows you to specify the width of the input exponent. This parameter is only available when the Format parameter is set to custom. The default value is 8.
Note: Only available for Floating to Floating Point function.
|
Input Mantissa | Integer | Allows you to specify the width of the mantissa. This parameter is only available when the Format parameter is set to custom. The default value is 23.
Note: Only available for Floating to Floating Point function.
|
Output Format | Single Double Custom |
Allows you to choose the floating point format of the output data values. The default value is single. |
Output Exponent | Integer | Allows you to specify the width of the output exponent. This parameter is only available when the Format parameter is set to custom. The default value is 8.
Note: Only available for Floating to Floating Point function.
|
Output Mantissa | Integer | Allows you to specify the width of the mantissa. This parameter is only available when the Format parameter is set to custom. The default value is 23.
Note: Only available for Floating to Floating Point function.
|
Fixed Point Data | ||
Width | 16 to 128 | The bit width of the fixed point data port. This parameter is only available when the Name parameter is set to Fixed to Floating Point. The default value is 32. |
Fraction | -128 to 128 | The bit width of the fraction. This parameter is only available when the Name parameter is set to Fixed to Floating Point. |
Sign | Signed Unsigned |
Choose if the fixed point data is signed or unsigned. This parameter is only available when the Name parameter is set to Convert. The default value is signed. |
Rounding | ||
Mode | nearest with tie breaking to even | The rounding mode. |
Relax rounding to round up or down to reduce resource usage | — | Choose if the nearest rounding mode should be relaxed to faithful rounding, where the result may be rounded up or down, to reduce resource usage. Only available for arithmetic functions |
Ports | ||
Generate Enable Port | — | Choose if the FP_FUNCTIONS Intel® FPGA IP core should have an enable signal. |
Parameter | Values | Descriptions |
---|---|---|
Target | ||
Goal |
|
If the Goal is the frequency, then the Target is the desired frequency in MHz. This, together with the target device family, determines the amount of pipelining. If the Goal is Combined then two Targets are displayed, one is the desired frequency in MHz, one is the target latency in cycles. When you set the Goal parameter to frequency, the default value is 200 MHz When you set the Goal parameter to latency, the default value is 2. If the Goal is Latency, then the Target is the desired latency. The report generates the achievable latency if it can't meet target latency. If the Goal is set to Manually Specify DSP Registers, you can manually select the register and function subblocks within the DSP IP core. |
Target | Any Positive Integer | Specify your target frequency and latency. |
Report | ||
Latency on Arria 10 is <x> cycles | — | This report shows the latency of the function. |
Resource Estimates:
|
— | This report shows the number of multipliers, LUTs, memory bits, and memory blocks utilized by the IP core. |
Check Performance | — | Click this to check if the design can achieve the target latency.
Note: Only available when Goal is set to Latency.
|