How to use chisel dsptools with floats - scala

Question

I need to convert a Float32 into a Chisel FixedPoint, perform some computation and convert back FixedPoint to Float32.
For example, I need the following:
val a = 3.1F
val b = 2.2F
val res = a * b // REPL returns res: Float 6.82
Now, I do this:
import chisel3.experimental.FixedPoint
val fp_tpe = FixedPoint(6.W, 2.BP)
val a_fix = a.Something (fp_tpe) // convert a to FixPoint
val b_fix = b.Something (fp_tpe) // convert b to FixPoint
val res_fix = a_fix * b_fix
val res0 = res_fix.Something (fp_tpe) // convert back to Float
As a result, I'd expect the delta to be in a range of , e.g
val eps = 1e-4
assert ( abs(res - res0) < eps, "The error is too big")
Who can provide a working example for Chisel3 FixedPoint class for the pseudocode above?

Answer 1

Take a look at the following code:
import chisel3._
import chisel3.core.FixedPoint
import dsptools._
class FPMultiplier extends Module {
val io = IO(new Bundle {
val a = Input(FixedPoint(6.W, binaryPoint = 2.BP))
val b = Input(FixedPoint(6.W, binaryPoint = 2.BP))
val c = Output(FixedPoint(12.W, binaryPoint = 4.BP))
})
io.c := io.a * io.b
}
class FPMultiplierTester(c: FPMultiplier) extends DspTester(c) {
//
// This will PASS, there is sufficient precision to model the inputs
//
poke(c.io.a, 3.25)
poke(c.io.b, 2.5)
step(1)
expect(c.io.c, 8.125)
//
// This will FAIL, there is not sufficient precision to model the inputs
// But this is only caught on output, this is likely the right approach
// because you can't really pass in wrong precision data in hardware.
//
poke(c.io.a, 3.1)
poke(c.io.b, 2.2)
step(1)
expect(c.io.c, 6.82)
}
object FPMultiplierMain {
def main(args: Array[String]): Unit = {
iotesters.Driver.execute(Array("-fiv"), () => new FPMultiplier) { c =>
new FPMultiplierTester(c)
}
}
}
I'd also suggest looking at ParameterizedAdder in dsptools, that gives you a feel of how to write hardware modules that you pass different types. Generally you start with DspReals, confirm the model then start experimenting/calculating with FixedPoint sizes that return results with the desired precision.

Answer 2

For others benefit, I provide an improved solution from #Chick, rewritten in a more abstract Scala with variable DSP tolerances.
package my_pkg
import chisel3._
import chisel3.core.{FixedPoint => FP}
import dsptools.{DspTester, DspTesterOptions, DspTesterOptionsManager}
class FPGenericIO (inType:FP, outType:FP) extends Bundle {
val a = Input(inType)
val b = Input(inType)
val c = Output(outType)
}
class FPMul (inType:FP, outType:FP) extends Module {
val io = IO(new FPGenericIO(inType, outType))
io.c := io.a * io.b
}
class FPMulTester(c: FPMul) extends DspTester(c) {
val uut = c.io
// This will PASS, there is sufficient precision to model the inputs
poke(uut.a, 3.25)
poke(uut.b, 2.5)
step(1)
expect(uut.c, 3.25*2.5)
// This will FAIL, if you won't increase tolerance, which is eps = 0.0 by default
poke(uut.a, 3.1)
poke(uut.b, 2.2)
step(1)
expect(uut.c, 3.1*2.2)
}
object FPUMain extends App {
val fpInType = FP(8.W, 4.BP)
val fpOutType = FP(12.W, 6.BP)
// Update default DspTester options and increase tolerance
val opts = new DspTesterOptionsManager {
dspTesterOptions = DspTesterOptions(
fixTolLSBs = 2,
genVerilogTb = false,
isVerbose = true
)
}
dsptools.Driver.execute (() => new FPMul(fpInType, fpOutType), opts) {
c => new FPMulTester(c)
}
}

Answer 3

Here's my ultimate DSP multiplier implementation, which should support both FixedPoint and DspComplex numbers. #ChickMarkley, how do I update this class to implement a Complex multiplication?
package my_pkg
import chisel3._
import dsptools.numbers.{Ring,DspComplex}
import dsptools.numbers.implicits._
import dsptools.{DspContext}
import chisel3.core.{FixedPoint => FP}
import dsptools.{DspTester, DspTesterOptions, DspTesterOptionsManager}
class FPGenericIO[A <: Data:Ring, B <: Data:Ring] (inType:A, outType:B) extends Bundle {
val a = Input(inType.cloneType)
val b = Input(inType.cloneType)
val c = Output(outType.cloneType)
override def cloneType = (new FPGenericIO(inType, outType)).asInstanceOf[this.type]
}
class FPMul[A <: Data:Ring, B <: Data:Ring] (inType:A, outType:B) extends Module {
val io = IO(new FPGenericIO(inType, outType))
DspContext.withNumMulPipes(3) {
io.c := io.a * io.b
}
}
class FPMulTester[A <: Data:Ring, B <: Data:Ring](c: FPMul[A,B]) extends DspTester(c) {
val uut = c.io
//
// This will PASS, there is sufficient precision to model the inputs
//
poke(uut.a, 3.25)
poke(uut.b, 2.5)
step(1)
expect(uut.c, 3.25*2.5)
//
// This will FAIL, there is not sufficient precision to model the inputs
// But this is only caught on output, this is likely the right approach
// because you can't really pass in wrong precision data in hardware.
//
poke(uut.a, 3.1)
poke(uut.b, 2.2)
step(1)
expect(uut.c, 3.1*2.2)
}
object FPUMain extends App {
val fpInType = FP(8.W, 4.BP)
val fpOutType = FP(12.W, 6.BP)
//val comp = DspComplex[Double] // How to declare a complex DSP type ?
val opts = new DspTesterOptionsManager {
dspTesterOptions = DspTesterOptions(
fixTolLSBs = 0,
genVerilogTb = false,
isVerbose = true
)
}
dsptools.Driver.execute (() => new FPMul(fpInType, fpOutType), opts) {
//dsptools.Driver.execute (() => new FPMul(comp, comp), opts) { // <-- this won't compile
c => new FPMulTester(c)
}
}