Bach­e­lor’s The­sis / Stu­dent Re­search Pro­ject

Ab­stract

The Abstract Com­puter Archi­tec­ture Des­crip­tion Lan­guage (ACADL) al­lows for fast mod­el­ing and per­for­mance eval­u­a­tion of DNN Hard­ware Ac­cel­er­a­tors. Cur­rently, the func­tional sim­u­la­tion of mapped DNNs is done in Python, which al­lows for easy use but comes with the draw­back of a slow sim­u­la­tion.

This stu­dent pro­ject’s goal is to au­to­mat­i­cally trans­late the be­hav­ior of an ACADL in­struc­tion in Python into C++ to speed up the func­tional sim­u­la­tion using the Python ab­stract syn­tax tree (AST).

Python be­hav­ior de­scrip­tion of the ADD In­struc­tion in add_function:

class ADD(Instruction):
    def __init__(self, id: int, rs1: str, rs2: str, rd: str):
        def add_function(architectural_state: ArchitecturalState):
            rs1 = self.read_registers[0]
            rs2 = self.read_registers[1]
            rd = self.write_registers[0]

            rs1_value = architectural_state.read_register(rs1)
            rs2_value = architectural_state.read_register(rs2)
            rd_value = rs1_value + rs2_value
            architectural_state.write_register(rd, rd_value)

            pc = architectural_state.read_register("pc0")
            architectural_state.write_register("pc0", pc + int(self.size / 8))

Trans­la­tion into C++:

void add_function(std::shared_ptr<Instruction> self, std::shared_ptr<ArchitecturalState> ag) {
    auto rs1 = self->read_registers[0];
    auto rs1_value = ag->read_register(rs1);

    auto rs2 = self->read_registers[1];
    auto rs2_value = ag->read_register(rs2);

    auto rd = self->write_registers[0];
    auto rd_value = rs1_value + rs2_value;

    ag->write_register(rd, rd_value);

    uint64_t pc = ag->read_register("pc0");
    pc += 4;
    ag->write_register("pc0", pc);
}

Ref­er­ences

• Mika Markus Müller, Alexan­der Richard Man­fred Borst, Kon­stan­tin Lübeck, Alexan­der Louis-Fer­di­nand Jung, Felix Wedlich, Oliver Bring­mann - Using the Ab­stract Com­puter Ar­chi­tec­ture De­scrip­tion Lan­guage to Model AI Hard­ware Ac­cel­er­a­tors
• Kon­stan­tin Lübeck, Alexan­der Louis-Fer­di­nand Jung, Felix Wedlich, Oliver Bring­mann - Work-in-Progress: Ul­tra-fast yet Ac­cu­rate Per­for­mance Pre­dic­tion for Deep Neural Net­work Ac­cel­er­a­tors

Re­quire­ments

• Python
• C++
• Suc­cess­fully at­teded the lec­ture “Grund­la­gen der Rech­ner­ar­chitek­tur” and/or “Par­al­lele Rech­ner­ar­chitek­turen” (op­tional)
• Linux (op­tional)

Con­tact

Lübeck, Kon­stan­tin

Bring­mann, Oliver