Uploaded by Abdulrhman Alshameri

Simics

advertisement
Peter S. Magnusson, Magnus Crhistensson, Jesper Eskilson,
Daniel Forsgren, Gustav Hallberg, Johan Högberg, Frederik
larsson, Anreas Moestedt.
Presented by Eduardo Cuervo

Simulation is an important step
◦ Research
◦ Evaluation
◦ Computer Design

Not enough to simulate only user level code
◦ Not accurate enough
◦ Need for Full System Simulation (slower)

Simulation must be able to interface with
detailed HW models timely.

Scope (What is modeled?)
◦ Full system
◦ User level

Level of abstraction
◦ Functional behavior (what)
◦ Timing behavior (when)

Realistic workloads
◦ Functional:Boot, run unmodified OS, benchmarks
◦ Timing: Support hardware engineering
◦ Fast enough to run real workloads



Full system simulator
Instruction-set level
Support for multiple
architectures
◦
◦
◦
◦
◦
◦

SPARC
Alpha
X86
Itanium
MIPS
ARM
Unmodified OS
support


Processor Models
Device Models
◦ Accurate enough for real drivers and firmware

Simics Central
◦ Creation of full-scale distributed system
◦ Router
◦ Communicates multiple Simics instances
 Multiple nodes of the same architecture per instance
◦ Synchronizes instances

Microprocessor design
◦
◦
◦
◦

Interaction with memory manager an scheduler
Approximate cache and I/O timing models
Scalable to full server workloads (TPC-C)
OoO Support, custom ROB but no pipeline
Memory studies
◦ Memory Spaces  Address Spaces
◦ Extendable with timing models

OS Development
◦ Implementation of very specific breakpoints

Debugging

Simics Central
◦ Synchronizes virtual time
◦ Simulation speed = speed of the slowest Simics
process

Configuration
◦ Object Oriented

Command Line Interface (CLI) and Scripting
◦ Built-in Python runtime environment
◦ Scripts tied to events (TLB misses, I/O operations)

Devices
◦ Timer, floppy, keyboard,mouse, DMA,interrupt,etc.
OBJECT cpu0 TYPE x86-hammer
{
freq_mhz: 3500
physical_memory: phys_mem0
}
OBJECT phys_mem0 TYPE memory-space
{
map: ((0xa0000, vga0, 1, 0,
0x20000),
(0x100000, mem0, 0, 0x100000,
0xff00000),
...
}
OBJECT con0 TYPE gfx-console
{
queue: cpu0
x-size: 720
y-size: 400
keyboard: kbd0
Mouse: kbd0
}
from sim_core import *
import conf
def break_handler(id):
if conf.cpu0.eax > conf.cpu0.ecx:
raise SimExc_Break
id = SIM_breakpoint(conf.phys_mem0,
Break_Physical, Break_Execute,
0x000f2501, 1, 0)
SIM_hap_register_callback(
“Core_Breakpoint” ,break_handler,
id)

HDL interface
◦ Link Simics to Verilog through C interface

Simics API
◦ Makes Simics extensible
◦ Write new device models, commands, routines

Memory
◦ Biggest performance challenge
◦ Simulator transaction cache


Speeds up loads, stores
and fetches
Pointers to simulated
memory
◦ Indexed by virtual address

No side effects on hit
◦ Alignment exception, TLB
miss, cache miss,
breakpoint



Interpreter cache
Hit inlined in the kernel
Most complex construct

Two event queues
◦ Step queue
 Triggered by program counter steps
◦ Time queue
 Resolution of a processor clock cycle

Mix of time-driven and event-driven
components




Specification
language: Sim Gen
Generates all
permitted
combinations
Better interpreter
than practical to
do manually
Outputs an
interpreter in C
// IA32/x86-64 add to left instruction
instruction ADD_L({REG}, {REG_OR_MEM})
pattern
op_h2 == 0 && opl == 0 && d == 1
&& opm == 0
syntax
“add {REG},{REG_OR_MEM}”
semantics
#{
ireg_t op1 = {REG};
ireg_t op2 = {REG_OR_MEM};
ireg_t dst = op1 + op2;
EFLAGS_ADD(dst,op1,op2,w,os);
SET({REG_W}, dst);
#}
attributes
type = IT_ALU

Os boot workloads
◦ Modeled with 7 processor architectures

Scalability shown on Ultra II Enterprise
Servers
◦ Increasing number of CPUs

Lower performance on OoO versions




IBM first emulator (7070)
PDP-11
G88
Gsim
◦ Based on g88
◦ Rewritten as the first version of simics

SimOS
◦
◦
◦
◦
MIPS-based processor
Similar goals and solutions
More general solution
Three CPU simulators
Full system simulation is required for
realistic workloads
 Simics offers a valuable simulation
tool for designing and evaluating HW
 Support for scripting, networking, and
multiple architectures are a great
advantage

Download