Hi,

This is a snippet of Exec, TLB and MemoryAccess traces generated for a very
simple test program

36679500: system.cpu.dtb: lookup 0x8ffc8, asn 0 -> hit ppn 0x49
36679500: system.ruby.cpu_ruby_ports-port1: Timing access caught for address
0x93dc0
36679500: system.ruby.cpu_ruby_ports-port1: Request found in 0 - 0x7ffffff
range
36681000: system.ruby.cpu_ruby_ports-port1: Hit callback needs response 1
36681000: system.physmem: Read of size 4 on address 0x93dc0 data 0xa
36681000: system.ruby.cpu_ruby_ports-port1: Sending packet back over port
36681000: system.ruby.cpu_ruby_ports-port1: Hit callback done!
36679500: system.cpu T0 : @main+48    : ldl        r1,16(r15)      : MemRead
:  D=0x000000000000000a A=0x11ff91dc0

1) Virtual and Physical Addresses

I looked into some posts related to this topic but didn't understand
completely.
Can someone explain what is 0x8ffc8?
tlb.cc does a virtual address to physical address mapping and this is what
is the virtual address it gets to translate, then what was the address
printed with cpu exec trace 0x11ff91dc0

2) Memory Accesses

Is the memory accessed even if there is a cache hit?
When I have multiple load instructions one after other to the same location,
for every such load instruction, an access to physical memory is printed in
the MemoryAccess trace. I was expecting after first load, data is available
in cache and so there wont be a memory access for the data. I am working
with caches and need to measure timing information and also check the
functionality of a modified cache protocol. How can I make sure that the
data hits in cache or there is a miss for a certain instruction?

3) Is there a way to track the protocol state changes happening for an
instruction in Ruby Memory system?

-- 
B Saratchandra Prasanth <http://www.cse.iitd.ernet.in/%7Emcs092094>
Graduate-student in Computer Science
Dept. of Computer Science & Engineering<http://www.cse.iitd.ac.in/index.shtml>
Indian Institute of Technology, Delhi <http://www.iitd.ac.in>
New Delhi - 110016
Ph. No. 9555918314
_______________________________________________
gem5-users mailing list
[email protected]
http://m5sim.org/cgi-bin/mailman/listinfo/gem5-users

Reply via email to