Commit Graph

742 Commits

Author SHA1 Message Date
Thanh C. Tran
fefc2a6646 Remove unneccesary “throw” 2016-11-22 15:36:50 +01:00
Thanh C. Tran
43ac482677 Modified report message 2016-11-22 15:07:47 +01:00
Thanh C. Tran
93bcda3c72 Add compatible assertion between Store Mode and Power Down mode 2016-11-22 14:27:57 +01:00
Éder F. Zulian
5520aa2682 Continuous testing python files appear in qtcreator 2016-11-21 09:56:19 +01:00
Éder F. Zulian
aa6b7af216 Address mappings and specs that may be useful
Also specificaiton file for a SAMSUNG_K4B4G1646Q_4Gb_DDR3-1066_16bit.
Nevertheless, this spec file is not necessarily 100% correct. Use this file at
your own risk! Before using this file, it's strongly recommended that you
double check if the ALL currents provided are for a DIMM (made of 4 chips x16
in this case) or a Device (one single chip x16). Currents were measured.
Additionally, the original fck was 800 MHz (DDR3-1600) but then underclocked
to fck 533 MHz (DDR3-1066).

Other changes:
Avoid duplicate trace files in Sim. manager.
2016-11-19 17:25:36 +01:00
Éder F. Zulian
b0506239b0 Memory size can achieve values not supported by unsigned int. (small fix!!!)
Using std::uint64_t.
2016-11-18 01:25:17 +01:00
Éder F. Zulian
57058207a0 small fix 2016-11-15 15:05:48 +01:00
Éder F. Zulian
e8bd29fcb6 Debug messages improved (small changes) 2016-11-15 15:01:42 +01:00
Éder F. Zulian
8421814766 Fixes (small ones) 2016-11-15 14:57:18 +01:00
Éder F. Zulian
bc73bb70df New test: run with base config and compare output. 2016-11-15 14:42:01 +01:00
Éder F. Zulian
3d09195010 Saving changes to python scripts
These are intermediate changes, just pushing them to access from home.
2016-11-15 14:41:45 +01:00
Éder F. Zulian
ac31e075f6 Set the memory ID for DRAMPower.
It seems that this field is not used by DRAMPower currently, but I could not
find a good reason to keep it uninitialized.
2016-11-11 11:52:00 +01:00
Éder F. Zulian
94063a4861 Set the memory type for DRAMPower.
If the memory type is not set, then it's considered invalid by DRAMPower what
would result in a bug for RDA and WRA commands. DRAMPower breaks those
commands in two: RD + PRE or WR + PRE, then the PRE is shifted in time. The
time shift could be zero.
2016-11-10 21:59:37 +01:00
Éder F. Zulian
9fc226c83e Single device simulation files.
The files added can be used to simulate a single 1Gbx8 DDR3-SDRAM device.
For more details refer to the files.
2016-11-09 22:26:37 +01:00
fzeder
8e42de9b35 Disabling bank-wise logic because it's not fully implemented.
As soon we get power analysis + bank-wise logic working we can change the default value again.
2016-11-09 20:22:45 +01:00
Éder F. Zulian
dddac71f49 Assertions added.
For memory modules:
- The least significant bits of the physical address are the byte offset of
  the N-byte-wide memory module.
- A single data word has N bytes.
- N = 2^(# bits for byte offset).

Assert that the number of bytes per burst match the burst length times the
data word size in bytes.
2016-11-09 19:19:09 +01:00
Éder F. Zulian
803c2845c2 Function to get the number of bytes in a burst 2016-11-09 18:10:42 +01:00
Éder F. Zulian
478ce8d79f Function to get the width of the data bus. 2016-11-09 17:44:13 +01:00
Éder F. Zulian
f9555acc32 Function to get the mem. size in bytes 2016-11-09 17:10:50 +01:00
Éder F. Zulian
bc1a957a63 Directory renamed 2016-11-09 16:29:37 +01:00
Éder F. Zulian
de0caeb6fb Scripts renamed 2016-11-09 16:26:05 +01:00
Éder F. Zulian
3196bea1a4 Git ignore updated to ignore __pycache__ folders 2016-11-09 16:17:20 +01:00
Éder F. Zulian
5eb7336261 Easy way to change the standard output and error output 2016-11-09 16:16:43 +01:00
Éder F. Zulian
b38180384a Prevent wrong burst size calculation.
For the case of a single x4 device the value assigned to "bytesPerBurst" would
be zero because of integer division.
2016-11-08 18:35:34 +01:00
fzeder
8d7cd7db87 Merge pull request #139 from trancong/fix_hardcoded_memory_size
Fix hardcoded memory size (issue #128)
2016-11-08 18:00:28 +01:00
Thanh Tran
989d00204b Reimplement calculation of Channel Memory Size to more readable and safer way 2016-11-08 15:54:59 +01:00
Thanh Tran
ac8982aa39 Rename the data bit width of one DRAM chip to more significant one 2016-11-08 15:49:45 +01:00
Thanh Tran
79b624a88b Add assertion when parsing Number of Device on DIMM 2016-11-08 15:46:39 +01:00
Thanh Tran
701152e0af Recalculate memory size, that take into account of bus width & #chips on a DIMM 2016-11-08 00:04:00 +01:00
Thanh Tran
c28c996271 Fix hardcoded memory size (issue #128)
Recalculate memory size using #Banks, #Rows, and #Columns from memory specs.
2016-11-07 22:03:56 +01:00
Éder F. Zulian
561a688b0f Removal of unused variable (issue #140). 2016-11-07 19:11:55 +01:00
Éder F. Zulian
8b0108f366 ACT is also valid after PREA.
This bug was masked because until now we always had a REF after PREA. So, we
never had a PREA followed by an ACT.
2016-11-07 13:50:30 +01:00
fzeder
7811bdf923 Merge pull request #138 from trancong/updated_TLM2_checker
Update the TLM2 protocol checker to the newest version from Doulos
2016-11-05 02:11:37 +01:00
fzeder
2314c88e82 Merge pull request #137 from trancong/test_script_compatible_osx
Fix test scripts to support both OS X and Linux System
2016-11-05 01:46:33 +01:00
Thanh Tran
e47b6a616b Fix test scripts to support both OS X and Linux System
Problem:
	1. Test Script cannot determine number of logical cores on OS X with `cat /proc/cpuinfo | grep processor | wc -l`
	2. sed command cannot run on OS X, e.g. sed: 1: "../../DRAMSys/tests/TLM ...": invalid command code .

Solution
	1. Add logic to determine number of logical cores on Testing Host’s OS
	2. Fix sed command to support both OS X and Linux by adding -i flag suffix to add to the backed-up file
	3. Cleaning backup files
2016-11-05 00:48:28 +01:00
Thanh Tran
369f7153a9 Update the TLM2 protocol checker to the newest version from Doulos
Basically it changes:
	1. Update error messages refer to IEEE std 1666-2011
	2. Add DMI post check
	3. Add check for Generic Payload Option in various checks
	4. Minor bug fix
More information, take a look at its source code prelude comments
2016-11-04 22:04:41 +01:00
Éder F. Zulian
f02c7169eb Gitignore updated 2016-11-02 10:17:05 +01:00
Éder F. Zulian
ec0de7fafd Simple test updated 2016-10-28 19:45:26 +02:00
Éder F. Zulian
26e1f6ac37 Fix for pull-request #131 2016-10-28 19:40:10 +02:00
fzeder
959efef573 Merge pull request #131 from tranan/master
Separate TracePlayer from Simulation module
2016-10-28 18:38:59 +02:00
tranan
b54f9582bb Merge remote branch 'upstream/master' 2016-10-26 15:10:42 +02:00
Matthias Jung
050573fa9c Merge branch 'master' of git.rhrk.uni-kl.de:EIT-Wehn/dram.vp.system 2016-10-26 11:20:28 +02:00
Matthias Jung
d51ed243ee updated MSC from Matthias' Diss. 2016-10-26 11:19:55 +02:00
fzeder
ab2699b34a Merge pull request #127 from sprado/master
Fix for the minimum distance between commands
2016-10-25 21:52:25 +02:00
Tran Anh Quoc
39ce0ad51c Separate TracePlayer from Simulation module 2016-10-25 01:19:24 +02:00
fzeder
07c6918bec Update README.md 2016-10-25 00:08:03 +02:00
Éder F. Zulian
fe1eca682f DRAMPower submodule updated with specific fix for mac
A header file was missing.
2016-10-25 00:06:18 +02:00
fzeder
c85ed403f3 Update README.md 2016-10-24 21:54:03 +02:00
Éder F. Zulian
8d7bc890e5 DRAMPower updated.
This is the top of the master branch (5878beedbc9123f3cc41481a4bb7431c7bca728b).
It contains the new window based power calculation and bankwise features!
2016-10-24 20:41:06 +02:00
Felipe S. Prado
9e1e285de1 Merge remote branch 'upstream/master' 2016-10-24 20:19:04 +02:00