forked from uvahotspot/HotSpot
-
Notifications
You must be signed in to change notification settings - Fork 1
/
README-6.0
111 lines (89 loc) · 6.31 KB
/
README-6.0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
HotSpot-6.0 Changes
-------------------
Contact: Robert West, [email protected]
The changes documented in this release are partly sponsored by
Defense Advanced Research Projects Agency(DARPA), Microsystems Technology Office (MTO),
under contract no. HR0011-13-C-0022.
This release of HotSpot incorporates the following changes/improvements:
-----------------------------------------------------------------------------------------------------
1) Calibration and validation
-----------------------------------------------------------------------------------------------------
- Acknowledgements: Pritish R. Parida at IBM and Pradip Bose at IBM
Please refer to our technical report for more details.
-----------------------------------------------------------------------------------------------------
2) New Feature: Fast steady-state solver based on SuperLU.
-----------------------------------------------------------------------------------------------------
To speedup grid-mode's steady-state simulations, we implemented a new, direct solver based on SuperLU.
SuperLU is an open source library for the direct solution of large, sparse systems of
linear equations. Documentations, installation instructions and source code can be found here:
http://crd-legacy.lbl.gov/~xiaoye/SuperLU/
How to Use :
-----------
a. Download and build SuplerLU.
Note that SuperLU requires BLAS (Basic Linear Algebra Subprograms) library.
b. Modify Makefile parameter SuperLUroot
SuperLUroot = /path/to/your/build
c. Build HotSpot with option SUPERLU=1
>> make SUPERLU=1
d. Run HotSpot simulation. (Input and output files are same as before)
According to our internal tests, this new direct solver could be 10x-100x faster than
the iterative solver used by previous versions of HotSpot. However, due to its large memory usage,
direct solver will significantly slowdown when the problem size (grid-size, number of layers in a
3D chip) is large. For this reason, we recommend using direct solver in these two situations:
a. Grid size is relatively small. (e.g. no larger than 256x256 for a 4-layer 3D chip).
If the host machine's memory is small (<8GB), even smaller grid-size is preferred.
b. detailed_3D mode is enabled.
-----------------------------------------------------------------------------------------------------
3) New Feature: Support layers with non-uniform thermal resistivity and heat capacity.
-----------------------------------------------------------------------------------------------------
- By Ayse K. Coskun, Katsutoshi Kawakami, Daniel Rossell, and Samuel Howes at Boston University.
- Contact: [email protected]
- Acknowledgements: Tiansheng Zhang and Fulya Kaplan at Boston University; Yasuko Eckert and Wei Huang at AMD;
David Atienza, Mohamed M. Sabry, and Yusuf Leblebici at EPFL; and Tajana Rosing at UCSD.
This feature is contributed by Prof. Ayse Coskun's research group in Boston University.
It enables modeling for layers with blocks that have a heterogeneous set of thermal
resistivity and heat capacity values. All major related code revisions are commented with "BU_3D".
To enable this new feature in simulation, cmd line option 'detailed_3D' should be set to 'on'.
Currently, heterogeneous layers could only be modeled in grid-mode with .lcf file specified.
There is a slight change in the format of the floorplan file (.flp) that allows units in
the floorplan to have different heat capacitance and resistivity (see the example below).
Note that specific-heat and resistivity are optional. Leaving them blank will force HotSpot
to use layers' default RC values (specified in .lcf file) for that block.
# Line Format: <unit-name> <width> <height> <left-x> <bottom-y> [<specific-heat>] [<resistivity>]
Unit1 0.002000 0.002000 0.000000 0.000000 4e6 .02294
How to Run :
-----------
Get steady-state temperature map:
>> ./hotspot -c hotspot.config -f <filename> -p <powerfile> -steady_file <steadyfile> \
-model_type grid -detailed_3D on -grid_layer_file <lcffile>
Transient simulation with steady-state results as initial temperature:
>> ./hotspot -c hotspot.config -init_file <steadyfile> -f <filename> -p <powerfile> \
-o <ttracefile> -model_type grid -detailed_3D on -grid_layer_file <lcffile>
Notice that since .lcf is provided, the .flp file specified by -f option will be overridden.
More detailed examples and instructions are available in subdirectory 3D_testcase.
-----------------------------------------------------------------------------------------------------
4) Secondary heat transfer path compatible with 3D.
-----------------------------------------------------------------------------------------------------
Previous version does not support secondary heat transfer path when .lcf file is specified.
This version relaxes this constraint and now secondary heat transfer path can be modeled along
with 3D stacks.
-----------------------------------------------------------------------------------------------------
5) Script to plot 3D chip.
-----------------------------------------------------------------------------------------------------
This is also a contribution made by BU.
3Dfig.pl has similar functionality to tofig.pl but instead reads in an .lcf files and creates a .FIG
file for each floorplan listed in the .lcf file. Additionally, it will show a unit's resistivity and
capacitance if it specified in the floorplan file.
3Dfig.pl must be used with Unix filetypes. To convert from a DOS file to unix, use the command 'dos2unix'.
-usage: 3Dfig.pl [-a <area ratio>] [-f <fontsize>] [-s <nskip>] <file>
Reads in a layer configuration file and creates a .FIG file for each floorplan listed.
Requires that all referenced files are in the same directory as 3Dfig.pl.
[-a <area ratio>] -- approx page occupancy by figure (default 0.95)
[-f <fontsize>] -- font size to be used (default 10)
[-s <nskip>] -- no. of entries to be skipped in input (default 0)
<file> -- input .lcf file (eg: example.lcf)
Alternatively, the command to convert 'example.lcf' into a PDF is 3Dfig.pl example.lcf | 3Dfig2pdf.sh output.pdf
-----------------------------------------------------------------------------------------------------
6) Discontinue the support for MS EXCEL interface
-----------------------------------------------------------------------------------------------------
Starting from version 6.0, we no longer maintain the Excel interface.