-
Notifications
You must be signed in to change notification settings - Fork 1
/
response.txt
55 lines (42 loc) · 3.3 KB
/
response.txt
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
Attached please find a revised version of our extended abstract accepted for presentation at PAW17.
Reviewer 1
----------
1. Comment: "[2] Please cite the OpenSHMEM 1.3 specification (Available at: http://openshmem.org)"
Response: The specification is now reference 10.
2. Comment: "Figure 1: The animation is probably not a good idea for the paper. Why not show a few frames instead?"
Response: Figure 1 now depicts one representative frame. Space does not permit multiple frames.
3. Comment: "Figures 2-: The 'X' marks are indistinguishable when printed. Please use different symbols.
Please also show error bars rather than plotting over the same point multiple times."
Response: The symbols have been changed to better differentiate distinct runs. We did not perform a sufficient
number of runs to estimate error bars and the different data points are in fact distinct. For example,
the close correspondence between the MPI_put and OSH_put points in Figure 2 stems results from separate
runs in each case and likely relates to commonalities in the underlying hardware or in the MPI and
OpenSHMEM implementations.
Comment: "What about the Intel Fortran compiler? Seems like this compiler should be used on KNL, and it should
support coarrays?"
Response: Section 1.2 now details our experiences with the Intel compiler, which produces executable programs
that crash or hang beyond small core counts.
Review 2
--------
1. Comment (a): "The animation in Figure 1 depicts atmospheric processes over North 105 America as simulated by ICAR".
Which animation? I would suggest take out animation (whatever that is) and put just a figure or snapshots.
Response: Figure 1 now depicts one representative frame. Space does not permit multiple frames.
2. Comment (b): "Gets: an image that needs data proactively gets the data from memory managed by a providing image
without the providing image’s active involvement,": what is an image?
I know what it is, but it is not defined anywhere before this. A reader that knows nothing about CAF
should at least be able to read what images are.
Response: The first paragraph in Section 1 now gives a brief overview of the CAF programming model, including the meaning of "image."
3. Comment (c): "Figure 3 (top) depicts results across multiple compilers, machines, and architectures.
In this case, only “puts” are tested, and only for the 2000x2000 domain.": why only puts? need to justify
this more.
4. Response: The second paragraph of Section 3 now explains the reasons only "puts" are shown for MPI.
Comment (d): in figure 3, lower right: title says runtime comparison, but there is no comparison.
Response: The word "comparison" has now been deleted.
5. Comment (e): line 392 "This work as funded in..." typo. "This work was funded in"
Response: This typographical error has been corrected.
Review 3
--------
Comment: "1.2 spelling of "alteranative" needs to be alternative."
Response: The spelling has been corrected.
Comment: "2.2 The sentence 'We compiled at NCAR using of the GCC...." is poorly worded. Restructure the whole sentence
Response: The cited sentence has been resphrased for greater clarity.