LLVM: A Compilation Framework for Lifelong Program Analysis & Transformation

• p. 11 Section 6, first paragraph

  mation gathered from programmers in the field

• p. 5 Section 3.1, first paragraph

  represent the end-user’s (not the developer’s) runs

• p. 11 Section 6, first paragraph

  idle-time optimization using profile infor-mation gathered from programmers in the field

• p. 1 Section 1, first paragraph

  using profile information collected from the end-user

Only the authors can say what is right here.

Judge's note: Page 11 image shows 'gathered from programmers in the field'. Capability #3 on page 2 says profiling is 'in the field so that it is representative of actual users'; Section 3.6 on page 7 says 'profile information gathered from end-user runs'; the same Conclusion paragraph uses 'programmers' for developers ('transparent to programmers'). The word is probably a slip for 'users' or 'programs'; the authors should confirm which.

Second check: another model agreed

• p. 2 Section 1, experimental results paragraph

  SPECINT 2000 C benchmarks

• p. 8 Section 4.1.1, text before Table 1

  Table 1 shows this statistic for the C benchmarks in

• p. 8 Table 1 rows

  177.mesa

• p. 8 Section 4.1.1, text introducing Table 1

  Table 1 shows this statistic for the C benchmarks in SPEC CPU2000

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: that 177.mesa, 179.art, 183.equake and 188.ammp belong to the floating-point part of SPEC CPU2000 and not to SPECint2000. From the files alone: page 2 prints 'SPECINT 2000', page 8 prints 'SPEC CPU2000', and the 68% is the mean of all 15 rows of Table 1 (recomputed as 68.04, matching the printed average). The two labels for the same benchmark set differ; the claim that 'SPECINT' is the wrong one relies on knowledge of the suite.

Second check: another model agreed

• p. 7 Section 4.1, first sentence

  We evaluate three important characteristics of the LLVM

• p. 7 Section 4, opening paragraph

  (a) the characteristics of the representation; (b) the speed of

• p. 9 Subsection heading

  4.1.4 How fast is LLVM?

Judge's note: Page 7: Section 4 lists (a) characteristics of the representation, (b) speed of whole-program analyses, (c) illustrative uses; Section 4.1 announces three questions. Page 9: 4.1.4 on speed sits under 4.1 'Representation Issues', and 4.2 is the applications. The count and structure do not match the announcement, but the authors may regard speed as a property of the representation, and no result is affected.

Second check: another model agreed

• p. 11 Section 5, paragraph on unified intermediate representations

  Architecture and language

• p. 11 Reference [4]

  The Architectural Neutral

Only the authors can say what is right here.

Judge's note: Rests on outside knowledge: which expansion of ANDF is the official name cannot be checked from the files. Viewed both crops: the body text and the reference entry do print the name differently. The inconsistency is internal and real, but either form could be a legitimate gloss, so only the authors can say which should stand.

Second check: another model agreed

• p. 4 Section 2.4, paragraph after Figure 1

  tees that the destructor for the Object object will be run

• p. 4 Figure 1

  AClass Obj;

• p. 4 Section 2.4, text after Figure 1

  the destructor for the Object object will be run

• p. 4 Sec 2.4 text after Figure 1

  destructor for the Object object will be run

Judge's note: Page 4 image shows 'Object' in monospace like an identifier. No identifier 'Object' appears in Figure 1 or Figure 2; the variable is Obj and the class is AClass. The evident fix is 'Obj'. The meaning is unaffected.

Second check: none (judge only)

• p. 4 Section 2.4, last paragraph of the page

  as allocating memory for exceptions5. Second, the runtime

• p. 4 Section 2.4, discussion of Figure 3

  such as allocating memory for exceptions5. Second, the runtime

Judge's note: Page 4 image, last paragraph of the right column. The sentence before 'Second,' is evidently the unlabelled first point and no 'First' appears in the paragraph. The meaning is recoverable, so this is cosmetic.

Second check: none (judge only)

• p. 4 Figure 2, comment under ExceptionLabel

  ; If unwind occurs, excecution continues

• p. 4 Figure 2 code comment

  excecution continues

Judge's note: The misspelling is visible in the Figure 2 code listing on the page 4 image.

Second check: none (judge only)

• p. 4 Section 2.4, paragraph after Figure 2

  a language-indepenent manner to the optimizer

• p. 4 Sec 2.4, para after Fig 2

  language-indepenent manner

• p. 4

  exposed in a language-indepenent manner to the optimizer

Judge's note: The misspelling is visible on the page 4 image, in the paragraph after Figure 2.

Second check: none (judge only)

• p. 4 Section 2.4, paragraph after Figure 2

  ception specifications, function try blocks, etc., and reqiring

• p. 4 Sec 2.4

  reqiring

• p. 4

  and reqiring complex semantics for the dynamic lifetime

Judge's note: The misspelling is visible on the page 4 image, in the paragraph after Figure 2.

Second check: none (judge only)

• p. 6 Section 3.1, Omniware sentence in the virtual-machines bullet

  a low-level represention that permits extensive static

• p. 6 Omniware bullet, Sec 3.1

  low-level represention

• p. 6

  low-level represention that permits extensive static

Judge's note: The misspelling is visible on the page 6 image, in the Omniware sentence at the top of the left column.

Second check: none (judge only)

• p. 7 Section 3.5, first paragraph

  Although that work is outside the scope if this paper

• p. 7 Sec 3.5

  outside the scope if this paper

Judge's note: Visible on the page 7 image, Section 3.5, first paragraph.

Second check: none (judge only)

• p. 8 Section 4.1.1, second paragraph

  DSA, including Automatic Pool Allocation [30]).

• p. 8 Section 4.1.1, second paragraph

  including Automatic Pool Allocation [30]).

• p. 8 Sec 4.1.1

  Automatic Pool Allocation [30]).

Judge's note: The page 8 image shows the stray ')' after [30]. The earlier '(DSA)' is closed on its own and the sentence has no other open parenthesis.

Second check: none (judge only)

• p. 12 Reference [12]

  on Prog. Lang. Design and Implemenation

• p. 12 Reference [12]

  Design and Implemenation

• p. 12 Refs [33], [12]

  Commmon Language Infrastructure

Judge's note: The crop of reference [12] on p. 12 clearly shows 'Implemenation' with the 't' missing; the text layer agrees. The 'Commmon' misspelling also mentioned in this cluster belongs to reference [33] and is judged under C018.

Second check: none (judge only)

• p. 12 Reference [33]

  Commmon Language Infrastructure

• p. 12

  Commmon Language Infrastructure, 2002.

Judge's note: The crop of reference [33] on p. 12 shows 'Commmon' with three m's; the text layer agrees.

Second check: none (judge only)

• p. 12 Reference [15]

  pages 13(4):451–490

Judge's note: The crop of reference [15] shows the stray 'pages' before '13(4):451–490'. Other journal entries visible in the crops, e.g. [35] '21(3):528–569' and [5] '33(5):301–312', use volume(issue):pages without the word. A bibliography formatting slip; the citation data themselves are not affected.

Second check: none (judge only)

• p. 12 Reference [39]

  Univ. of Illinois at Urbana-Champ aign

Judge's note: The crop of reference [39] shows a visible full-width gap in 'Urbana-Champ aign', and the text layer also has the space. Reference [31], seen in another crop, prints 'Urbana-Champaign' with no gap, so this is not a font or kerning effect.

Second check: none (judge only)