Threads Cannot Be Implemented as a Library

• p. 6 Section 5.1, last paragraph of right column

  since the number of threads is less than the number of hardware threads

• p. 6 Section 5.1, description of the Pentium 4 machine

  2 processors with 2 threads each

• p. 6 Section 5.1, description of the Pentium 4 machine

  2GHz, 2 processors with 2 threads each

• p. 6 text on Figure 4

  since the number of

Judge's note: Page 6 gives the Pentium 4 machine as 2 processors with 2 threads each, i.e. 4 hardware threads, and the experiments use 1, 2 and 4 threads (the next sentence says "even with 4 threads"). "No more than" would be accurate. I found no statement elsewhere in the paper that changes the hardware thread count. Loose wording only.

Second check: another model agreed

• p. 1 Footnote 2

  threads communicate communicate primarily

• p. 1 footnote 2

  communicate communicate primarily

• p. 1 Footnote 2

  communicate communicate

• p. 1 footnote 2

  concurrent threads communicate communicate primarily

Judge's note: Viewed the crop of footnote 2 on page 1. The word "communicate" is printed twice in a row on the same line, so this is not an extraction or line-break artefact.

Second check: code check

• p. 2 Section 2 Overview, first paragraph

  at least two of which have lead, to subtly incorrect

Judge's note: Viewed the crop of Section 2 on page 2; "have lead" is printed. Section 4 on page 3 uses the correct form ("the first problem led to an actual failure").

Second check: none (judge only)

• p. 3 Section 4.1, penultimate paragraph

  Resolving it essential requires a programming-language-defined

• p. 6 section 5.1

  fully synchronized versions in fig- ures 3

• p. 4 section 4.2

  it is know to the compiler

• p. 3 section 4.1 end

  Resolving it essential requires

Judge's note: Both slips are visible on the page 3 and page 6 images. The other items bundled into this cluster ("have lead", "it is know", "libary") are real but are judged under C002, C006 and C010; this cluster is confirmed for the two items unique to it.

Second check: none (judge only)

• p. 3 Section 4.2, bit-field example

  struct { int a:17; int b:15 } x;

Judge's note: The page 3 image shows the declaration exactly as quoted. The second struct on page 4 terminates its last member ("char h; } x;"). The paper's "taking some liberties with the C notation" remark is attached to a different expression on page 4. Cosmetic syntax slip in an illustrative snippet.

Second check: none (judge only)

• p. 4 Section 4.2, paragraph before the 8-char struct

  where it is know to the compiler that x is 64-bit

• p. 4 Section 4.2

  it is know to the compiler that x is 64-bit

Judge's note: Visible in the left column of the page 4 image.

Second check: none (judge only)

• p. 4 Section 4.3, first sentence

  updates were there were none in the source code

• p. 4 Section 4.3 first sentence

  introduce variable updates were there were none in the source code

• p. 4 Section 4.3 first paragraph

  updates were there were none

Judge's note: Visible in the first sentence of Section 4.3 on the page 4 image.

Second check: none (judge only)

• p. 4 Section 4.2, left column

  sanctioned by the Pthreads standards, since it prohibits

Judge's note: Visible on the page 4 image. Elsewhere the paper writes "the Pthreads standard" in the singular (pages 2 and 5), though footnote 10 does say "the Posix standards". A minor number-agreement slip; either form could have been intended.

Second check: none (judge only)

• p. 5 Section 5, third paragraph

  libary call

• p. 5 section 5, first paragraph

  per libary call

Judge's note: Visible in the left column of the page 5 image.

Second check: none (judge only)

• p. 5 Section 5, last paragraph of left column

  The continued use of the“double-checked locking” idiom

Judge's note: On the page 5 image (bottom of the left column) "the" and the opening quotation mark are run together.

Second check: none (judge only)

• p. 6 Section 5.1, paragraph beginning "Thus N copies"

  we will see speed-ups due the additional threads

Judge's note: Visible in the left column of the page 6 image, just before footnote mark 15.

Second check: none (judge only)

• p. 6 Section 5.1, paragraph describing Figure 1

  other synchronization,while the last

Judge's note: Viewed the crop of page 6. The printed line reads "other synchronization,while the last" with no space after the comma; the text layer agrees (main.txt line 864). It is in the published page, not an extraction artefact.

Second check: none (judge only)

• p. 6 Section 5.1, paragraph beginning "Repeating the byte-array experiment"

  fully synchronized versions in fig- ures 3

• p. 6 Section 5.1, right column, paragraph on the hyper-threaded Pentium 4

  less promising results for the fully synchronized versions in fig- ures 3

• p. 6 text before Figure 4 discussion

  fully synchronized versions in fig-

Judge's note: Viewed the crop of page 6. The right column prints "in fig-/ures 3. Here the single-threaded version..." across a line break. Only one figure number follows, and the Figure 3 caption ("HT P4 execution time for byte array (secs)") matches the hyper-threaded Pentium 4 byte-array experiment being described. The pointer goes to the right figure, so this is a number-agreement slip only.

Second check: none (judge only)