Closed Bug 942311 Opened 9 years ago Closed 9 years ago

Correct from int32x4 to uint32x4

Categories

(Core :: JavaScript Engine, defect)

x86
Linux
defect
Not set
normal

Tracking

()

RESOLVED WONTFIX

People

(Reporter: nmatsakis, Unassigned)

References

Details

Attachments

(1 file)

Somewhat embarassing, but when implementing the proposed SIMD types I had the mistaken impression we wanted int32x4, when in fact we want uint32x4. This should be corrected.
Blocks: 904913
Attached patch Bug942311.diffSplinter Review
Attachment #8337034 - Flags: review?(till)
Hi,

The type is actually int32x4. There was some waffling but its been fixed to int32x4.
I should have mentioned the design document I sent to you was a little out of data so that you could save time for this patch. Sorry about that. The document was written in September, after that:
  1) Float32x4Array and Int32x4Array was defined in John's proposal.
  2) uint32x4 is renamed to int32x4.
  3) SIMD.add and SIMD.addu32 are changed to SIMD.float32x4.add and SIMD.int32x4.add.
As those changes does not affect the engine implementation too much, I did not update the document.

I am rebasing with V8 master, after that, I could send out a code review request to Google V8 team so that you could look at that for a reference implementation, at least, at API level.
Comment on attachment 8337034 [details] [diff] [review]
Bug942311.diff

Canceling review based on comments 2 and 3
Attachment #8337034 - Flags: review?(till)
Huh. OK. The strawman http://wiki.ecmascript.org/doku.php?id=strawman:simd_number is out of date, in that case.
Here is the V8 SIMD implementation URL: https://codereview.chromium.org/90643003/.
Closing because it seems like I was mistaken and code is correct as is.
Status: NEW → RESOLVED
Closed: 9 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.