Hi there. The 4.6-2011.10 release causes an output miscompare failure in 175.vpr from SPEC 2000. The fault didn't exist in 2011.09 and has cleared in 2011.11.
Does this ring a bell with anyone? Andrew, could it be related to your widening multiplies fix?
-- Michael
On 16/11/11 22:03, Michael Hope wrote:
Hi there. The 4.6-2011.10 release causes an output miscompare failure in 175.vpr from SPEC 2000. The fault didn't exist in 2011.09 and has cleared in 2011.11.
Does this ring a bell with anyone? Andrew, could it be related to your widening multiplies fix?
As far as I can see, there were no widening-multiplies related changes between 2011.09 and 2011.10, so no, I don't think so.
Andrew
On Fri, Nov 18, 2011 at 3:02 AM, Andrew Stubbs andrew.stubbs@linaro.org wrote:
On 16/11/11 22:03, Michael Hope wrote:
Hi there. The 4.6-2011.10 release causes an output miscompare failure in 175.vpr from SPEC 2000. The fault didn't exist in 2011.09 and has cleared in 2011.11.
Does this ring a bell with anyone? Andrew, could it be related to your widening multiplies fix?
As far as I can see, there were no widening-multiplies related changes between 2011.09 and 2011.10, so no, I don't think so.
The fault cleared with the 4.6.2 merge. That implies that it was introduced with the previous release branch merge. That's a worry.
-- Michael
linaro-toolchain@lists.linaro.org
-
Andrew Stubbs -
Michael Hope