Alice, Re: "If you recall, the whole 179.art fits into US-III L2
cache. If for some stupid reason some portions of old code
used malloc() back and forth, and newer versions
of compiler eliminate this, 10x and more is possible.
For much smaller caches perhaps no one was doing this
check, but for huge 8MB they might start looking into
it. Alpha is doing well on SPEC mostly because of its
huge caches."
I don't buy the bullsh!t about it fitting in the cache. The IBM 1.3GHz Power4, widely recognized as currently the most powerful CPU in both integer and floating point, has 128MB of L3 cache, and it takes 153 seconds to finish 179.art.
spec.org
Your cache theory does not explain why the UltraSparc III 1050MHz can finish that same application in 31.8 seconds. The 179.art binary (not to mention the entire Spec binary) is much smaller than 128MB, so it should easily fit in either of the Power4 or USIII caches, yet the USIII finishes the application in nearly 1/5 the time!
There is no way that the USIII hardware could have suddenly caught up and surpassed Power4 on this benchmark by such a wide margin, simply by adding compiler enhancements.
Re: "Being caught in benchmark cheating is a big risk for
a big company."
Yes. And soon Sun will be caught for their fraudulent use of this benchmark, and be forced to have their scores stricken from the records.
wbmw |
