For applications like this several runs should be done to get a good 'mark'. The text drawing (quartz) bug is still present if you run it a couple of times the characters are mapped and the test result is about 4x larger than a first time run. Note: Although there weren't any/many intel specific optimizations performed, the intel part was compiled w/ gcc 4 while the ppc w/ 3.3? Sounds like it should be fair. The step forward is just that we can get a test result from an intel based mac. Download Mac Os Benchmark shareware, freeware, demo, software, files.
#Novabench for mac os lion software#
So, I think a markdown for the benchmark is required. NovaBench is a free benchmark software that helps in testing, comparing and sharing. Do you round each task up, round down, leave it alone? A lot of the old reviews can't get the hardware to do this sort of test again. For third party sites that did reviews based on the previous baseline set it becomes a task. This is find for the Xbench site (simple recalculation) and for the application except smaller numbers mean slightly lower levels of precision.
Now everything must be recalculated and adjusted. Most of the benchmarks on current machines that exist on the web and many older ones use the older standard. While I am glad that more developers are releasing UB applications, I think this may represent a 1 step to the side two steps back approach for Xbench, merely because of the change in the baseline machine.Ī Baseline was established and existed for months prior to this. The current version is released as a universal binary and is only 52k larger than its 644k predecessor. Xbench is a highly popular 'benchmarking' suite for Mac OS X.
0 kommentar(er)
