There's obviously something wrong with the implementation in this test. Re-running the test will show the correct result, mostly.vecLib needs to be thrown out on this test or rewritten to give consistent results. I'll have to run again.Sometimes the vecLib FFT score hits some bug and gives a very small score, for some reason. It is extremely low, and drags down my whole CPU score. Quote:Originally posted by gcc:quote:Originally posted by Zapchud:quote:Originally posted by thehardcard:Whoa! What is up with my vecLib FFT score. Plus the SATA drives are going nowhere but up in speed anyway.Edit: by 'typo' do you mean that the internal rate should be 683MB/sec instead of Mbits? However, the speed still isn't that bad all things considered. I have a few of its nearly identical brothers the CudaIV.That was my initial thoughts as well (considering that this is the 'lowend' 1.6 G5). Its damn near silent through and a nice drive if you don't care too much about performance. The 7200.7 is pretty slow because its basically a budget part for OEMs. If that is indeed the case, then perhaps Apple's device driver is still undergoing some tweaks.quote:Thats a typo. Only with a 4 way raid setup would you probably reach this.Certainly I understand this, which is why I went into detail about the theoretical versus actual speeds, but the fact that the SATA drive is going to be virtually the same speed as the ATA100, at least at the mechanism level means that the only improvements that will be seen is at the serial interface and driver level-with this particular drive. Quote:BTW In reference to the post ahead of me:I don't think there are ANY SATA drives that can even come close to saturating 150Mbytes /sec. Quote: BTW I just ordered a 36GB 10000rpm SATA WD Raptor drive for my G5. It looks like theres some write cacheing going on. Anyway those HD scores are probably wrong. I have a few of its nearly identical brothers the CudaIV. It also means that there isn't much difference between that Seagate SATA drive and their ATA100 drive, which is why you're seeing similar performance from both on the x-bench test.That drive actually has an internal transfer speed lower than the theoretical speed of Firewire 800. Still the average transfer rate is around 58MB/sec, so this may never become an issue in the current drive design. However, you'll note that the actual drive componentry is shared between the SATA and ATA100 drive. Considering that the interface max theoretical speed is 150 MB/sec for a SATA drive, this seems unnecessarily limiting for the drive. That means that the maximum actual transfer rate is going to be 85.375 MB/sec. Well, the specs look good except for one thing-the max internal transfer rate is 683 _Mbits_ according to the spec sheet. Well, the specs look good except for one thing-the max internal transfer rate is 683 Mbits according to the spec sheet. im not really sure what those numbers mean, or how they relate to what xbench gives us, but hopefully someone will. Not only does the scores jump up and down depending on how good Xbench's hair day is, but the way it compares different CPU-architectures really makes it the most irrelevant synthetic benchmark to date.Hell, even Let1kWindowsBloom is a better benchmark, at least, it's benching a real world operation Quite obviously - View image here: - This test must be one of the most unreliable ones I've ever seen. Quote:Originally posted by Eug Wanker:Processor PowerPC 970 1.60 GHz - CPU Test 118.6Processor PowerPC G4 1.00 GHz - CPU Test 121.73 W00t! My TiBook's CPU speed is faster than a G5 1.6 GHz! Quote:Originally posted by gcc:From the asm profiles I've looked at on Xbench most of the tests are far too trivial to make any assessment about performance on real world apps.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |