- WD3000GLFS and -HLFS (10k rpm, 16MB) (built arround 2010)
- Samsung 1TB HD103UJ (7.200rpm, 32MB) (years 2008, 2009)
- three different types of Hitachi 1TB drives with 7200rpm and 16MB and 32MB, and a Cinemastar 5400rpm. (years, 2007, 2008, 2009, 2010 and 2011)
- 3TB Toshiba, 64MB (was the fastest topping out at 39MB/s) (2013)
- 80GB Seagate 7200rpm 8MB, 160GB WD-YD 7200rpm 16MB, 160GB 2,5" 5400rpm Fujitsu 8MB (all built before 2008)
- 1TB 2,5" Toshiba, 5400rpm, 8MB (2013)
So the variety speaks for the Drives not being the bottleneck, I guess
See my answer
to Aqua: "variety of drives"
and my first post: "1) it's about internal SATA-PCI, which should be a "bit" faster than firewire, 2) different PCI-Cards tested 3) PowerMac is the same with every drive and Card, so no change there. 4) under variety of drives transfer speeds do not much differ"
JapaMac for example has a PowerMac G4 AGP or GigabitEthernet and he tested all ports once with AJA-Test and other benchmark tests, he got 100MB/s (even 114MB/s once) for SATA. http://www.jcsenterprises.com/Japamacs_Page/Hard_Drive_Benchmarks.html
So in his example the Mac is the same and he uses even older/slower drives like the WD6400AAKS. So how come that two persons with the same Mac and same PCI-Cards and comparable SATA-Drives get a difference of 60MB/s???
Oh, and since he uses a WD Raptor, which was built until 2006 or 2007 (most often sold as 74GB and 80GB version, some rare 150GB models), it must be slower anyway. I have a 300GB Velociraptor in my tests, which is newer and bigger (-> dense platters). If his Raptor is a 1st or 2nd generation Raptor, it is PATA internally with a SATA connector outside (yes WD made such fake SATA drives).
No matter what hardware anyone uses on a G4, the paltry 100-133,167 mhz bus is always going to slow it down considerably.. hense why I got rid of my G4 Powermacs. I think the G4 is very much underpowered and its due to the architecture.
SATA drives using a PCI card in a G4 are going to run ok, but don't expect a suddent burst in speed compared to say a Mac Pro.
First thank you, all! You make me confident, that it is not me making a mistake somewhere (I can't imagine what mistake anyway since it is plug and play. But people elsewhere like to point out, that the user is the biggest source of mistakes).
What I ask myself again is:
why then, are people claiming that they get 100MB/s speeds on the same PowerMac G4 Sawtooth ("AGP", with 100MHz-BUS)? I trust Japamac, that he did do tests, but what really comes to mind is, that it must be the benchmark utilities (Quickbench and ALA) making people think they have 100MB/s speeds on a theoretical SATA-150MB connection.
Btw. I have tried the 500MHz stock CPU, a Sonnet 7447 2x1,8GHz and a Sonnet 1x1,2GHz 7455, I could see that the 500MHz CPU is overal 5MB slower, when processing large files. 5MB is not siginficant, I know.
I am not saying, that a Dual 1,8GHz is making everything different and neglecting what you said, when you just transfer files and I see that the CPU is somehow oversized on a 100MHz FSB. Of course people swear their Upgrade CPUs do much. I guess they really roll out their advantages in let's say converting DV-files to another format. (I plan to test this the next weeks btw., also watching power consumption).
Thank you again. I hope Japamac says something on the topic, since he still posses his Sawtooth G4 and maybe he can reperform his test with a real file (other than "virtual" benchmarking). A pitty that there is no message function on this forum.
PS: on RAM usage, other than music authoring software and photoshop, RAM doesn't seem to be that big of a deal on a PowerMac. I watched activity monitor, while converting an iMovie-file to h.264 (took like 8 hours with the Dual 1,8GHz), the RAM was never filled over 512MB and I have 2GB installed. OS X is very nice to RAM, too it seems.
If you are so concerned about your data transfer/exchange rates across different protocols maybe you should check ou the website XLR8yourMac
Maybe you can find the real world empirical data you search for there.
Don't get me wrong, I really appreciate your help and effort, but you seem to not have read my first post.
Unless you use a real file and transfer it manually and stop count the time, the test is meaningless. Expecially with XBench. It is widely known for being total crap (the most crap for graphics card. ATI 7000 gets higher rates than Nvidia 7800GTX or modern High-End cards). I did tests with Xbench sometime, too. I had three different IDE-PCI-Controllers and wanted to know, which is the fastest.
For one drive and one PCI-Card, I had vastly differing results, doing 10 tests of the identical euqippment. Try it yourself and do 10 test of the Seagate drive you have, with the controller you have and with XBench.