I hope that this is the correct forum to post. The problem that I will describe only occurs on laptops with the express chipset. I've tested it on several laptops. I will just paste what I had asked in a microsoft forum. Hope someone can help. Thanks.
Laptop: Acer Aspire Ethos 8951G - 9824
OS: Microsoft Windows 7 Home Premium SP1
Processor: Intel(R) Core(TM) i7-2630QM CPU @ 2.00GHz, 2001 Mhz, 4 Core(s), 8 Logical Processor(s)
BIOS: INSYDE V1.13, 30/12/2011
SMBIOS: Version 2.7
Memory: 16.0 GB DDR3
Graphics: NVIDIA GeForce GT555M 2GB
Internal HDDs: 2 Seagate Momentus 750 GB 300 MB/s IO transfer rate, SATA 3Gb/s interface
IDE ATA/ATAPI controllers: Intel Mobile Express Chipset SATA AHCI Controller
All drivers updated to the latest
To safeguard very sensitive data, on a large partition of the secondary HDD, I have installed an On-the-fly-encryption system, FreeOTFE, considered one of the best disk encryption systems out there. Once mounted, it is completely transparent, and any program such as Office, media players etc. can directly access any file. From experience on other FreeOTFE installations and from my own analysis of the encryption algorithms, the computational overhead is minimal, even on the strongest encryption available, and certainly on my current system, it should not even register.
Here is the problem.
When transfering large datasets using just explorer from the primary HDD to the encrypted partion of the secondary HDD, the first few datasets transfer with the expected performance and speed. However, the speed suddenly begins to decline until it stabilizes at a painful 600K/s. This also occurs when moving one large dataset: while watching the progress on anything above 250MB the first have moves quickly and then it rapidly declines. Other copying tools such as Robocopy, teracopy experience the same thing. Once the I/O speed has been locked, any further operations (Office, media, navigating directories etc.) also move at that speed until a reboot.
I have installed this software on Vista and XP, and the drive never behaved like this. Also, I see the same thing when I have connected external HDDs with encrypted partitions by USB.
Some people have suggested DMA/PIO problem, that the IO is slipping to PIO. But I am sure how this applies to a SATA interface. And even if it is, the usually suggested solution is to open up device manager, IDE ATA/ATAPI controllers and check for the ATA channel 0 status, and under the advanced tab, change it to DMA. But as you can see from my specs, under that tab only thing listed is the chipset, and there is nothing to set in its properties. Another helpful suggestion from this article: http://winhlp.com/node/10, which suggested to check the kernel times in Task Manager. It stated that if the kernel times are around half the total load when doing something like a file copy, then it is likely that the machine is running in PIO mode. I have done this. On a large file copy or multiple file copies (with nothing else running including anti-virus), the kernel times go from nothing to dominating the load.
Finally, this problem is not restricted to this laptop. I have a similarly configured HP, with, in particular, the Express Chipset, and it experiences exactly the same problems. I also tested both laptops using other encryption software such as TrueCrypt, and the same result. Therefore, I hypothesise that the problem is related to the chipset.
So the question is under what conditions does the performance of the chipset degrade to something like PIO, and is there a solution / workaround.
When it comes to performance issues this is something that is handled by the computer manufacturer.
What I recommend is trying the hard drive without the encryption software installed. Also you can update the BIOS. You can also try different hard drives. Above are just a few suggestions you may want to try furthermore you can contact the Acer Support Department using the following link http://us.acer.com/ac/en/US/content/support
thanks for teh response, but it is not an issue limited to the manufacturer. I have now tested this on 6 machines of different manufacturers, three with the Mobile Express Chipset SATA AHCI Controller and three with IDE controllers, but otherwise similarly configured. In each case, the machines with the chipsets experienced extremely degraded performance, whereas the IDE machines were fast. BIOS was updated in all.
Without the encryption software, all performed similarly. I cannot understand why the software would degrade performance on the chipset machines and not the IDE ones.
As I see it, the fundamental question is under what circumstances does SATA degrade to something like PIO, and is there a fix or work around.