I am not sure how the frequency scaling works on the core i5 760 and if it is controlled by the hardware or if there is some linux involvement.
When doing copies to a usb hard drive I noticed high CPU usage % but the CPU was reported running at 1.2Ghz, so I decided to run a little test.
I copied over 29 GB of data from a sata II 7.2k rpm drive to an external sata II 7.2krpm drive which was attached via usb. I used the inbuilt graphical system monitor to check the cpu load and kinfocenter for the cpu speed. During the test which ran for a little less than 7 minutes, the cpu load varied but appeared to be generally in the 60% to 80% range with normally 2 cores running at 100%. The cpu speed display automatically refreshes and I periodically refreshed manually, never moved from 1.2GHz. This leaves me with the impression that the copy performance is being limited by the CPU. I calculated the data transfer as being fractionally under 70MBytes/s. I am also a little worried about the amount of CPU being used for the copy, even at 1.2Ghz. 2 cores and a bit of 2 others seems a little high
I regularly backup the entire system, about 7 or 8 file systems. This tends to run on one core and drives it mostly at 100% doing compression. Again the system mainly reports the CPU speed as 1.2GHz with 1 core 100% busy. This process sometimes results in the cpu speed changing to 2.8GHz
The system contains 3 sata II disks and has 4GB of DDR3 1333Mhz RAM.
This does not look right to me. Is the i5 760 not well supported or is there another problem?
I didn’t hear any problem with the iCore series CPU with linux. Do you have the same symptoms if you do the copy in terminal ? Can you try in GNOME too ?
I used the inbuilt graphical system monitor to check the cpu load and kinfocenter for the cpu speed. During the test which ran for a little less than 7 minutes, the cpu load varied but appeared to be generally in the 60% to 80% range with normally 2 cores running at 100%. The cpu speed display automatically refreshes and I periodically refreshed manually, never moved from 1.2GHz. This leaves me with the impression that the copy performance is being limited by the CPU.
I look at that evidence and suspect the exact opposite; it runs at 1.2 GHz because running faster wouldn’t improve performance and it is limited by the speed of IO to and from the disks. Running the CPU faster won’t make data pass to and from the disk faster, whether the bottleneck is the USB part or the disk mechanics, but if the CPUs aren’t being saturated with the current situation the speed won’t step up to the next level.
Are you backing up with Dolphin? I find it relatively slow and I would not backup with Dolphin myself. For example, I find the older Konqueror faster (and more stable).
The copy was cp * … from the command line followed by sync and wrapped by date for the timing. I had already thought that the load might be exactly on the lower limit of a core but I do find it strange that the cpu load can it seems, exaclty match 100% on 2 cores. I also still find the amount of CPU used to do the copy amazing. If it were running at full speed the copy is in effect taking a little bit less that 25% of a 2.8 GHz i5 760
It may be that you need about 12.5% of a 760 to read data off a sata II disk and anothe 12.5% to write the data to the USB disk. This leads me to the thought that an SSD to SSD copy is likely to eat the entire CPU.
I have just tried an internal sata II disk to disk copy of the same data. This is a bit faster than the internal to usb copy, 80 MBytes/sec compared to 70 MBytes per sec but the cpu load looks similar and the cpu speed stays at 1.2GHz. Maybe the scheduler loads a core until there is no wait time and then spills work to the next core
My motherboard optionally supports USB 3 or SATA III. As far as I can see the system would have difficulty with a loaded sata II subsystem. The same appears to be true for USB 3 although I suspect that the majority of systems would use usb 3 devices for backup and data transfer and I think that the maximum transfer speed is likely to be in the 70 - 80 MB/s range of the standard sata II desktop disks.