I do have four computers, all networked via a router/switch with gigabit LAN capability (Linksys WRT 350N.) The three desktops in the network have gigabit LAN capability. Their characteristics are shown in the table below. All network cables are Cat 6, 3 to 6 ft long. Transferring a 5 GB file between DM-4 and DM-5, the actual transmission speeds were about 15 to 20 Mb/s. Disconnecting DM-4 and 5 from the network and connecting them via a Cat 6 cross-over cable, the transmission rate rose to 45Mb/s. The Speed & Duplex settings on DM-4 and DM-5 are ‘Auto Negotiation’. All software firewalls have the IP range of the computers in their 'trusted zone'. All hard disks are SATA.

How can I realize transmission rates closer to the design rate of 1000 Mb/s?


Computer: DM-5 DM-4 DM-3
Motherboard: Asus Intel Intel
P5B Deluxe Wi-Fi D955XBK D875PBZ
OS Vista Ultimate Vista Ultimate Win XP SP2
Hard Disks 4 2 2
Disk size 250 GB each 250 GB each 250GB
150GB
Disk Configuration RAID 5 RAID 1 Non-RAID

1Gb networks are somewhat of a misnomer. You will never get near 1Gb/s. Yours is actually not bad. There are all kinds of bottlenecks in networks - PCI bus speed is a big one. According to the P5B manual, you have PCI 2.2 which has a "theoretical" speed of 528Mb/sec (66MHz) - so there's half your 1Gb to start. Then, PCI bus speed is shared/divided by all PCI devices - including your sound device (whether PCI or on-board). TCP/IP overhead (header/trailer data) consumes a lot, as does the drive controller, OS, OS overhead, background activity (including anti-malware scanners). And even though you have the firewalls set to Trusted, each packet is still analyzed to see if it meets that criteria. It is not like you set the flood gates to open and all comes through unchecked. And of course, this is happening on both ends of the cable. CPU speed, RAM speed and amount, are factors too. Even the HD buffer size can affect overall speed. And USB devices compete for bandwidth too. And, of course, the transfer speed will be determined by the slowest PC.

You might realize a little faster speed if you disconnect your router from the Internet, disable all scanners and firewalls, then unload all other startup programs.

Finally, proper connector termination is critical for CAT-6 cables. Did you test your cables with a good tester? If not, you should, or at least swap cables to see if it makes a difference.

Bus Speed Chart

Thanks for your reply.

I realized some of what you say this morning, when I transferred some large files on both DM-4 and DM-5 from one disk partition to another on the same machine. The transfer rate there was about 40 Mb/s.

I'm not sure I understand what you say about connector termination. I got brand new Belkin 'Gold' cables & would assume their connectors are appropriately 'teminated'. Of 'termination' I only know in connection with SCSI devices. I also doubt that that's the problem, when I first hgot the new router I had Cat 5 cables and got about the same transfer rate as with the Cat 6 cables.

Thanks again for your reply.

Peter Kroeger

By connector termination, I mean how the RJ-45 connectors are attached to the cable by the cable maker. I do NOT mean a terminator, such as is used on the SCSI bus. I am glad you asked for clarification.

Sadly, you cannot assume pre-made cables are made properly - even the top line of the premium brands can have problems - that's why I always make my own cables - if I don't get a good signal from end-to-end, I only have me to blame. Plus, I don't have to buy a 25 foot cable for a 11' run.

Note that transferring from one partition to another on the same hard drive can actually give some of the slowest times. With different drives, you have one drive reading while the other drive is writing. If copying on the same drive, the read/write heads can only be reading or writing at any given moment, plus the arm the heads are mounted on have to swing back and forth from the source location to the destination location - all of which slows things down. The fastest transfer would be between two separate drives in the same computer.

Thxs for your reply.

Re the TR-45 cables, without test equipment, I can't really do too miuch about it. Though, with the Cat5 cables giving me about the same transmission rates & with faulty cable attachments (hopefully) the exception I doubt that that's the problem.

I see what you're saying about transferring data within the same disks. Here I'm remaining within an environment with 3 Gb/s design transmission rate & if I got say 1 Gb/s, I'd be elated. Just for the heck of it I made a run on DM-3, which has two independent hard disks, copying from one to the other & got only 25 Mb/s. But then, it's SATA design speed is only 1 Gb/s vs. 3 for the other two.

Peter K

without test equipment, I can't really do too miuch about it.

Except swap cables and see if you get the same results.

As far as transferring between two HDs in the same machine, again, it is not like you are attaching two HDs together and letting them exchange data - the controller, RAM, CPU, and OS are all adding to the overhead.