EHD transfer rates

[fletchmath]

I've found a few threads discussing transfer speeds for EHD, but nothing real clear; if I missed a good post, please accept my apologies and point me in the right direction.

I am wondering if the transfer speeds I see are typical. I haven't measured real precisely, but a "typical" 1-hr HD show (around 4GB, give or take) seems to take about 9.5 minutes to archive to the EHD.

So, I'm getting somewhere between 40 and 80 Mbs transfer speeds. The "rated" speed of USB 2.0 is 480 Mbs. That's a big enough gap that I wonder if I could do better.

I have a vip722 and a mybook essential 750GB EHD. I'm using the USB cable that came with the EHD.

So, what transfer speeds are others getting?

Cheers!

 

[KAB]

You're about normal. The receivers control the transfer rate.

 

[8bitbytes]

Even though we have USB 2.0, the programmers seem to have limited the transfer rate to Normal Mode, not Fast Mode. Maybe this is to prevent overworking the CPU when you are recording two shows, transferring a show to EHD, and watching something already recorded.

 

[TheKrell]

Even though we have USB 2.0, the programmers seem to have limited the transfer rate to Normal Mode, not Fast Mode. Maybe this is to prevent overworking the CPU when you are recording two shows, transferring a show to EHD, and watching something already recorded.

I think 40Mbps is already much faster than USB 1.1, if that's what you meant by "Normal Mode." That tops out at 12Mbps. I always thought this (40Mbps) relatively slow speed (compared to USB 2.0) was due to the encryption. That has got to hurt.

Speaking of which, does anybody know if Dish is gratuitously encrypting OTA recordings transferred to the EHD?

 

[8bitbytes]

The fastest operational speed typically seen with USB 2.0 is 40MB/s or 320mb/s. 480mb/s is the theoretical top but because of the software handling process for moving data, 60MB/s just isn't going to happen. Encryption should only cut that speed in half at most; usually only a 10% reduction is speed is seen with the basic encryption used here.

40mb/s would be 5 MB/s or about 1/8th top speed which is what normal mode runs at.

USB 3.0 is coming soon and will increase top speed to gigabits per second but will require a new chipset.

 

[fletchmath]

Thanks for the input, folks. If they are limiting the bandwidth artificially to keep from overwhelming the system, well, then there's room for improvement. It would be great to see the throttle removed when there's nothing going on to interfere with.

It can't be the encryption scheme, unless they're using something really inefficient; you can apply strong encryption on the fly without reducing throughput by 10%, let alone 80%.

 

[sgarringer]

No, what 8bitbites said is correct. You're getting confused by the 480 megabit per second of USB and the 40 megabyte per second you measured.

There are overhead associated with both the USB2 protocol and also with the encryption. 40 MB/sec is pretty good for USB2.0. There is not going to be any improvement.

 

[TheKrell]

I think everybody but sgarringer is using lower case "b" to indicate bits, and an uppercase "B" to indicate bytes.

As far as guessing 10% speed reductions due to encryption... Where did that number come from? These receivers do not have anything equivalent to a fast general-purpose processor inside. Their only programmable processor might not be very fast. 100's of MHz is typical of DVD player processors, for example. The only real horsepower needed in these receivers is in the MPEG decompression, and that's going on inside a non-programmable ASIC. You should be thinking of a 100MHz 486 without floating point coprocessor, not a 3GHz P4! How much would that slow down the transfer? I don't have any such beast left alive to try it out, but 80% reduction in throughput (due to encryption alone) seems about right. Heck, back in the early 90's, a 486 couldn't even keep up with the original slow Ethernet speeds, which were 10Mbps. Ethernet cards without their own buffers would typically drop every other packet. Throughput stunk beyond belief.

40mb/s would be 5 MB/s or about 1/8th top speed which is what normal mode runs at.

What are you telling us? There is a "normal mode" for USB2 that runs at only 1/8 of typical (top) speeds?

 

[fletchmath]

No, what 8bitbites said is correct. You're getting confused by the 480 megabit per second of USB and the 40 megabyte per second you measured.

There are overhead associated with both the USB2 protocol and also with the encryption. 40 MB/sec is pretty good for USB2.0. There is not going to be any improvement.

If I was seeing 40 MB/sec, I would never have written -- that would be excellent throughput indeed. I was seeing something close to 40 Mbit/sec, which is a bit slower.

 

[fletchmath]

As far as guessing 10% speed reductions due to encryption... Where did that number come from? These receivers do not have anything equivalent to a fast general-purpose processor inside. Their only programmable processor might not be very fast.

I was thinking about stream ciphers like the one used in wpa2 (RC4? something like that). I was assuming that the processing power in most consumer-grade wireless routers does not exceed what's available in a box like the 722. On the other hand, I didn't think very hard about the throughput requirements (does the dvr push more bits than a 802.11n router? I guess so), nor did I think particularly hard about whether such ciphers (or something similar to them) would be suitable for this job. I do know that modest hardware is able to apply the wpa2 stream encryption with something like 10% overhead.

So, I suppose like most internet posters, I was speaking at least partly out of my ass. Sorry about that.

Ken