what is the fastest way to connect to the Empeg ?
I just got mine from Sonic Blue today, yay me.
I know that USB is faster that Ethernet on paper,
But I remember someone saying something about Ethernet being a protocol with less overhead? or something?
Blah, Blah, Blah...
I was wondering if anyone found Ethernet to ACTUALLY be faster than USB.

This is a FAQ Entry.

I think that FAQ entry could do with some editing Tony. For some people at least USB is more like twice as fast as Ethernet and it really is worth investing some effort in getting an Ethernet setup working.

I know I am not the only one who has found this, the USB on some machines just does not like emplode and empeg.

I think you meant to say that ethernet is TWICE as fast as USB. Which would be correct. It should be. That's the way the current implementation works. USB can be made to go faster, but this isn't something that's included with the current software set.

And USB on paper (original post) doesn't mean much. There are various classes of USB devices, each with their own speed. Then there are devices that use a certain class that should be relatively quick, but aren't: iomega USB Zip drive - USELESS.

empeg current USB 400-something KB/s. Ethernet 900-something KB/s. That info comes from Hugo and is plastered right in my jukebox matrix page where I've compared the empeg to both the Phatbox and Neo players. Tony might have a link to it somewhere on a FAQ page.

Bruno

So I did, too late to edit it unfortunately.

I thought only the 2.x software had the ethernet advantages and that was only for large transfers...

One thing I DID notice that ethernet is faster at is the small stuff like retreiving databases and basic communicating with the player.

Well, I have both USB and ETHERNET setup and I don't know that I can tell a difference in either, just yet

I'm on day 3 of adding music to the unit, 10 Gig so far,
I did it all on USB,
Will switch to Ethernet for the next 10 GB, and see if there's much difference...

Anyone else have REAL WORLD experience that that either protocol is faster?

regards - dcosta

Umm, the fact that altman, the designer of the product, says as much isn't proof enough?

Ok, how about the fact that I transfered 30GB with both ethernet as well as USB? And let me tell you, the ethernet wound was quite a bit faster. I wish I had written down the times, but I wasn't doing it for an excersize in benchmarking the empeg.

If you can use ethernet, use it - USB is next to useless for the empeg, IMO. Its use is limited to emplode. With ethernet you've got access to do pretty much anything to the player.

Bruno

USB is next to useless for the empeg, IMO. Its use is limited to emplode.

Boy, this Hijack, VNC and other stuff got us so spoiled... If we consider empeg an MP3 music player, then USB pretty much does what it should, no? Of course, I have connected my empeg via USB exactly once, when copying a large file from my nephew's network-challenged PC....

Well, I have both USB and ETHERNET setup and I don't know that I can tell a difference in either, just yet... [snip] 10 Gig so far, I did it all on USB

Umm....

Peter

There's no question that Ethernet is faster, assuming your network isn't loaded down.

Rob

Yeah, I didn't feel like stressing the obvious part of his post.

Bruno

I've only been running Hijack for two weeks. I've been following it since the very early releases back in November, but hadn't bothered to install it (I was waiting for one of those kenwood steering remotes). When Mark added so much extra functionality (enough to make the IR translation a secondary feature!) I knew I could wait no more.

I used USB only while trying to troubleshoot the problems I initially had with ethernet. And the problems were solved by forcing my machine down to 10Mbit communication with my switch. I'm sure the 10Mbit change has impacted my brother's net surfing and usage of my machine's resources, but he hasn't complained yet.

Bruno

Ethernet is greatly faster for me. Even better, I can just slap my baby onto a remote hub and be good to go

>Boy, this Hijack, VNC and other stuff got us so spoiled... If we consider empeg an MP3 music player, then USB pretty much does what it should, no?<

What do you mean by that? I haven't used mine yet, but what else can it be other than an MP3 music player?

Speaking of ethernet issues.. My new mk2a doesn't get along with my old Linksys 5-port 10/100 hub -- the hub refuses to forward frames to/from it to the other (newer Linksys 8-port, works great!) hub I have. Oddly enough, my Mk2 has no such issue, though.

Anyway, I just swapped out the pair of hubs for a new 16-port Nway switch (needed more ports anyway; only one left on the new setup..).
What'cha connected to?

I'm using just a cheapy D-Link DSS-8+ 10/100 switch. My main purpose in buying it was just to continue running 100 to a couple of machines and still use my printer @ 10. The empeg sync issue was the only trouble I've ever had. I could sync a bit of data, but could never finish anything large. After setting my machine to 10, I can do 20GB without a problem.

Bruno

Kinda Funny.
I have a 60 GB Rio Unit and the User's Manual states:
USB is the fastest way to connect your Rio Player to your PC.

I guess that's accurate Just maybe misleading... ?

It may be fast to physically connect it, so that part's true. If you don't have a NIC, you can't use ethernet. Plus the player doesn't ship with an ethernet cable. For wide compatibility (provided there isn't a bug in emplode...) USB is the easiest and most straight-forward connection.

Bruno

> My new mk2a doesn't get along with my old Linksys 5-port 10/100 hub -- the hub refuses to forward frames to/from it to the other (newer Linksys 8-port, works great!) hub I have.

Funny. My 8-port Linksys 10/100 does not work with my MK2a at all, but my 5-port 10/100 Linksys works fine.

I guess consistantcy is not Linksys's strong suit.

Well, my NFG 5-port is the original "grey colour" with the jacks on the front, whereas the good 8-port is "Revision 2", blue and black in colour with the jacks on the back.

-ml

Linksys BEFSR41 4 port router. Works poifectly, no tweakage necessary.

I guess consistantcy is not Linksys's strong suit.

No, it's not. They market network cards under the same model number that require different drivers based on the revision of the card. This caused me great grief when I needed a cheep Linux compatible card. The box said it was, but the revision inside didn't work with the existing modules out at the time. I couldn't wait, so I had to take it back. I've also ran into the same problem when trying to help a friend find Windows 2000 drivers. There was no way to find what revision he needed beyond opening the box. Why they can't even provide drivers that can determain the card on their own is beyond me. But because of this, and not very good reliability, I have long since abandoned Linksys products. 3Com or Intel seems to be decent, but at a cost.

3Com does the same versioning thing with their NICs. Have for years. I've been running a couple of 3Com (a few years old now) and a couple of Dlink without a problem though. The 3Com cost more than 4 times the price of the DLink.

Bruno

They market network cards under the same model number that require different drivers based on the revision of the card. This caused me great grief when I needed a cheep Linux compatible card. The box said it was, but the revision inside didn't work with the existing modules out at the time.

They do that with NICs, too?

You should have seen Hugo cursing them over the problems with the Linksys USB10T USB-to-Ethernet adapter that they wanted to support for the Jupiter. Same thing: Outside of Hugo's box said all the same things as my box did. When queried by the driver software, both units return the same ID number. But they used different chips under the surface, and the drivers had to be completely different ones.

I don't know how they managed to work around it in the end, but they did. Both of our adapters now work on the HSX-109.

Until Linksys revises it again, of course.

With all my cards, 3Com has either kept the same chipset/drivers, or created slightly different models. In my house I have a 3c905, 3c905b and a 3c905c. And a few others (Like an ancient ISA card in my Linux box). The one complaint I could possibly have with the 3Com products was lack of Pocket PC drivers for their AirConnect cards early on. But they still had them out faster then most of the other cards at the time.

"You should have seen Hugo cursing them over the problems with the Linksys USB10T USB-to-Ethernet adapter that they wanted to support for the Jupiter"

I cursed just as much (well, maybe half :P ) from the different chipsets that they used, as only ONE works with the 3com audrey. Guess which chipset I ended up with in both the adapters I bought?

Have you had much (any?) experience with Netgear NICs? I've had good success with their hubs and switches for small applications, and with their NICs, but I don't know if they arbitrarily change chipsets like so may other of the discount NIC companies do. I'm guessing not, since it's run by a well-known company in the networking world (Bay), but I don't really know for sure. If I had more info in this aspect about their NICs, I would feel very comfortable recommending all of their products to just about anyone.

I believe the card in my media system right now is a NetGear. No problems that I can think of off hand with it, it just sits there chugging away when someone uses the network on that machine.

No real experience though to say good/bad about their products though.

I meant if you knew whether they change chipsets or not. Right now, Netgear is the only value line that I can recommend or will consider using. I'm hoping to find that they exceed in that area, as well.

After years of messing with different home networking setups (and some small client installations) I would recommend Netgear - they seem to be pretty robust and sensible, in a 'you know what you're getting' kind of way - drivers work as expected etc. They are nothing special, and as stated above 3com/Intel probably have the best cards but for a small/home office setup I'd go with Netgear every time.

One of the good thinks about Netgear, compared to other cheap kit is the boxes. I far prefer a decent metal case on my hubs to a cheap and tacky plastic one. The metal Netgear cases just feel so much better made.

And they're such a great shade of blue, too .

Yes, nearly as nice a shade of blue as the link light on the back of the Mk2, which is nearly as nice as the signal indicator on my Geodesy (http://www.morpheous.com/)...

Well, it seems that we got a little off topic here, but hey, so what.
I've been using Ethernet to send music to the player.
I did three trials to benchmark the "uploading tunes" portion of syncronization.

The average of the three was 24.6 Megabytes per minute.
- Call it 25 Megabytes per minute
I know this is not the standard unit for transfer rates,
but I feel it is a more useful benchmark number.
If you have 100 Megabytes of stuff to put on the player,
it'll take 4 minutes (or so)

This benchmark does not include time for
reading / deleteing / rebuilding database, etc,
as those times will vary depending on database size,
ammount of data on the unit, disk size, etc.

I haven't gotten around to benchmarking my USB connection,
but I suspect that they will be significantly slower.

That sounds quite slow. For me, emptool reports between 700Kbytes/sec and 1Mbyte/sec (occasionally, late at night when the Ethernet switch is otherwise idle). That's 40-60 MB/s in your units.

For me, emptool reports between 700Kbytes/sec and 1Mbyte/sec (occasionally, late at night when the Ethernet switch is otherwise idle).

Same here. And if anyone can figure out why it varies so wildly (from 700-950Kbytes/sec) from one tune to the next, I'd be very interested. Especially if it led to a technique for always getting the 950 value.

Using tar and netcat to hose a lot of fids onto a car player gave me 700Kbytes/sec on average. It's thus possible that the 950 value was due to buffers filling that had drained slightly between songs, and doesn't represent a sustainable rate.

Peter

When I'm uploading, I can hear whether or not it's a fast one. I get noise from my display card appearing on my (on-mobo) mixer, too quiet to hear when I'm playing the empeg through it, but I can just detect the percentage meter changing. Maximum throughput occurs when a track is long enough that the per-track overhead is small, but not so long that transfer is stalled when it gets ahead of disk access. I'm assuming that's what we're waiting for; half a second or so at a time. I assuming it's the writing on the empeg that's the issue, rather than reading (I'm reading from a local Reiser FS). Of course, normal network traffic is also an issue, as I hinted in my earlier post.

700 Kbits per secons, eh ?

I was using a crossover cable for my trials, so Network traffic is not a variable, must be something in windows, or with my network card / driver,
I'm using win98 SE with a Linksys EtherFast 10/100 Card

Anyone have tips for tweaks I might make?
(other than using linux)

Anyone have tips for tweaks I might make?

I don't think you've mentioned which version of the player software you're running. My 700Kbytes/sec, and I suspect tms13's too, were on public betas of the forthcoming v2.0 of the player software. The current recommended version, 1.03, is slower than the 2.0 betas.

Peter

Sorry,

I have a 60 Gig Rio running 2.00 beta 11 and I used the Logo editor to apply v200.hijack.v200b11.mk2.zImage

In reply to:

---

My 700Kbytes/sec, and I suspect tms13's too, were on public betas of the forthcoming v2.0 of the player software.

---

Yes, I'm tracking the beta releases. On the host side, I'm not tracking quite as keenly: Linux 2.4.9-ac9.

Figured I'd bring this back up....

I've been doing some testing, actually some timing.
There are like 10 ways to connect to the Empeg these days,
with all kinds of great software,
But the question remains : What's the Fastest ???
This thread has lead me to belive that Ethernet is fastest,
coming in around 700 - 900 KiloBytes/second.

The only way I was ever able to get that speed was using straight FTP.
When I use Emplode to upload songs I only get around 300 - 400 KiloBytes/second and
When I use Jemplode to upload songs I only get around 450 - 550 KiloBytes/second.

The only good way I've been able to calculate this is to record the size of the data I'm uploading, then pay attention to the synchronization routine and start my stopwatch when songs strat uploading and stop it when they stop.

I know It's not a hardware issue, because I can get 700 - 900 KiloBytes/second over FTP.

Now if the syncronization can't do the FTP speed, I say why not ?
Why can't I get the 700 - 900 KiloBytes/second I get on FTP ?

like to know if anyone has used Emplode or Jemplode to upload data at a rate greater than 600 KiloBytes/second ?

Hey, if you don't want to time all your uploads and calculate the speeds, goto http://www.dumeter.com/ and download thier software, i've been using it for about 3 years now. Works very well!

Remember that the empeg only has an ISA bus, not much RAM, and doesn't have a huge amount of speed optimization in its circuitry. They were more concerned with making it reliable and durable than blazingly fast. It's possible you may have reached the limit of what it can do realistically.

Why the fascination with raw upload speed? For any given player, you'd only need to do a huge upload a few times during its lifespan.

You should be able to get that level of speed, emplode will use "fast channel" (instead of the normal blocked protocol) over ethernet, opening a separate tcp pipe and just dumping the file down it. If the fast channel fails though, it'll go back to the slow method.

Hugo

Tony you need a bigger drive!

I agree for the most part however - it is the database rebuild that really takes forever...

Yes but this ISA bus and low RAM machine IS capable of 900 KB/s over FTP,
It's just dissapointing to know that uploading during synchro is about half the speeed of achievable FTP speed.
I'm sure everyone here would love to double their upload speed during synchro.