cut the time uploading to the MK2 in almost half?

a relevation just cale to me. i haven't tried to search if this was mentioned before, but for the ~year that i've been frequenting this board, it hasn't been mentioned.

if we use both the LAN port AND the USB port, similar to a dual WAN load balancing router, we can effectively upload tunes 1.75x faster (guesstimate). each acts as an independent channel. one track on eth, one on usb, keep it load balanced, and at the end, rebuild the database. to reduce back-and-forth hdd head movement, reserve each hdd for a seperate channel on dual drive systems.

what it would take is a rewrite of (j)emplode and probably the player software to accept more than one connection to use both at once, and i'm not sure if there would be any USB limitations... but if there is, the usb-ethernet emulation project would take care of it, right?

comments, ideas, or attempts to blow this pipe dream away welcome =)

I believe this has been discussed before, and the answer was that you're not going to get much speed benefit from doing something like that, because limitations in the player's hardware architecture make it so that you'd start to hit speed bottlenecks in getting the data to/from the IDE drives. Anyone have a link to that old thread on this topic?

i remember that a USB2 conversion discussion where altman says that you cant saturate the bus... but usb2 will make it go faster than ethernet. but it leads me to believe that the bus can handle at least 2x the speed its going right now.

As Tony says, this has come up before at least once. The answer is almost certainly that neither Ethernet, nor USB, nor IDE in an Empeg has DMA hardware, and that the current best-case transfer of about 700Kbytes/sec is thus CPU-bound from all that PIO. However, I don't remember writing anything in the car-player that disables USB when Ethernet is active, or vice versa, so the Jemplode folks could possibly try it out without any player changes if they really, really wanted to. There could be other reasons it won't work, though.

Peter

USB generates an interrupt every 64 bytes (ie, one packet). USB could *probably* be bumped up to getting almost 1Mbyte/sec with a different protocol and a FIQ interrupt handler - but at the moment, USB is speed limited by the number of interrupts that have to be processed, and the interrupt load when disk transfers are going on.

Maybe I need to talk to Patrick about a USB2 mainboard swap for the empeg... we could do a new board with the latest DSP, an Xscale, 100mbit ethernet and USB2

Hugo

Maybe I need to talk to Patrick about a USB2 mainboard swap for the empeg... we could do a new board with the latest DSP, an Xscale, 100mbit ethernet and USB2

Screw it, why don't you guys just release empeg Mk3 kits at that point!

In reply to:

---

Maybe I need to talk to Patrick about a USB2 mainboard swap for the empeg... we could do a new board with the latest DSP, an Xscale, 100mbit ethernet and USB2

---

Hmm,
That almost smells/sounds like a repackaging of the Pearl mainboard given all those features....

In reply to:

---

Maybe I need to talk to Patrick about a USB2 mainboard swap for the empeg... we could do a new board with the latest DSP, an Xscale, 100mbit ethernet and USB2

---

... while your at it why don't ya throw in a color lcd and built in gps?

Maybe I need to talk to Patrick about a USB2 mainboard swap for the empeg... we could do a new board with the latest DSP, an Xscale, 100mbit ethernet and USB2

Oh, talk about a major tease. Everytime I think about what could make the empeg better, it usually boils down to "but a mainboard swap could be so much easier".

If that was even a semi-serious thought, I'd like to put in some requests now

More memory.
Removable flash drive capability for userland apps- SDcard or similar.
USB2 *host* capability.
Buffered external display capability.
Bluetooth (Hey someone else asked...)

Removable flash drive capability for userland apps- SDcard or similar.

But if you've got USB host, you can attach your own flash reader...

I suppose that's true...

Maybe I need to talk to Patrick about a USB2 mainboard swap for the empeg... we could do a new board with the latest DSP, an Xscale, 100mbit ethernet and USB2

How long would a kit like that take to solder up anyway

How long would a kit like that take to solder up anyway

How good are you at soldering 256 pin micro-BGA (Ball Grid Array) packages?

In reply to:

---

How good are you at soldering 256 pin micro-BGA (Ball Grid Array) packages?

---

Only more reasons to release the Empeg MK3! :-)

* more memory
* 100Mbit ethernet
* USB2 host
* External display
* color LCD display
* Wi/Fi and/or bluetooth
* build in tuner
* Build in S/PDIF
* build in DVD player
* lit buttons/knob
and on and on and on....

Your describing Camelot (more or less)

In reply to:

---

Your describing Camelot (more or less)

---

Got a link or something?
I've done a little bit of searching but can't find anything....

I'm hoping it's not double din and could be fitted in the exsisting housing for mk2...

Did a bit of searching eh? Did you try the keyword "Camelot"??
This was one of the first things I came up with...

http://empeg.comms.net/php/showthreaded.php?Cat=&Board=empeg_general&Number=133284&page=&view=&sb=&o=

Damn! You beat me to it!

Your describing Camelot.

It's only a model...

Matthew

And at least 4 years away...

And if you have USB Host you cannot attach to a PC

What you want is a USB OTG port (and yes it can run at high speed)

ROFL Good call, Patsy.

ROFL Good call, Patsy.

At the San Fran dinner with Rob and Toby recently, it was explained that the "It's only a model" line was, in fact, why the product was code named Camelot.

In reply to:

---

Did a bit of searching eh?

---

Euh... only did some searching on google thinking it was an product in mass volume production....

just giving you a hard time....

Come on guys, Hugo's throwing us a bone and you guys are asking for the moon! How about asking about some features that would keep us under the $2000 dollar mark and not require a whole new software build.

Improved DSP w. Bass and Treble
SubOut
Digital Audio Out
100base Ethernet
USB2.0
Higher res or larger size display?

I'm just guessing here...., but what would be some features that would really help the product without adding capabiliies that only 2% of the user base would use or that would shoot the costs up so high it would never get made.

Improved DSP w. Bass and Treble

Um, the current DSP has bass and treble, it's just that no one's implemented it in the software yet.

Improved DSP w. Bass and Treble

Bass and Treble could be implemented on the exising DSP. But I would like to see a true multichannel DSP so we could set independant delays on all speakers.

Bass and Treble could be implemented on the exising DSP

You were too slow today, Grasshopper. We shall try again, tomorrow.

<cue Kung Fu music>

Oops, sorry. I knew that, but didn't write it clearly. I was under the impression that there was some reason it wasn't turned on in the empeg software. I think I read once that with the great EQ we have, the development team thought there was no need for a simple treble and bass adjustment. So, what I meant to say was "Updated DSP (only because Hugo mentioned it) and while you're at it, enable bass and treble adjustment" but that didn't fit into a list format .

That was a very nice humble pie I just ate

Hugo's throwing us a bone

Actually I think Hugo is just taunting us for his own amusement. Of course he is more that welcome to prove me wrong

That was a very nice humble pie I just ate

I honor your humility with a new avatar, Grasshopper.

lol. nice.

Thank you. That one took some research.

In reply to:

---

Um, the current DSP has bass and treble, it's just that no one's implemented it in the software yet.

---

Alright already! I've been meaning to and will definitely start working on it tonight. It's not quite as simple as you would think though. Possibly the loudness control might get a bit munged in the process. Would people be willing to ditch the loudness (or have it work differently, possibly not correctly) in favour of bass and treble?

Not to say I can't.... but I'm kinda covering myself

I was just kidding about the gps and color lcd, I'd be happy with a 100base ethernet only

BTW... Love the avatar Tony...

Wow.

You really have lost a lot of weight!

Would people be willing to ditch the loudness (or have it work differently, possibly not correctly) in favour of bass and treble?

Yes. In fact, if we could also have some control over the slope and cutoff point of the loudness control, that would be ideal.

When playing with the RTA stuff, I noticed that the loundess control was only really affecting the deepest of the sub-bass frequencies. It had a very sharp frequency peak in the near-subsonic range, and not much above that. I'd rather have it more smoothly increase all the low frequencies instead of just the sub-bass.

You really have lost a lot of weight!

Mmm. Hair, too.

Well it seems the loudness may not be affected by the bass and treble after all. However the FM bass boost will but I don't see that being much of issue with peope.

I'm having a few minor issues with calculating some of the DSP coefficients but once I sort that out we should be good to go.

As for affecting the slope and cutoff of the loudness, that's probably a separate project.

Also what kind of frequency points are we looking at here? Examples in the manual are 125Hz and 5kHz. Since it basically needs to be pre-calced, we can't have a variable frequency unfortunately. -6db to +6db enough? I only have 4 bits each to play with using genixias code, although Mark said he would give more bits once bass and treble were done properly.

ISTR that somebody wanted +/-9dB range, and somebody else wanted 0.5dB resolution. Can't win

I'd recommend starting by duplicating the existing range though. It's hard to say what values to use though - the existing EQ fudge is based upon center frequencies and Q factors, whereas the DSP algorithm is based on cutoff frequencies and order (2nd or 3rd IIRC). But I think off the top of my head that a 2nd order at the listed values should be a reasonable approximation to the existing controls. Let me check...

Hmm...

Bass is currently Fc=100Hz, Q=0.2
Treble, Fc=10KHz, Q=0.2

Q figures look wrong (always did) but sound right. Go figure.

Maybe, maybe not. Things sometimes move faster than expected

Hugo

Well, yes and no. Camelot was also the name because "it's the holy grail" (of in-car mp3 players). If you connect to a camelot from emplode, the player icon *is* a castle facade with a very obvious little triangular support keeping it upright.

Hugo

Ahh... I was wondering what the castle icon was for in Emplode's resources. Makes sense now

- Trevor

OK, I just had to see it.

Absolutely priceless....

This IS a really silly thread.

mmmmmh - castle - nice