Home arrow Support arrow Forums

Luminary Micro Forums

<< Start < Prev 1 2 Next > End >>

rocksoft

Senior Boarder
Click here to see the profile of this user

2008/05/31 05:22

Again...more ram!

I have developed a board with the LM3S6569, I think this is a great little device. However it's application base is limited by the amount of RAM. I think this device with 512k or more ram and 64kb flash would be just right. External SPI flash is reasonably cheap and fast enough to swap code in and out of ram. The current amount of ram limits network stack performance...

..or, of course, make a device with an external bus... I think a generic device with some of the I/O mappable to external address space would be a good move...

I actually don't use most of the peripherals for the specific functions I'm really just using it as an embedded ethernet device and using the I/O pins for some external SRAM would be very good. Of course I could do this in software but the speed of this would limit it's usefulness.

Just a thought.

Liam Redmond.

login or register to reply

cb1

Platinum Boarder
Click here to see the profile of this user

2008/05/31 08:15

Re:Again...more ram!

Like both your idea and the manner in which you detailed/justified.

My firm is nearing the completion of a low cost device which talks to LMI devices via an 8 bit parallel bus - and provides ALL of the glue-logic to directly attach various sizes/types of economical Sram. Plan to introduce 1st as a small "daughterboard" - later will provide either/both custom ic and ram for your own population.

Key appeal of this approach is that ALL LMI devices can employ large, bulk Sram and optimal transfer rates. (plus several other, desirable "goodies.")

login or register to reply

rocksoft

Senior Boarder
Click here to see the profile of this user

2008/05/31 10:11

Re:Again...more ram!

Sounds like a good product and I'm sure many people will be interested in it when it's ready. No doubt you'll give us all a "heads-up" when it ready.

I have an SPI flash device fitted at the moment and plan to experiment with compatible FRAMs and SRAMs of the same footprint to try and implement a virtual memory system, for emergency situations!

But this would never be anywhere near as good as a parallel bus arrangement so I think your product will be useful.

It's a shame the devices don't have native bus interfaces though!

login or register to reply

cb1

Platinum Boarder
Click here to see the profile of this user

2008/05/31 15:37

Re:Again...more ram!

We faced this "no external bus" problem
several years back with ARM7 devices. Eventually makers responded - with huge pin-count devices. And guess what - while "all in one" package is desirable we were never able to equal the performance of our custom, Sram-Control ic! And our ic frees the micro from having to "step" addresses - while enabling unique Sram navigation. (important in several growing applications)

Lastly - many applications do fine without large Sram - internal or external. Our approach enables LMI users to purchase & master fewer devices - speeding & reducing development efforts & costs! Focused buying saves you still more - enabling volume price discounts, too.

login or register to reply

jrmymllr

Platinum Boarder
Click here to see the profile of this user

2008/09/11 06:10

Re:Again...more ram!

I second (or third?) the request for more RAM, although I'd prefer it onboard. I have found that the 6965 is very unique in that it contains an ethernet PHY, full speed flash, and as much RAM as you'll find anywhere in this class. The only real exception is one of the Atmel ARM7 devices has 128K, but no PHY and slower flash.

For my application, 64K RAM is just OK, but 128K would be great.

login or register to reply

cb1

Platinum Boarder
Click here to see the profile of this user

2008/09/11 09:42

Re:Again...more ram!

jrmymllr-

Of course "on chip" is best - but in its absence could you live with an asic-sram combo which you "talk to" via 8 bit parallel or spi? My firm has automated the addressing process so that you "just" pump data after assigning a starting address. Believe that my asic can out-perform the data transfer rates you can achieve via parallel or spi. Is this of any interest?
(also - we produce extreme contrast TFT and graphic mono Lcds - I would like to provide one for your I2S project - gratis of course)

login or register to reply
<< Start < Prev 1 2 Next > End >>