Hi, Has anyone out there got first hand experience of working with the PXA270 and dealing with Marvell? I have a customer who is considering the PXA270 for a new design and from what I've seen on the Marvell web site (i.e. total lack of datasheets or information on development boards etc.) and read in other discussions here (apparently you have to sign an NDA to get datasheets), that may not be a good choice for a small company. Any advice would be welcomed. Thanks.
Thanks for your replies, PeteS and Mateusz. I probably should have been more specific about my question. I'm not too worried about getting hold of datasheets (as you mentioned they can be found on the web) and I know that there are plenty of good development kits and single board computers out there based on the PXA270 (we are planning to use one of them). My concern comes after we have prototyped the system using an off-the-shelf board (e.g. Arcom Zeus) and then we want to design our own board, what sort of support will we get from Marvell and how long will the PXA270 be around?
The one time I considered a product of theirs (early this or late last year) and was naive enough to ask for data, I was handed a questionaire which included questions like "how much is your yearly revenue". Who do those morons think they are, the tax authorities or what. Probably best to avoid them like the plague if there is a reasonable alternative (I don't know the particular part you are considering).
Not anymore. It seems Intel deleted the application processor stuff from their servers. I think the 5year support period ended recently. Documentation can be found on some old intel mirrors. e.g.
Our NDA experiences with Marvell are: They are responding very fast, but we didn't sign. We stopped considering any of the new Marvell-only PXAs because of this strange company policy. I personaly think customers should show Marvell, that such behaviour is incredible annoying and leads for small development companies to either a significant risk or a large expenses for legal advice. Especially at the first stage of evaluating target plattforms for a future design.
that makes so far 3:0 against Marvell :-). I am not surprised, my nose proved right again. Actually it did not take much of my nose (which I do have for hollow companies).
There seems to be a growing number of people thinking they can buy and control the world given the enormous amount of cash they have access to - which might be generally true _if_ they had the brains to do it _and_ "enormous" would be "sufficient"... (it probably takes extra brains to evaluate that, I guess).
I was involved in a project using one of their wireless products, and my experiences with Marvell were entirely negative. I would never work with one of their products again, except under duress. We dropped plans to upgrade a PXA260 product to a PXA270 when Intel sold the line to Marvell, and decided to use NXP's LPC3180 instead (the vector floating point unit in the 3180 helped the decision, since our app is FP intensive).
I'm finishing up a project that uses the PXA270. In hindsight, I wouldn't touch any embedded processor from Intel or Marvell again. I'm a little concerned about Intel's memory products too; They've been selling off those lines too.
We started our project when Intel owned the PXA. We quickly learned that with a volume of several thousand per year, Intel tech support was insultingly silent. And I mean literally silent. Make a call, leave a message, no reply ever even after repeats. Send an email, no reply, even after repeats. It's like talking to the wall.
We were naively hopeful when Marvell took over. We started calling Marvell after the transition and they told us they don't know anything about the PXA270 - dead end. My several dealings with Marvell on similar matters taught me they don't know anything about anything.
Both companies are cut from the same cloth. If your project isn't some insanely huge revenue stream for them, you'll get more help if you talk to a rock.
As for the PXA270 itself:
the chip basically works which isn't significant
it has a few problems that makes QNX run a little too slow
the docs are huge, verbose, tough to index, have significant errors
errata list is excessively long
Conclusion: There are better choices for most of us.
"Tom Lucas" wrote in message news: email@example.com...
W, too, used a LogicPD board to get started. But we very quickly outgrew it. It didn't really do much for us except verify our tools worked and some idea of performance. I don't recall it providing any useful code or examples. A few years back though,
Well the Logic Loader was quite useful to get off the ground and prove the hardware. Sharp provided some good example code but the Lolo didn't come with much. However, the user guides and other documentation were plenty good enough to begin writing code for use on top of Lolo.
I moved on to using my own OS fairly quickly because the overhead of the Lolo was a bit much the performance I needed but I could have used it to boot my code if I'd wanted to and it provides lots of good ways of getting a program into the board if you didn't have JTAG probe.
We use fairly low volumes so for us the card engines are more cost effective to buy in than to design ourselves. Plus their support is very good and they are nice people as well.
Perhaps likely if the rock was a sentient silicon-based life form.
Many of us at one time or another have pondered the nature of consciousness; that internal 60 cycle background hum, that feedback loop that compares the recent internal state with the current state and projected future state. Some researchers have proposed that all processing systems can be considered to possess consciousness on a continuum, and that perhaps all biological and perhaps physical systems are also candidates as well; not in some neo-animist sense but in a definition involving organization and complexity.
As a kid, I used to wonder how the IBM 1620 felt when it was being shut down at the end of a many-days long silver isotope half-life computation run. I would appreciate your thoughts on this subject as well as any personal insights.
My Maxim experience is limited to only small quantities, not large enough (as yours apparently have been) to run into delivery issues.
Apart from that Maxim are by far not as bad as Marvell - and they are certainly not hollow, they do develop their products for a long time (best comparators - by a huge margin - good refs etc stuff). They are more of a pain to deal with than ADI (but almost anybody is that), and perhaps more of a pain than BB (OK, TI), but I would take their product if it is superior (hopefully I am not just giving you the chance for "I told you so" before we know...:-).
I felt I had to give them credit for being helpful when I asked for help (sending me more samples when they could not sell me small quantities etc.).
I have dealt with Marvell on products other than the PXA series and I am joyfully under no NDA on those other products.
They remind me strongly of the attitude (strongly denied of course) at Maxim where if you aren't going to buy a million or so parts, you can just go and wait outside until we decide to make some (because someone else *is* buying a million) and we might give you support if we feel like it (or we'll charge for it because we aren't selling you a million parts - they charge by raising the price to eye-watering levels, of course)