ARM, Single-Chip PC, or other architectures for audio stuff?

Do you need floating point support?

Have a look at Renesas RZ series. Very promising parts: Cortex-A9, 400MHz, up to 10MB(!) on-chip SRAM, Double Precision FPU, LQFP.

But, samples only for now. Full availability somewhere this year?

We were looking for fast double precision floating point, and this one is one of the few 'smaller' chips that might have a chance meeting our performance requirements (need to test). Beware of 'light' versions of the FPU in some Cortex processors, that takes 10 times more clock cycles for a math operation than a 'none light' version.

--
Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail) 

Clothes make the man.  Naked people have little or no influence on society. 
 Click to see the full signature
Reply to
Stef
Loading thread data ...

I wonder if you looked at the Intel Quark SOC used in the Galileo board:

formatting link

Reply to
Paul Rubin

Windows may well have better audio latency than Linux. It all depends.

--
Les Cargill
Reply to
Les Cargill

And, by design, Xmos definitely has a much better jitter than either.

Reply to
Tom Gardner

No, had not found that one, thanks for the pointer.

Looks a bit more difficult to use than the RZ, but haven't studied it in detail yet. And ofcourse there's no 10MB on chip SRAM.

Also 400 MHZ, and the FPU seems promising:

The internal floating-point unit performs floating-point operations on the 32-, 64- and 80-bit arithmetic formats as specified in IEEE Standard 754.

Certainly worth a closer look, thanks.

--
Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail) 

modem, adj.: 
 Click to see the full signature
Reply to
Stef

Vaporware.

Mouser shifted expected delivery again now to 20. Feb. First exected delivery was Nov 13...

--
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de 

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt 
 Click to see the full signature
Reply to
Uwe Bonnes

In stock here, with reviews from customers who have received them:

formatting link

Reply to
Paul Rubin

All I can tell you is that we have FAE out a lot and I'd expect they would immediately refuse any cooperation if they'd see a use that violate their company policies.

[...]

A Windows-based single-chip or tiny-board computer would also work, and might even be ideal. No matter what one might say about MS, stuff is generally backwards-compatible over decades. I can still run all my DOS programs from the 80's.

:-(

[...]

Yup, and stuff lives really long there. 30 years or more. or in the case of the DC-3 it can easily be 70 years.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Well, we have Intel really close by and they just announced a layoff. One huge Campus is only 10 miles from here. So maybe I should be looking at a tiny board with an Atom processor :-)

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

The project would easily run more than half a year. Relocating someone would be rather expensive, especially if they must maintain their usual (freaking expensive) residence in the Bay Area.

[...]

Not really. The LPC4300 series is available in gullwing QFP.

Regarding SMT we were early adopters in the mid-80's. But I did stay clear of some of the not-so-great "advances" in technology such as pressfit connectors.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

That would be ok. What is your experience with their support? Paul didn't sound too cheerful about it.

That is what I liked so far about the M4 from NXP: Supposedly it does a multiply-accumulate in one clock-cycle, meaning about 5nsec when run full bore.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

None so far, haven't used Renesa cpu's before. If we use them (we are still in the process of finding the right cpu) we will have support through our disti (Silica in this case). Have had good experiences with them in the past (for other products).

I did not see any comment from Paul on the Renesas part, he only came with a possible alternative. Did I miss something?

That is integer only, no problem if you're doing integer ofcourse. The ARM7 already supported this (in 1 or 2 clocks IIRC) with the SMLAL instruction (32*32+64=64). I still remember that as we could not get the compiler to emit that specific instruction at the time and had to code the filter routine in asm.

--
Stef    (remove caps, dashes and .invalid from e-mail address to reply by mail) 

In the misfortune of our friends we find something that is not displeasing 
 Click to see the full signature
Reply to
Stef

Sacramento isn't that bad a drive from the bay area. Not ok for a daily commute, but I could imagine someone just getting a cheap room/craashpad in Sac. for the duration of the gig, staying there on weeknights and driving back home for weekends. I did something like that (different location but same idea) a while back and it worked fine.

Reply to
Paul Rubin

It was not Paul Rubin but Paul from the UK, in a post yesterday.

It'll be integer in our case. The M4 would be overkill but I've seen too many situations where I was assured that a uC would handle all the algorithms that I could possibly wish for and then it was like trying to cram 15 people into a Fiat 500, even after paring down my algorithms.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Yes, that would be possible. From here to, say, Mountain View it can be anywhere between 3h and 5h one way. It's crazy. I am so glad I don't have to live there.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

I'm very surprised and think you should try harder. They simply could not have created such an open sharing community without support, even if their web team are shite - there must be good support folk, you just have to find the right entry point.

Anyhow, I place more importance on community support. A good community means that a lot of folk have succeeded in figuring things out, with or without good vendor support. It speaks volumes.

Clifford Heath.

Reply to
Clifford Heath

But what is that entry point? Not a forum but real FAEs. I find it strange business behavior if a contact email is publicly stated and then nobody bothers to even answer. It also did not bounced, just went into a big garbage pile. Not only once. All they'd have to do is point the requester in the right direction if the published amils address turns out to be the wrong path.

I place more importance on corporate support. I often work in very regulated business and there one must have iron-clad design history. A discussion thread in a community does not survive an audit while a written or emailed statement by an engineer of the manufacturer usually does. Also, many of my questions can really only be answered by the guys who did the IC design. Radiation tolerance levels and stuff like that.

A week ago I had that with a switcher IC mfg. Calling them resulted in a taped message that they do not provider support over the phone. At that point I'd have kicked out the chip but it was not my call. They only offered a forum where I got no answer for one question and non-helpful ones for another. Only until I rattled the cage (something I then tend to do) did I get through to a senior FAE who could answer my question within minutes. Before, we had lost several days because of this restrictive support access. I prefer companies that don't place such restrictions.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

I meant "non local" the way it is done nowadays, you may never physically meet. Unless the programming of the thing would take the programmer constantly pressing/turning knobs at your side which cannot be pressed/turned remotely you don't really need him there.

You will probably find some stuff but your options do get narrowed a lot. Then using a tiny - 0.8 or even 0.5mm pitched - BGA will definitely be mechanically better than the qfp (much lower mass to accelerate). Like in the phones nowadays, you know - the housings and displays may crack but I haven't heard of a BGA chip getting detached without breaking the board or sort of.

Dimiter

------------------------------------------------------ Dimiter Popoff, TGI

formatting link

------------------------------------------------------

formatting link

Reply to
dp

That is how I typically work. But it's different on this project.

We need to work together on a large tester with fluid, pumps, machines and hoses. We have to look at subtle effects together, discuss possible ways to combat those in software and try it out right then and there. This is why the person also would need to have a generalist attitude, not a "That's not my turf, I am a programmer" behavior. Interestingly, with most SW folks that I know this is not a problem.

Sometimes one has to suffer for good reasons :-)

It's not about mass, it's about flexing of circuit boards at the moment when the unit hits a tile floor or is otherwise being banged up. A flexible substrate with a large non-flexible part on there is generally not a good idea. Small BGAs are usually ok though.

I have, lots of times. A whole niche industry has sprung up where individuals set up show and re-ball, re-work, whatever when a video game console or some other device has become erratic (usually after a fall). Don't know this company, just an example, it seems there's whole little industries evolving from that trend:

formatting link

A whole chip won't become detached. All it takes is one ball to become iffy. Then you have effects like a device that sort of works again when you apply pressure here or there.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Have to re-qualify that: If the substrate is very flexible such as with a Kapton flex circuit, BGAs can be perfectly ok. But that's expensive.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.