Lattice XP2 finally announced

Maybe that's because Lattice only announces devices when you can actually buy them and not 12 months before a "normal" customer can get his hands on some engineering samples? :)

cu, Sean

--
My email address is only valid until the end of the month.
Try figuring out what the address is going to be after that...

LOL

actually yes. I have just looked at Xilinx website online shop and all distributors

there is ABSOLUTLY no deliver information for ANY spartan-3A or 3AN device yet. And S3A is not new anymore. Even S3AN is out for some time, still NO BUYING options, nothing nothing nothing. Xilinx has really mis-understood the meaning of "online shop"...

lets hope XP2 devices are readily available, the RAM to flash transfer seems nice feature, also there are again nice small foot print package options offered

XP and S3AN have no good package options, so XP2 fills the gap here

Antti

Another example:

Look in the online shop of the XC3S500E in hand solderable PS2008. Nill, Zilch, Nada...

Then go to digikey.: On stock ...

IS XP2 in some online shop, with online pricing?

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

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

Uwe

If you like something hand solderable for XC3S500E have a look at out Craignell modules

We also have a new version coming that will allow operation down to 3V and going up to 5.5V supply and IO. These new versions are a few weeks away yet. As usual educational and bulk order discounts available and they will be in our on-line shop when the new batch are available.

If you need something more advanced we have Darnaw1 as well.

John Adair Enterpo> Antti wrote:

I Just checked their website, but this device is not yet available on their store. XP however has online pricing, and availability for some of the parts. I suppose it is just a matter of Mouser (powering the web store) buying the parts.

Luc >Antti wrote:

I already have a board with XC3S500-PQ208 running, bought at digikey.

-- Uwe Bonnes snipped-for-privacy@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt

--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

I already have a selfmade board with XC3S500-PQ208 running. The 500E was bought at digikey. Programming is done via USB by a FT2232 and a modified xc3sprog

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

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

My FAE, who is a good chap, assures me that the S3A is just about there

- approx 6 weeks before the parts I am interested in get into distribution.

No doubt Steve, who is also a good chap, will jump in with the authoritative picture.

Tim,

Unfortunately, Steve Knapp has left Xilinx to pursue an opportunity that intrigued him. We will miss him.

As far as I am aware, the Spartan 3A, 3AN rollout is proceeding just fine. There are ES available now, with production coming very soon.

Distributors will not stock ES, as they can not return it.

Peter and I have made some suggestions (like accepting unsold ES back from distributors), but I don't know how far that goes (what may work for LX330T ES, may not work for 3S400A).

In any event, we are still trying to work on the "store."

As always, if anyone has an issue with getting parts, Peter and I will intervene and try to find out what the problem is.

If you are willing to place an order, the distributor will order from us (immediately), and your parts will be on their way. The only case where they may not be on their way is if you have ordered production parts, and that part is not in production yet, or you have ordered some package/part/RoHS combination that needs to be assembled (because no one is ordering it).

Be very careful about leaded, vs "un-leaded" (RoHS), as we can not ship any non-RoHS to the EU any longer (the fences are up!).

Austin

OOPS!

As usual, I was a bit too low key:

"The Spartan-3A family is now released to production two months ahead of schedule. The entire family is shipping on all speed grades, all temperature grades, and all part and package types! We have production volumes on the shelf now with over XXX,XXX units in the pipe today. Lead times are short and order entry is open."

Also, of some interest:

"Not only was the power reduced on the Spartan-3A family but we were also able to reduce the power numbers across the entire Spartan-3 Generation. In both Spartan-3 and Spartan-3E we have made reductions across the devices that are as much as 50% improvement in our power numbers. On the Spartan-3E platform, the changes have been made to the data sheet already. By mid-May, you will see the changes made to the Spartan-3 data sheet."

Enjoy,

Austin

That power note is a little unclear. Is it a decrease in the MAX column, from better testing data [so no change in typicals], or is it a process change, that has resulted in both MAX, and TYP values decreasing.

Also, how does such a change affect the precision of mA/MHz power-models ?

-jg

Jim,

Basically, the entire Spartan 90nm process node has benefited from UMC getting to the point where we are in really good shape on leakage.

As you know, speed and leakage are not something that can be separated easily. Make it fast, it gets leaky. Make it slow, it gets less leaky.

Through a lot of just plain old hard work, the process is so stable (and varies so little), and predictable, that we can specify a much lower static power level.

When you are a large enough company to interest your fab in meeting your needs, well, very good things happen as a result.

If you are too tiny, and not able to command the fab's attention, well, then you "gets what you gets." Basically, whatever comes out of the fab, you either pay for it, or scrap it.

Today at DAC, I heard nothing but good stories about UMC - about how good they are. Yes, UMC is part of Xilinx' success. We could not have done it without them. But, Toshiba is also now part of that success, too. We are careful to approach those fabrication facilities that have the best possible reputation, where being a supplier to Xilinx is something they are actively involved in, consciously working on every day.

So, just when our competition announces their "super low" static power numbers, we also drop our specification to make their announcement unimportant. Magic? No, just good old hard work.

I feel sorry for the other FPGA vendors: they have no ability whatsoever to get anything from their fab partner, other than whatever their partner is presently making money from. So, if cell phones are the big money maker, you can have a cell phone process.

If microprocessors are the product, you get a microprocessor process.

That works for a cell phone company, or a microprocessor company, but it is really tough on a FPGA company.

One year, one month, and counting.

Austin

Thanks Austin, but in all the above, you seem to have overlooked answering my questions :)

-ie have UMC changed the process since you started, ( which means newer silicon is measurably different from older silicon ) ? or did you start with more spec-headroom on MAX values (as one does, on any new process), but the process itself has not actually changed, and thus the typicals are also unchanged, but the better experience allows the MAX to be decreased ?

MAX can also be decreased with some marketing push, where some yield rejects are allowed, to hit soms magic number. Doing this brings the TYPs up closer to MAX.

Is this change ["as much as 50% improvement"] purely static Icc, or does it impact mA/MHz ?

-jg

Jim Granville,

I thought I was clear. Sorry.

No change to the process. Just tighter, cleaner, process controls.

As you may be aware, as dimensions shrink, it becomes increasingly difficult to control variation. Even on a single wafer, one can experience 6 sigma variation (+/- 3). In effect, all possible process corners may happen on the same wafer, sometimes on the same die.

It is only after tens of millions of shipped Spartan 3 family die that the process becomes well enough known that we are able to take advantage of it, and not lose yield by a tightening of the specifications.

Austin

Hi Austin So that means the MAX column has come down, but the TYP column has not changed ? What about the mA/MHz values ? - Tho I'd imagine they are rather less process dependant than leakage values

-jg

Well, I designed a board 2 years ago to use V4FX40 parts. Is there any sign of them yet? Lol

Jon

If he's reading this, many thanks to Steve for all the help and support over the years. And good luck for the future. If at first you don't succeed, try and try and Triscend again.

I wish Xilinx would have an inventory page like TI has, e.g.

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

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt
--------- Tel. 06151 162516 -------- Fax. 06151 164321 ----------

did Steve work at Triscend?

I felt bad when Triscend was killed.. they had nice new prouduct lineup ready, but then all that got killed. sure some of the features planned for Triscend latest chips are now part of V4-V5, but having one less player in FPGA field is still a bit sad. I think Triscend products (those that never come out) would have had success I am referring here to triscend never announced FPGA+ARM

+ethernet+ADC chips.

Antti