Sign in

username:

password:



Not a member?

Search fpga-cpu



Search tips

Subscribe to fpga-cpu



fpga-cpu by Keywords

Altera | CISCifying | IDE | ISA | Java | JHDL | JTAG | LBU | MicroBlaze | PAR | PCI | RISC | SoC | Spartan | Transputers | Verilog | VHDL | Virtex | VLIW | WebPack | Xilinx | Xsoc | YARD-1A


Ads

Discussion Groups

See Also

DSPFPGAElectronics

Discussion Groups | FPGA-CPU | Re: FPGA routing - was - RE: Re: POP-11 (PDP-11/40 in an FPGA)


Advertise Here

This list is for discussion of the design and implementation of field-programmable gate array based processors and integrated systems. It is also for discussion and community support of the XSOC Project (see http://www.fpgacpu.org/xsoc).

Re: FPGA routing - was - RE: Re: POP-11 (PDP-11/40 in an FPGA) - woodelf - Oct 16 21:42:36 2007

Austin Franklin wrote:

> I'm not sure how long it's been since you've done any FPGA work, but though
> that was true up through the 3k and less true with the 4k series of Xilinx
> parts, the later serieses have far more routing resources, and routing is
> typically not an issue, especially for complex designs.

Why does everybody assume Xilinx is the *only* brand of FPGA's. I used
Altera 10K.

> Even in the 3k and especially the 4k, if you were savvy with the design
> tools, you could implement efficient placement (and mapping of logic) so
> that you could get very full designs, upwards of %98 in my experience.

I doing my new project in CPLD's (amtel's) since they don't require a ROM boot
and I can get them 84 PLCC packages, and program them with a demo board I got for $99
a few years back. I get about 80% routing on the CPLD's.

> Regards,
>
> Austin
Ben.

To post a message, send it to: f...@yahoogroups.com
To unsubscribe, send a blank message to: f...@yahoogroups.com



(You need to be a member of fpga-cpu -- send a blank email to fpga-cpu-subscribe@yahoogroups.com )


FPGA routing - was - RE: Re: POP-11 (PDP-11/40 in an FPGA) - rtstofer - Oct 17 0:58:36 2007


> Why does everybody assume Xilinx is the *only* brand of FPGA's. I used
> Altera 10K.

I certainly prefer the licensing terms for WebPACK ISE over the Altera
equivalent. Part of the Altera terms is that you get an annual
license revocable at will. I like the idea that, with Xilinx, what I
get is what I get. I may never get an upgrade but at least the
license won't get yanked.

I went through that debacle with UCSD Pascal when the university
decided to yank back all the licenses. I don't want to run into that
again.

So, I give Altera a pass. I also like the price of the Spartan 3
Starter Board ($150 for 1M gates) and Digilent has a nice product
line. www.digilentinc.com

I have looked at some Altera development boards but nothing really
caught my attention in terms of capability and price.

Richard

To post a message, send it to: f...@yahoogroups.com
To unsubscribe, send a blank message to: f...@yahoogroups.com



(You need to be a member of fpga-cpu -- send a blank email to fpga-cpu-subscribe@yahoogroups.com )

RE: FPGA routing - was - RE: Re: POP-11 (PDP-11/40 in an FPGA) - Austin Franklin - Oct 17 9:32:55 2007

Hi Ben,

> Why does everybody assume Xilinx is the *only* brand of FPGA's. I used
> Altera 10K.

I didn't assume anything, I simply stated what my experience was with
specific Xilinx families in relation to routing problems.

Regards,

Austin

To post a message, send it to: f...@yahoogroups.com
To unsubscribe, send a blank message to: f...@yahoogroups.com



(You need to be a member of fpga-cpu -- send a blank email to fpga-cpu-subscribe@yahoogroups.com )

Re: FPGA routing - was - RE: Re: POP-11 (PDP-11/40 in an FPGA) - Charles Steinkuehler - Oct 20 6:10:14 2007

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Austin Franklin wrote:
> Hi Ben,
>
>> Why does everybody assume Xilinx is the *only* brand of FPGA's. I used
>> Altera 10K.
>
> I didn't assume anything, I simply stated what my experience was with
> specific Xilinx families in relation to routing problems.

Xilinx isn't the only...just the first.

My history with FPGAs and routing:

Xilinx:
- -------
The Xilinx 2K series worked (routed) pretty well. 90+ percent
utilization, although that wasn't all that much in today's world (100
flip-flops in each and every part!), and we were mostly hand-routing
with the chip editor.

The Xilinx 3K series had far too little routing resources (basically,
the resources of the 2K seris, but more logic), and could not get past
70% capacity or so for general purpose designs...even worse in the
larger parts.

In the Xilinx 4K series, they over-compensated, and you could route just
about *ANYTHING* in the initial 4K parts. Near 100% logic utilization,
and you still had most of the routing available! They fixed that with
later cost-reduced 4K series parts, which are more typical (80-ish
percent utilization before routing starts to get tight).

I haven't done any serious work with the newer Xilinx parts (Virtex,
Spartan, etc), but in the simple designs I've worked on with these parts
I haven't run into any issues.

AT&T/Lucent/Lattice:
- --------------------
We switched to the 1C series parts when the Xilinx 3K line didn't get
the job done. These parts *ROCKED*...95+ percent utilized with little
or no performance degradation due to routing. Toolchain based on the
NeoCad environment, IMHO the best FPGA design environment yet (until
Xilinx bought them and killed the product to prevent designs from being
easily ported across vendors).

The later 2 and 3 series parts continued the excellence of the initial
family, although I never pushed them quite as hard as the 1 series (they
were bigger & faster, so I didn't need to).

I haven't done any serious designs with this family since Lattice took
them over.

Altera:
- -------
My initial introduction to Altera came when switching jobs. I had been
doing big, complex designs in Lucent Orca parts and took over a 10K
based design. The 10K family /sucked/ by comparison...slow,
insufficient routing resources, and no distributed memory. The 10KA and
10KE are a little better, but not much. Starting with the Stratix
family, the Altera parts start to become a lot more usable. Usage above
80% is possible without serious routing induced speed delays. I'm
currently working with the S1GX family, but have dabbled some with the
S2GX parts, which look to be as good or better from a routing perspective.

Actel:
- ------
My only serious work with these parts was in the mid '90s. They worked
well, but the anti-fuse parts were awkward to use (program once, test,
then throw away). Otherwise, they routed fine, had decent speed, and
were critical for anything you couldn't easily do in a PAL/CPLD that
needed to be up at system boot time (like the boot and glue logic for
the CPU that configured the SRAM based FPGAs!). :)

All in all, I wish the ORCA family had taken off a bit better. Those
were my favorite parts to work with, the toolchain in the mid-90s was in
many ways better than what's shipping today from Altera/Xilinx, and I
always like to root for the underdog. :)

Nowadays, it mostly seems like an Altera/Xilinx world. As with
Democrats vs. Republicans, two choices doesn't seem like enough to
guarantee a good result! :-)

- --
Charles Steinkuehler
c...@newtek.com
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHFjBAenk4xp+mH40RAk3SAJ9Xr7gPoeFbhvcRmFN4FrTCtaAeKACfcUuI
ZPIyYVpJeBblALZxRwVX1X8=
=wuhu
-----END PGP SIGNATURE-----
To post a message, send it to: f...@yahoogroups.com
To unsubscribe, send a blank message to: f...@yahoogroups.com



(You need to be a member of fpga-cpu -- send a blank email to fpga-cpu-subscribe@yahoogroups.com )