Site Home   Archive Home   FAQ Home   How to search the Archive   How to Navigate the Archive   
Compare FPGA features and resources   

Threads starting:
1994JulAugSepOctNovDec1994
1995JanFebMarAprMayJunJulAugSepOctNovDec1995
1996JanFebMarAprMayJunJulAugSepOctNovDec1996
1997JanFebMarAprMayJunJulAugSepOctNovDec1997
1998JanFebMarAprMayJunJulAugSepOctNovDec1998
1999JanFebMarAprMayJunJulAugSepOctNovDec1999
2000JanFebMarAprMayJunJulAugSepOctNovDec2000
2001JanFebMarAprMayJunJulAugSepOctNovDec2001
2002JanFebMarAprMayJunJulAugSepOctNovDec2002
2003JanFebMarAprMayJunJulAugSepOctNovDec2003
2004JanFebMarAprMayJunJulAugSepOctNovDec2004
2005JanFebMarAprMayJunJulAugSepOctNovDec2005
2006JanFebMarAprMayJunJulAugSepOctNovDec2006
2007JanFebMarAprMayJunJulAugSepOctNovDec2007
2008JanFebMarAprMayJunJulAugSepOctNovDec2008
2009JanFebMarAprMayJunJulAugSepOctNovDec2009
2010JanFebMarAprMayJunJulAugSepOctNovDec2010
2011JanFebMarAprMayJunJulAugSepOctNovDec2011
2012JanFebMarAprMayJunJulAugSepOctNovDec2012
2013JanFebMarAprMayJunJulAugSepOctNovDec2013
2014JanFebMarAprMayJunJulAugSepOctNovDec2014
2015JanFebMarAprMayJunJulAugSepOctNovDec2015
2016JanFebMarAprMayJunJulAugSepOctNovDec2016
2017JanFebMarAprMayJunJulAugSepOctNovDec2017
2018JanFebMarAprMayJunJulAugSepOctNovDec2018
2019JanFebMarAprMayJunJulAugSepOctNovDec2019
2020JanFebMarAprMay2020

Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Custom Search

Messages from 141025

Article: 141025
Subject: Re: Xilinx GbE performance
From: OutputLogic <evgenist@gmail.com>
Date: Tue, 2 Jun 2009 11:51:47 -0700 (PDT)
Links: << >>  << T >>  << A >>
On Jun 2, 9:53=A0am, Antti <Antti.Luk...@googlemail.com> wrote:
> Hi
>
> does anybody have real and realistic performance figures for Xilinx
> GbE solution with XPS_TEMAC/MPMC ?
>
> we need to get 60% of GbE wirespeed, UDP transmit only but it seems
> like real hard target to reach :(
>
> MPMC has memory latency of 23 cycles (added to EACH memory access
> cycle) so the ethernet
> SDMA takes a lot of bandwith already, there is another DMA writing
> data at same speed, and the
> PPC itself uses the same memory too
>
> Antti


Antti,

Take a look at http://www.treck.com
They do high performance TCP/IP
But if all you need is UDP, you don't have to use XPS at all. You can
design custom logic that does DMA from the memory controller directly
into the GbE MAC.
I've done it with very good performance results. Email me for more
info.

-outputlogic

--
visit http://outputlogic.com : tools that improve productivity
--

Article: 141026
Subject: Re: Xilinx GbE performance
From: "Antti.Lukats@googlemail.com" <Antti.Lukats@googlemail.com>
Date: Tue, 2 Jun 2009 12:03:56 -0700 (PDT)
Links: << >>  << T >>  << A >>
On 2 June, 21:47, Jan Pech <inva...@void.domain> wrote:
> On Tue, 2009-06-02 at 11:32 -0700, Antti.Luk...@googlemail.com wrote:
> > On 2 June, 21:05, Jan Pech <inva...@void.domain> wrote:
> > > On Tue, 2009-06-02 at 09:53 -0700, Antti wrote:
> > > > Hi
>
> > > > does anybody have real and realistic performance figures for Xilinx
> > > > GbE solution with XPS_TEMAC/MPMC ?
>
> > > > we need to get 60% of GbE wirespeed, UDP transmit only but it seems
> > > > like real hard target to reach :(
>
> > > > MPMC has memory latency of 23 cycles (added to EACH memory access
> > > > cycle) so the ethernet
> > > > SDMA takes a lot of bandwith already, there is another DMA writing
> > > > data at same speed, and the
> > > > PPC itself uses the same memory too
>
> > > > Antti
>
> > > With custom Ethernet core + MPMC we get data rates slightly above
> > > 100MBps, depending on MTU. The single memory is shared by MicroBlaze/=
PPC
> > > for code and data access, at least one streaming data source (custom =
PIM
> > > for NPI) and the custom Ethernet IP (MAC + some packet composers,
> > > decoders, etc.) again connected to NPI.
> > > We rejected to use XPS_TEMAC because its low performance. The problem=
 is
> > > I lost my benchmark results. Sorry.
>
> > > Jan
>
> > hum.. my current task is to optimize a XPS_TEMAC based system
> > (with 1 single DDR2 chip as main memory!)
> > to reach about 580MBps
>
> > :(
>
> > I have never said that to be possible, but i need money :(
> > and if the goal cant be reached there will be none...
>
> > over 100MBps is sure possible (with XPS_TEMAC too)
> > =A0but 580MBps is beyong doable i think for sure
>
> > Antti
>
> Just a simple calculation:
> 125000000 / 1024 / 1024 =3D 119.2MBps
> It is without protocol overhead, FCS, IFGs. How do you want to exceed
> limit of Gigabit Ethernet?
>
> Jan

ups silly me :(
B is as byte

I wanted say we need 580Mbit/s

Antti




Article: 141027
Subject: Re: Xilinx GbE performance
From: "Antti.Lukats@googlemail.com" <Antti.Lukats@googlemail.com>
Date: Tue, 2 Jun 2009 12:05:40 -0700 (PDT)
Links: << >>  << T >>  << A >>
On 2 June, 21:51, OutputLogic <evgen...@gmail.com> wrote:
> On Jun 2, 9:53=A0am, Antti <Antti.Luk...@googlemail.com> wrote:
>
> > Hi
>
> > does anybody have real and realistic performance figures for Xilinx
> > GbE solution with XPS_TEMAC/MPMC ?
>
> > we need to get 60% of GbE wirespeed, UDP transmit only but it seems
> > like real hard target to reach :(
>
> > MPMC has memory latency of 23 cycles (added to EACH memory access
> > cycle) so the ethernet
> > SDMA takes a lot of bandwith already, there is another DMA writing
> > data at same speed, and the
> > PPC itself uses the same memory too
>
> > Antti
>
> Antti,
>
> Take a look athttp://www.treck.com
> They do high performance TCP/IP
> But if all you need is UDP, you don't have to use XPS at all. You can
> design custom logic that does DMA from the memory controller directly
> into the GbE MAC.
> I've done it with very good performance results. Email me for more
> info.
>
> -outputlogic
>
> --
> visithttp://outputlogic.com: tools that improve productivity
> --

hm

we DO NOT COPY the UDP buffers around at all
they are pre-formatted by custom DMA

so i do not see much where treck would give us any gain

Antti





Article: 141028
Subject: Re: Xilinx GbE performance
From: "Antti.Lukats@googlemail.com" <Antti.Lukats@googlemail.com>
Date: Tue, 2 Jun 2009 12:16:29 -0700 (PDT)
Links: << >>  << T >>  << A >>
On 2 June, 21:50, Jan Pech <inva...@void.domain> wrote:
> On Tue, 2009-06-02 at 20:47 +0200, Jan Pech wrote:
> > On Tue, 2009-06-02 at 11:32 -0700, Antti.Luk...@googlemail.com wrote:
> > > On 2 June, 21:05, Jan Pech <inva...@void.domain> wrote:
> > > > On Tue, 2009-06-02 at 09:53 -0700, Antti wrote:
> > > > > Hi
>
> > > > > does anybody have real and realistic performance figures for Xili=
nx
> > > > > GbE solution with XPS_TEMAC/MPMC ?
>
> > > > > we need to get 60% of GbE wirespeed, UDP transmit only but it see=
ms
> > > > > like real hard target to reach :(
>
> > > > > MPMC has memory latency of 23 cycles (added to EACH memory access
> > > > > cycle) so the ethernet
> > > > > SDMA takes a lot of bandwith already, there is another DMA writin=
g
> > > > > data at same speed, and the
> > > > > PPC itself uses the same memory too
>
> > > > > Antti
>
> > > > With custom Ethernet core + MPMC we get data rates slightly above
> > > > 100MBps, depending on MTU. The single memory is shared by MicroBlaz=
e/PPC
> > > > for code and data access, at least one streaming data source (custo=
m PIM
> > > > for NPI) and the custom Ethernet IP (MAC + some packet composers,
> > > > decoders, etc.) again connected to NPI.
> > > > We rejected to use XPS_TEMAC because its low performance. The probl=
em is
> > > > I lost my benchmark results. Sorry.
>
> > > > Jan
>
> > > hum.. my current task is to optimize a XPS_TEMAC based system
> > > (with 1 single DDR2 chip as main memory!)
> > > to reach about 580MBps
>
> > > :(
>
> > > I have never said that to be possible, but i need money :(
> > > and if the goal cant be reached there will be none...
>
> > > over 100MBps is sure possible (with XPS_TEMAC too)
> > > =A0but 580MBps is beyong doable i think for sure
>
> > > Antti
>
> > Just a simple calculation:
> > 125000000 / 1024 / 1024 =3D 119.2MBps
> > It is without protocol overhead, FCS, IFGs. How do you want to exceed
> > limit of Gigabit Ethernet?
>
> > Jan
>
> Or did I get you wrong and you talk about Mbits per second? I was
> talking about Mbytes per sec.
> If it is so, your goal should be reachable using xps_ll_temac instead of
> xps_temac.
> Jan

oh, i am talking wrong today
yes Mbit/sec or Mbps
and sure XPS_LL_TEMAC with ALL hardware options tuned to maximum
and we do not copy buffers, and do not calc UDP checksum with PPC

but, even TRECKs marketing booklet promised only 355 Mbps for MTU1500
abd i need 580Mbps

Antti








Article: 141029
Subject: Re: VHDL synthesis difference bwetween tools
From: Pauli =?iso-8859-1?Q?Per=E4l=E4?= <nobody@nowhere.com.invalid>
Date: Tue, 02 Jun 2009 22:25:23 +0300
Links: << >>  << T >>  << A >>
Antti <Antti.Lukats@googlemail.com> writes:

>
> sure the code is maybe not fully correct, nice, i would never write
> like this,
> but it works with 3 different type of tools, and fails with the 4th

Me neither. I would think it like this:

			if DidPause = "01" or (Pause = "00" and DidPause(1) = '0') then
				if Rst_r = '1' or Inst_Skip = '1' then
					-- Skip/flush: NOP insertion
					Inst <= (others => '0');
				else
					Inst <= ROM_Data;
				end if;

Since I have no access to any VHDL tool at the moment, I cannot be quite
sure about the correctness but it gives the general idea how I used to
write VHDL. Or maybe there is some particular reason this could not be 
done so?

> so question is, is the code "wrong enough" to allow Magma synthesis
> to generate valid as per VHDL code, that doesnt work?
>

I do not have any personal experience about Magma, but different
interpretation of synthesizable code between synthesis tools does not
surprise me at all. My students were very good at finding those kind of
problems where another tool might even refuse to continue when the other
did not find anything wrong :)

- Pauli

Article: 141030
Subject: Re: VHDL synthesis difference bwetween tools
From: "Antti.Lukats@googlemail.com" <Antti.Lukats@googlemail.com>
Date: Tue, 2 Jun 2009 12:32:41 -0700 (PDT)
Links: << >>  << T >>  << A >>
On 2 June, 22:25, Pauli Per=E4l=E4 <nob...@nowhere.com.invalid> wrote:
> Antti <Antti.Luk...@googlemail.com> writes:
>
> > sure the code is maybe not fully correct, nice, i would never write
> > like this,
> > but it works with 3 different type of tools, and fails with the 4th
>
> Me neither. I would think it like this:
>
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 if DidPause =3D "01" or (=
Pause =3D "00" and DidPause(1) =3D '0') then
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 if Rst_r =
=3D '1' or Inst_Skip =3D '1' then
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0 -- Skip/flush: NOP insertion
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0 Inst <=3D (others =3D> '0');
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 else
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =
=A0 =A0 Inst <=3D ROM_Data;
> =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 =A0 end if;
>
> Since I have no access to any VHDL tool at the moment, I cannot be quite
> sure about the correctness but it gives the general idea how I used to
> write VHDL. Or maybe there is some particular reason this could not be
> done so?
>
> > so question is, is the code "wrong enough" to allow Magma synthesis
> > to generate valid as per VHDL code, that doesnt work?
>
> I do not have any personal experience about Magma, but different
> interpretation of synthesizable code between synthesis tools does not
> surprise me at all. My students were very good at finding those kind of
> problems where another tool might even refuse to continue when the other
> did not find anything wrong :)
>
> - Pauli

it seems that all other except magma syntheis think this way
another way is to create feedback mux, it would create also
working code

whatever magma did, did not work

Antti




Article: 141031
Subject: Re: Open Source FPGA circuit design.
From: nico@puntnl.niks (Nico Coesel)
Date: Tue, 02 Jun 2009 20:31:44 GMT
Links: << >>  << T >>  << A >>
Rich Webb <bbew.ar@mapson.nozirev.ten> wrote:

>On Mon, 1 Jun 2009 13:15:13 -0700 (PDT), "jack.gassett"
><jack.gassett@gadgetfactory.net> wrote:
>
>>Hello all,
>>
>>I just wanted to post that I have released an open source FPGA circuit
>>design. Please feel free to use it as is or to quick start your own
>>designs. It is released under the Creative Commons Attribution license
>>exactly like the Arduino. This is a work in progress so any feedback
>>is appreciated.
>>
>>Here are some specifications:
>>-Schematic and Board layout in EAGLE.
>
>Just curious, but why EAGLE and not one of the open source schematic
>capture and board layout apps like gEDA or Kicad?

I'd recommend using PCB (also open source) instead of KiCad. PCB works
better (don't forget to turn the DRC on).

-- 
Failure does not prove something is impossible, failure simply
indicates you are not using the right tools...
                     "If it doesn't fit, use a bigger hammer!"
--------------------------------------------------------------

Article: 141032
Subject: Re: Xilinx GbE performance
From: "MM" <mbmsv@yahoo.com>
Date: Tue, 2 Jun 2009 17:18:16 -0400
Links: << >>  << T >>  << A >>
<Antti.Lukats@googlemail.com> wrote in message news:244fac96-a937-48b4-949e-
>
> but, even TRECKs marketing booklet promised only 355 Mbps for MTU1500
> abd i need 580Mbps

Antti,

I think Treck numbers assume TCP/IP. I am actually in the middle of 
evaluating the same thing. I have a design similar to the one described in 
XAPP1041 running on a custom V4FX60 board and I seem to be getting the 
numbers you are looking for (raw Ethernet Tx traffic) although it is early 
for me to say whether they are "real", i.e. I haven't yet analyzed properly 
what the Xilinx perf_app software does.


/Mikhail 



Article: 141033
Subject: error with xps_ll_temac
From: namth299@gmail.com
Date: Tue, 2 Jun 2009 20:25:37 -0700 (PDT)
Links: << >>  << T >>  << A >>
hello!

I'm using xps_ll_temac in my project to send media from ML507 to
internet. But after serveral minutes the system automatic reset and an
error occur with message: "incorrect configuration: xps_ethernetlite
drivers not present?". I don't know why because I'm using
xps_ll_temac? On the other hand, when I pressed reset button there was
the same error occured. Have you got any advance?

Thanks!

Article: 141034
Subject: BRAM/LUT Comparison
From: Venkat <venkat.japan@gmail.com>
Date: Tue, 2 Jun 2009 22:28:58 -0700 (PDT)
Links: << >>  << T >>  << A >>
Hi all,

I am wondering what would be the distinct advantage of implementing
Memory using the Dedicated BRAM instead of using the LUTs (assuming I
have plenty of LUTs at my disposal). Will Speed be a criterion? While
I am working on that deeply, if someone had prior experience
researching this point, will be glad to hear your views.

Thanks in advance,
Venkat.

Article: 141035
Subject: Re: Micron SODIMM Type Variation
From: vcar <hitsx@163.com>
Date: Tue, 2 Jun 2009 22:44:51 -0700 (PDT)
Links: << >>  << T >>  << A >>
Anybody could give some advice? Thanks a lot!

Article: 141036
Subject: Re: BRAM/LUT Comparison
From: Frank Buss <fb@frank-buss.de>
Date: Wed, 3 Jun 2009 07:55:45 +0200
Links: << >>  << T >>  << A >>
Venkat wrote:

> I am wondering what would be the distinct advantage of implementing
> Memory using the Dedicated BRAM instead of using the LUTs (assuming I
> have plenty of LUTs at my disposal). Will Speed be a criterion? While
> I am working on that deeply, if someone had prior experience
> researching this point, will be glad to hear your views.

If you have plenty of LUTs, you can use it. Synthesizing time might be
really slow, but a big advantage is that you don't have a read latency like
with BRAM, so might be easier to build fast caches or registers with it. I
use LUTs for small register memory, too, e.g. in my signal generator:

http://www.frank-buss.de/SignalGenerator/

-- 
Frank Buss, fb@frank-buss.de
http://www.frank-buss.de, http://www.it4-systems.de

Article: 141037
Subject: Re: BRAM/LUT Comparison
From: goouse@twinmail.de
Date: Tue, 2 Jun 2009 23:01:37 -0700 (PDT)
Links: << >>  << T >>  << A >>
On 3 Jun., 07:28, Venkat <venkat.ja...@gmail.com> wrote:
> Hi all,
>
> I am wondering what would be the distinct advantage of implementing
> Memory using the Dedicated BRAM instead of using the LUTs (assuming I
> have plenty of LUTs at my disposal). Will Speed be a criterion? While
> I am working on that deeply, if someone had prior experience
> researching this point, will be glad to hear your views.
>
> Thanks in advance,
> Venkat.

Hi Venkat,
speed will surely be a point as the distributive memory grows.
The other point is size. IT makes no sense to waste LUTs for memory if
your BRAMs are empty.
And you need a lot of LUTs to make a of RAM the same size as a BRAM-
Block.
Use each memory wisely to take advantage of it's special abilities.

some generic examples:
If you need some kB of RAM for average bus sizes of 8,16,32 (+parity)
BRAMs are most likely the best choice.
if you need a wide ram with limited size (e.g. 256 x 4) Distributive
RAM is your friend.

The first example will be found quite often in processor related
designs and video processing.
The second one is mainly used in DSP or cryptographic applications.

Of course there are always unique problems that need special handling
of your ressources.
You need to deal with these individually.

Have a nice synthesis
  Eilert


Article: 141038
Subject: Re: BRAM/LUT Comparison
From: Mike Treseler <mtreseler@gmail.com>
Date: Tue, 02 Jun 2009 23:04:52 -0700
Links: << >>  << T >>  << A >>
Venkat wrote:

> I am wondering what would be the distinct advantage of implementing
> Memory using the Dedicated BRAM instead of using the LUTs

BRAM is synchronous and easy to use, and it often goes /unused/.
While LUTs and flops are good for almost anything,
BRAM is designed to
be RAM.

      -- Mike Treseler

Article: 141039
Subject: Re: BRAM/LUT Comparison
From: Venkat <venkat.japan@gmail.com>
Date: Tue, 2 Jun 2009 23:17:55 -0700 (PDT)
Links: << >>  << T >>  << A >>
On Jun 3, 11:01=A0am, goo...@twinmail.de wrote:
> On 3 Jun., 07:28, Venkat <venkat.ja...@gmail.com> wrote:
>
> > Hi all,
>
> > I am wondering what would be the distinct advantage of implementing
> > Memory using the Dedicated BRAM instead of using the LUTs (assuming I
> > have plenty of LUTs at my disposal). Will Speed be a criterion? While
> > I am working on that deeply, if someone had prior experience
> > researching this point, will be glad to hear your views.
>
> > Thanks in advance,
> > Venkat.
>
> Hi Venkat,
> speed will surely be a point as the distributive memory grows.
> The other point is size. IT makes no sense to waste LUTs for memory if
> your BRAMs are empty.
> And you need a lot of LUTs to make a of RAM the same size as a BRAM-
> Block.
> Use each memory wisely to take advantage of it's special abilities.
>
> some generic examples:
> If you need some kB of RAM for average bus sizes of 8,16,32 (+parity)
> BRAMs are most likely the best choice.
> if you need a wide ram with limited size (e.g. 256 x 4) Distributive
> RAM is your friend.
>
> The first example will be found quite often in processor related
> designs and video processing.
> The second one is mainly used in DSP or cryptographic applications.
>
> Of course there are always unique problems that need special handling
> of your ressources.
> You need to deal with these individually.
>
> Have a nice synthesis
> =A0 Eilert

Hello Eilert and Frank,

Thank you so much for the responses.

So the deciding factors would be Size of the Memory and Efficient
Utilization of the Available Resources. On the Speed factor, from
Eilert's views, speed is going to slow down on LUT based RAMs as the
size grows larger. However for reasonably smaller size, the speed on
LUT would be higher? I came across one paper where the DES algorithm
has been implemented with/without BRAMs and the figures show a better
Data Rate for LUT Implementation. The reason quoted is the number of
critical paths (larger for BRAM based than LUT based). Any idea on
this point?

Appreciate your time.
Venkat.

Article: 141040
Subject: Re: Has anyone tried to install a Xilinx floating license? The
From: Allan Herriman <allanherriman@hotmail.com>
Date: 03 Jun 2009 11:23:45 GMT
Links: << >>  << T >>  << A >>
On Tue, 02 Jun 2009 09:06:18 -0700, OutputLogic wrote:

> On Jun 1, 3:34 pm, phil hays <philh...@dont.spam> wrote:
>> soar2morrow wrote:
>>
>> Has anyone tried to install a Xilinx floating license?
>>
>> Yes, three days ago. It really was no problem. And this was on Fedora,
>> which isn't supported.
>>
>> > Now, just try to find XLCM on Xilinx's website (all you will find are
>> > references to it.
>>
>> If the Xilinx tools are installed on the system, open a command prompt
>> and type xclm.
>>
>> > To further complicate issues, it seems like the only way to get help
>> > of ANY KIND is to submit a web case. Just try to talk to someone on
>> > the phone! I actually did; guess what they said: SUBMIT A WEB CASE!
>>
>> I'll be happy to solve problems like this for you for $85/hour plus
>> expenses and travel. Send me an email. I'm very familiar with the
>> Xilinx tools and FPGA design.
>>
>> --
>> Phil Hays
>> (phil_hays at eeei.gro (fix the order for email)
> 
> 
> I installed the floating license yesterday on Windows 64-bit server. So
> far so good.
> Speaking of licenses, we needed a few node-locked licenses in addition
> to the floating one. It made things much more expensive and as the
> result we want to re-evaluate our commitment to Xilinx. Does anybody
> have a recent experience with Altera licensing cost.

The Altera license was roughly half the cost, the last time I bought one.

The Xilinx license is more flexible though - you can get a node locked 
license and run multiple concurrent instances of the software on that 
machine.  (You can verify this for yourself - the license file has 
"uncounted" in it.  The local FAE confirmed that this is not a mistake, 
and we will be holding him to this.)

Cheers,
Allan

Article: 141041
Subject: Xilinx ISE doesn't recognize a signal added in Xilinx Platform Studio
From: Real SoPC New-Be <marteno_rodia@o2.pl>
Date: Wed, 3 Jun 2009 04:50:02 -0700 (PDT)
Links: << >>  << T >>  << A >>
Hello,
I'm working on a system combining a MicroBlaze, memory and some user
RTL logic. I'm not the creator, I've received all the files and I'm
supposed to introduce a few changes... However, I've encountered a
problem. I'll try to describe it.

In the design, there are a few clock signals used, generated by a
clock generator IP. I was supposed to add a new clock signal in the
generator and feed with it a clock input of one of the system
components. I've done it. But... I have to add a constraint for this
new clock in the UCF file. Once I've added a constrained (manually),
I've received an error in Xilinx ISE (clock signal I've added is
unknown). And Xilinx ISE doesn't allow me to add a constraint in GUI
because of the same reason.

What the hell is wrong? Maybe there is a simple obvious thing I don't
know. Anyway, iif you have any idea - please, tell me! Thanks in
advance.

*** ADDITIONAL INFO ***
The exact error message I've received is as follows:

ERROR:ConstraintSystem:59 - Constraint <NET
   "MB_SYSTEM/clock_generator_0/clock_generator_0/PLL0_CLK_OUT<5>"
TNM_NET =
   MB_SYSTEM/clock_generator_0/clock_generator_0/PLL0_CLK_OUT<5>;>
   [MB_ENCODER.ucf(287)]: NET
   "MB_SYSTEM/clock_generator_0/clock_generator_0/PLL0_CLK_OUT<5>" not
found.
   Please verify that:
   1. The specified design element actually exists in the original
design.
   2. The specified object is spelled correctly in the constraint
source file.
<end of msg>

As far as I know, the specified design element does actually exist in
the original design!

Article: 141042
Subject: Re: Xilinx GbE performance
From: "john.orlando@gmail.com" <john.orlando@gmail.com>
Date: Wed, 3 Jun 2009 05:33:53 -0700 (PDT)
Links: << >>  << T >>  << A >>
<snip>
> > > > hum.. my current task is to optimize a XPS_TEMAC based system
> > > > (with 1 single DDR2 chip as main memory!)
> > > > to reach about 580MBps
>
> > > > :(
>
> > > > I have never said that to be possible, but i need money :(
> > > > and if the goal cant be reached there will be none...
>
> > > > over 100MBps is sure possible (with XPS_TEMAC too)
> > > > =A0but 580MBps is beyong doable i think for sure
>
> > > > Antti
>
> > > Just a simple calculation:
> > > 125000000 / 1024 / 1024 =3D 119.2MBps
> > > It is without protocol overhead, FCS, IFGs. How do you want to exceed
> > > limit of Gigabit Ethernet?
>
> > > Jan
>
> > Or did I get you wrong and you talk about Mbits per second? I was
> > talking about Mbytes per sec.
> > If it is so, your goal should be reachable using xps_ll_temac instead o=
f
> > xps_temac.
> > Jan
>
> oh, i am talking wrong today
> yes Mbit/sec or Mbps
> and sure XPS_LL_TEMAC with ALL hardware options tuned to maximum
> and we do not copy buffers, and do not calc UDP checksum with PPC
>
> but, even TRECKs marketing booklet promised only 355 Mbps for MTU1500
> abd i need 580Mbps
>
> Antti

Antti,
The USRP2 (http://en.wikipedia.org/wiki/
Universal_Software_Radio_Peripheral) is a software-defined radio that
uses a Spartan III + gigE PHY chip to reach 800 Mbits/sec sustained.
I believe the MAC in their FPGA has a few limitations (only supports
1000 Base-T) but was originally based on the opencores tri-mode MAC
(though significant modifications were needed to make it reliable,
IIRC).  The other caveat here is that the USRP2 guys push raw ethernet
frames into a PC...i.e., they don't use TCP or UDP.  I believe their
analysis showed that they needed a custom network layer to support the
sustained high data rates.

So I wouldn't give up hope on making something work here at 580 Mbits/
sec.  All of the USRP2 code (software + HDL) is open-sourced, and
should be available through their subversion repositories.

Good Luck,
John

Article: 141043
Subject: Re: Xilinx GbE performance
From: OutputLogic <evgenist@gmail.com>
Date: Wed, 3 Jun 2009 06:26:51 -0700 (PDT)
Links: << >>  << T >>  << A >>
On Jun 3, 5:33=A0am, "john.orla...@gmail.com" <john.orla...@gmail.com>
wrote:
> <snip>
>
>
>
>
>
> > > > > hum.. my current task is to optimize a XPS_TEMAC based system
> > > > > (with 1 single DDR2 chip as main memory!)
> > > > > to reach about 580MBps
>
> > > > > :(
>
> > > > > I have never said that to be possible, but i need money :(
> > > > > and if the goal cant be reached there will be none...
>
> > > > > over 100MBps is sure possible (with XPS_TEMAC too)
> > > > > =A0but 580MBps is beyong doable i think for sure
>
> > > > > Antti
>
> > > > Just a simple calculation:
> > > > 125000000 / 1024 / 1024 =3D 119.2MBps
> > > > It is without protocol overhead, FCS, IFGs. How do you want to exce=
ed
> > > > limit of Gigabit Ethernet?
>
> > > > Jan
>
> > > Or did I get you wrong and you talk about Mbits per second? I was
> > > talking about Mbytes per sec.
> > > If it is so, your goal should be reachable using xps_ll_temac instead=
 of
> > > xps_temac.
> > > Jan
>
> > oh, i am talking wrong today
> > yes Mbit/sec or Mbps
> > and sure XPS_LL_TEMAC with ALL hardware options tuned to maximum
> > and we do not copy buffers, and do not calc UDP checksum with PPC
>
> > but, even TRECKs marketing booklet promised only 355 Mbps for MTU1500
> > abd i need 580Mbps
>
> > Antti
>
> Antti,
> The USRP2 (http://en.wikipedia.org/wiki/
> Universal_Software_Radio_Peripheral) is a software-defined radio that
> uses a Spartan III + gigE PHY chip to reach 800 Mbits/sec sustained.
> I believe the MAC in their FPGA has a few limitations (only supports
> 1000 Base-T) but was originally based on the opencores tri-mode MAC
> (though significant modifications were needed to make it reliable,
> IIRC). =A0The other caveat here is that the USRP2 guys push raw ethernet
> frames into a PC...i.e., they don't use TCP or UDP. =A0I believe their
> analysis showed that they needed a custom network layer to support the
> sustained high data rates.
>
> So I wouldn't give up hope on making something work here at 580 Mbits/
> sec. =A0All of the USRP2 code (software + HDL) is open-sourced, and
> should be available through their subversion repositories.
>
> Good Luck,
> John


I'm using UDP and getting sustainable 600-700Mbits/sec. In fact this
number is limited by the PC side: either a network card or a stack.

- outputlogic

Article: 141044
Subject: Re: Xilinx ISE doesn't recognize a signal added in Xilinx Platform Studio
From: "MM" <mbmsv@yahoo.com>
Date: Wed, 3 Jun 2009 12:09:38 -0400
Links: << >>  << T >>  << A >>
Is your top level design in ISE or EDK? Can you post your system.mhs file?


/Mikhail 



Article: 141045
Subject: Re: BRAM/LUT Comparison
From: General Schvantzkoph <schvantzkoph@yahoo.com>
Date: 3 Jun 2009 17:36:42 GMT
Links: << >>  << T >>  << A >>
On Tue, 02 Jun 2009 22:28:58 -0700, Venkat wrote:

> Hi all,
> 
> I am wondering what would be the distinct advantage of implementing
> Memory using the Dedicated BRAM instead of using the LUTs (assuming I
> have plenty of LUTs at my disposal). Will Speed be a criterion? While I
> am working on that deeply, if someone had prior experience researching
> this point, will be glad to hear your views.
> 
> Thanks in advance,
> Venkat.

If it's bigger than 32 deep then use BRAM otherwise use LUT RAM. Small 
LUT RAMs are faster, the read is asynchronous, and they place less burden 
on the place and router. The advantage of BRAM is size, if you need lots 
of RAM then you use BRAM.

Article: 141046
Subject: Re: Micron SODIMM Type Variation
From: LittleAlex <alex.louie@email.com>
Date: Wed, 3 Jun 2009 10:36:47 -0700 (PDT)
Links: << >>  << T >>  << A >>
On Jun 2, 10:44=A0pm, vcar <hi...@163.com> wrote:
> Anybody could give some advice? Thanks a lot!

My advice is to read the data sheets for the chips mounted on the
modules.

AL

Article: 141047
Subject: Re: Has anyone tried to install a Xilinx floating license? The documentation (UG631 (v 11.1.0) April 27, 2009) says that the required
From: "MM" <mbmsv@yahoo.com>
Date: Wed, 3 Jun 2009 14:11:28 -0400
Links: << >>  << T >>  << A >>
"Allan Herriman" <allanherriman@hotmail.com> wrote in message 
news:004237cd$0$9747$c3e8da3@news.astraweb.com...
>
> The Xilinx license is more flexible though - you can get a node locked
> license and run multiple concurrent instances of the software on that
> machine.

Do you know if the same can be done with a floating license, i.e. running 
multiple instances on a single machine with a single license?


/Mikhail



Article: 141048
Subject: Secure netlist
From: "maxascent" <maxascent@yahoo.co.uk>
Date: Wed, 03 Jun 2009 13:11:59 -0500
Links: << >>  << T >>  << A >>
How would I create a create a secure netlist that I could give to a 3rd
party. Ideally I would like the edif output from Synplify but in a
protected form.

Cheers

Jon

Article: 141049
Subject: Re: Secure netlist
From: Mike Treseler <mtreseler@gmail.com>
Date: Wed, 03 Jun 2009 11:18:25 -0700
Links: << >>  << T >>  << A >>
maxascent wrote:
> How would I create a create a secure netlist that I could give to a 3rd
> party. Ideally I would like the edif output from Synplify but in a
> protected form.


http://download.cnet.com/windows/encryption-software/



Site Home   Archive Home   FAQ Home   How to search the Archive   How to Navigate the Archive   
Compare FPGA features and resources   

Threads starting:
1994JulAugSepOctNovDec1994
1995JanFebMarAprMayJunJulAugSepOctNovDec1995
1996JanFebMarAprMayJunJulAugSepOctNovDec1996
1997JanFebMarAprMayJunJulAugSepOctNovDec1997
1998JanFebMarAprMayJunJulAugSepOctNovDec1998
1999JanFebMarAprMayJunJulAugSepOctNovDec1999
2000JanFebMarAprMayJunJulAugSepOctNovDec2000
2001JanFebMarAprMayJunJulAugSepOctNovDec2001
2002JanFebMarAprMayJunJulAugSepOctNovDec2002
2003JanFebMarAprMayJunJulAugSepOctNovDec2003
2004JanFebMarAprMayJunJulAugSepOctNovDec2004
2005JanFebMarAprMayJunJulAugSepOctNovDec2005
2006JanFebMarAprMayJunJulAugSepOctNovDec2006
2007JanFebMarAprMayJunJulAugSepOctNovDec2007
2008JanFebMarAprMayJunJulAugSepOctNovDec2008
2009JanFebMarAprMayJunJulAugSepOctNovDec2009
2010JanFebMarAprMayJunJulAugSepOctNovDec2010
2011JanFebMarAprMayJunJulAugSepOctNovDec2011
2012JanFebMarAprMayJunJulAugSepOctNovDec2012
2013JanFebMarAprMayJunJulAugSepOctNovDec2013
2014JanFebMarAprMayJunJulAugSepOctNovDec2014
2015JanFebMarAprMayJunJulAugSepOctNovDec2015
2016JanFebMarAprMayJunJulAugSepOctNovDec2016
2017JanFebMarAprMayJunJulAugSepOctNovDec2017
2018JanFebMarAprMayJunJulAugSepOctNovDec2018
2019JanFebMarAprMayJunJulAugSepOctNovDec2019
2020JanFebMarAprMay2020

Authors:A B C D E F G H I J K L M N O P Q R S T U V W X Y Z

Custom Search