FPGA Central

Hi,

We will soon be using the Nios II as embedded controller and we would
like to add a High Speed (HS) USB 2.0 hosting feature that is capable
or providing a sustained transfer rate of 20 MBytes/sec. to an external
HS USB 2.0 device.

I realize that there are a limited number of HS USB 2.0 hosts
devices/IP cores currently available. The Phillips ISP1761 is the only
HS USB 2.0 component I was able to locate so far and it's not clear to
me if this part is readily available. Also, the FPGA-based HS USB 2.0
host IP cores look like they are just now becoming available
(www.asics.ws) but they may be cost-prohibitive.

Can anyone share their experiences in implementing either a HS USB 2.0
or Full Speed (FS) USB 1.1 host? Is a sustained transfer speed of 20
MBytes/sec. achievable with the assumptions that there are no major
transfer bottlenecks in the HS USB 2.0 host component or the associated
HS USB 2.0 device (e.g. USB Hard Disk). If not, what is a more
reasonable transfer speed goal that has a high probability of success?

The choice of which OS with Nios II (or no OS at all) may also be
influenced by the max. sustained HS USB 2.0 transfer speed that can be
achieved. However, it's also possible that in order to meet the 20
MBytes/sec. goal, the Nios II will have to be "removed" from the data
I/O path so that custom FPGA circuitry can handle the transfers
directly with the external HS USB 2.0 host transceiver or the IP core.

Any information and/or opinions would be helpful.

Sincerely,
Brad.

<[email protected]> schrieb im Newsbeitrag
news:[email protected] oups.com...
> Hi,
>
> We will soon be using the Nios II as embedded controller and we would
> like to add a High Speed (HS) USB 2.0 hosting feature that is capable
> or providing a sustained transfer rate of 20 MBytes/sec. to an external
> HS USB 2.0 device.
>
> I realize that there are a limited number of HS USB 2.0 hosts
> devices/IP cores currently available. The Phillips ISP1761 is the only
> HS USB 2.0 component I was able to locate so far and it's not clear to
> me if this part is readily available. Also, the FPGA-based HS USB 2.0
> host IP cores look like they are just now becoming available
> (www.asics.ws) but they may be cost-prohibitive.
>
> Can anyone share their experiences in implementing either a HS USB 2.0
> or Full Speed (FS) USB 1.1 host? Is a sustained transfer speed of 20
> MBytes/sec. achievable with the assumptions that there are no major
> transfer bottlenecks in the HS USB 2.0 host component or the associated
> HS USB 2.0 device (e.g. USB Hard Disk). If not, what is a more
> reasonable transfer speed goal that has a high probability of success?
>
> The choice of which OS with Nios II (or no OS at all) may also be
> influenced by the max. sustained HS USB 2.0 transfer speed that can be
> achieved. However, it's also possible that in order to meet the 20
> MBytes/sec. goal, the Nios II will have to be "removed" from the data
> I/O path so that custom FPGA circuitry can handle the transfers
> directly with the external HS USB 2.0 host transceiver or the IP core.
>
> Any information and/or opinions would be helpful.
>
> Sincerely,
> Brad.
>

ISP1761 is readily available, the proove is there
http://www.hydraxc.com

ISP1761 (in BGA) is on lower side of the PCB. non BGA version has been
available for some time already, the BGA version (0.5mm pitch!) did become
available April/June 2005.

and yes, you are about correct for max sustained dataflow the high speed
traffic should bypass the processor, that could be done by dedicated DMA and
ISP1761

Antti

Antti,

why this is not clearly stated in the datasheet ? To me, it seems a big plus
of that little board!

Has stack support been implemented on ucLinux ?


"Antti Lukats" <[email protected]> ha scritto nel messaggio
news:[email protected]...
>
> <[email protected]> schrieb im Newsbeitrag
> news:[email protected] oups.com...
>> Hi,
>>
>> We will soon be using the Nios II as embedded controller and we would
>> like to add a High Speed (HS) USB 2.0 hosting feature that is capable
>> or providing a sustained transfer rate of 20 MBytes/sec. to an external
>> HS USB 2.0 device.
>>
>> I realize that there are a limited number of HS USB 2.0 hosts
>> devices/IP cores currently available. The Phillips ISP1761 is the only
>> HS USB 2.0 component I was able to locate so far and it's not clear to
>> me if this part is readily available. Also, the FPGA-based HS USB 2.0
>> host IP cores look like they are just now becoming available
>> (www.asics.ws) but they may be cost-prohibitive.
>>
>> Can anyone share their experiences in implementing either a HS USB 2.0
>> or Full Speed (FS) USB 1.1 host? Is a sustained transfer speed of 20
>> MBytes/sec. achievable with the assumptions that there are no major
>> transfer bottlenecks in the HS USB 2.0 host component or the associated
>> HS USB 2.0 device (e.g. USB Hard Disk). If not, what is a more
>> reasonable transfer speed goal that has a high probability of success?
>>
>> The choice of which OS with Nios II (or no OS at all) may also be
>> influenced by the max. sustained HS USB 2.0 transfer speed that can be
>> achieved. However, it's also possible that in order to meet the 20
>> MBytes/sec. goal, the Nios II will have to be "removed" from the data
>> I/O path so that custom FPGA circuitry can handle the transfers
>> directly with the external HS USB 2.0 host transceiver or the IP core.
>>
>> Any information and/or opinions would be helpful.
>>
>> Sincerely,
>> Brad.
>>
>
> ISP1761 is readily available, the proove is there
> http://www.hydraxc.com
>
> ISP1761 (in BGA) is on lower side of the PCB. non BGA version has been
> available for some time already, the BGA version (0.5mm pitch!) did become
> available April/June 2005.
>
> and yes, you are about correct for max sustained dataflow the high speed
> traffic should bypass the processor, that could be done by dedicated DMA
> and
> ISP1761
>
> Antti
>
>
>
>
>
>
>
>
>
>
>
>
>
>

"Antonio Pasini" <[email protected]> schrieb im Newsbeitrag
news:[email protected]...
> Antti,
>
> why this is not clearly stated in the datasheet ? To me, it seems a big
plus
> of that little board!
>
> Has stack support been implemented on ucLinux ?
>

there is some linux code available for the isp1671 but it has to be adapted
and thats not completed.
But the goal is to support the isp1671 from uClinux of course

Antti

> "Antti Lukats" <[email protected]> ha scritto nel messaggio
> news:[email protected]...
> >
> > <[email protected]> schrieb im Newsbeitrag
> > news:[email protected] oups.com...
> >> Hi,
> >>
> >> We will soon be using the Nios II as embedded controller and we would
> >> like to add a High Speed (HS) USB 2.0 hosting feature that is capable
> >> or providing a sustained transfer rate of 20 MBytes/sec. to an external
> >> HS USB 2.0 device.
> >>
> >> I realize that there are a limited number of HS USB 2.0 hosts
> >> devices/IP cores currently available. The Phillips ISP1761 is the only
> >> HS USB 2.0 component I was able to locate so far and it's not clear to
> >> me if this part is readily available. Also, the FPGA-based HS USB 2.0
> >> host IP cores look like they are just now becoming available
> >> (www.asics.ws) but they may be cost-prohibitive.
> >>
> >> Can anyone share their experiences in implementing either a HS USB 2.0
> >> or Full Speed (FS) USB 1.1 host? Is a sustained transfer speed of 20
> >> MBytes/sec. achievable with the assumptions that there are no major
> >> transfer bottlenecks in the HS USB 2.0 host component or the associated
> >> HS USB 2.0 device (e.g. USB Hard Disk). If not, what is a more
> >> reasonable transfer speed goal that has a high probability of success?
> >>
> >> The choice of which OS with Nios II (or no OS at all) may also be
> >> influenced by the max. sustained HS USB 2.0 transfer speed that can be
> >> achieved. However, it's also possible that in order to meet the 20
> >> MBytes/sec. goal, the Nios II will have to be "removed" from the data
> >> I/O path so that custom FPGA circuitry can handle the transfers
> >> directly with the external HS USB 2.0 host transceiver or the IP core.
> >>
> >> Any information and/or opinions would be helpful.
> >>
> >> Sincerely,
> >> Brad.
> >>
> >
> > ISP1761 is readily available, the proove is there
> > http://www.hydraxc.com
> >
> > ISP1761 (in BGA) is on lower side of the PCB. non BGA version has been
> > available for some time already, the BGA version (0.5mm pitch!) did
become
> > available April/June 2005.
> >
> > and yes, you are about correct for max sustained dataflow the high speed
> > traffic should bypass the processor, that could be done by dedicated DMA
> > and
> > ISP1761
> >
> > Antti

Antti,

Thanks for the great feedback. The HydraXC looks like a powerful
module that could be incorporated into several types of embedded
sytems. Too bad it doesn't incorporate a NIOS II soft-core processor
on a Stratix I/II ;-)

Other than the indication that the HydraXC has a USB OTG mini-A/B
connector, there is nothing else on the datasheet to indicate that this
board contains the ISP1761, or any other component(s), that supports HS
USB 2.0 hosting. How can you know from the information on the
website/datasheet that a USB 2.0 hosting capability if provided?

Based on the assumption that a HS USB 2.0 OTG capability is provided,
do you know of any HS USB 2.0 transfer speed benchmarks available for
the HydraXC using the dedicated DMA channel? I didn't see any
benchmark information on their datasheet.

I know uClinux for Nios II currently only supports the Full Speed
ISP1161A1 and not the ISP1761 components from Philips Do you know
which of the listed operating systems that run on the HydraXC support
the ISP1761? Since the ISP1761 looks like it is readily available, I
start looking into obtaining an evaluation board/kit. Have you done
any work with the ISP1761 yourself?

The mounting of a 0.5 mm pitch BGA component is non-trivial. We have
successfully mounted BGA components with a 0.8 mm pitch so making the
jump to a0.5 mm pitch component should be achievable.

Brad.

[email protected] wrote:

> Hi,
>
> We will soon be using the Nios II as embedded controller and we would
> like to add a High Speed (HS) USB 2.0 hosting feature that is capable
> or providing a sustained transfer rate of 20 MBytes/sec. to an external
> HS USB 2.0 device.

Brad,

Our USB 2.0 HS host can sustain close to 56Mbytes/sec.

Best Regards,
rudi
================================================== ===========
Rudolf Usselmann, ASICS World Services, http://www.asics.ws
Your Partner for IP Cores, Design, Verification and Synthesis
****** Certified USB 2.0 HS OTG and HS Device IP Cores ******

> I realize that there are a limited number of HS USB 2.0 hosts
> devices/IP cores currently available. The Phillips ISP1761 is the only
> HS USB 2.0 component I was able to locate so far and it's not clear to
> me if this part is readily available. Also, the FPGA-based HS USB 2.0
> host IP cores look like they are just now becoming available
> (www.asics.ws) but they may be cost-prohibitive.
>
> Can anyone share their experiences in implementing either a HS USB 2.0
> or Full Speed (FS) USB 1.1 host? Is a sustained transfer speed of 20
> MBytes/sec. achievable with the assumptions that there are no major
> transfer bottlenecks in the HS USB 2.0 host component or the associated
> HS USB 2.0 device (e.g. USB Hard Disk). If not, what is a more
> reasonable transfer speed goal that has a high probability of success?
>
> The choice of which OS with Nios II (or no OS at all) may also be
> influenced by the max. sustained HS USB 2.0 transfer speed that can be
> achieved. However, it's also possible that in order to meet the 20
> MBytes/sec. goal, the Nios II will have to be "removed" from the data
> I/O path so that custom FPGA circuitry can handle the transfers
> directly with the external HS USB 2.0 host transceiver or the IP core.
>
> Any information and/or opinions would be helpful.
>
> Sincerely,
> Brad.

Rudi,

Do you have any published benchmarks for your USB 2.0 HS host running
on the Altera Stratix II EP2S90 FPGA?

Also, to your knowledge to you know of any USB 2.0 devices that can
sustain close to the 56 MByytes/sec rate (e.g. Hard Disk).

Brad.