Searching \ for '[EE]:: USB 3 plans' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/index.htm?key=usb+plans
Search entire site for: ': USB 3 plans'.

Exact match. Not showing close matches.
PICList Thread
'[EE]:: USB 3 plans'
2007\09\19@184250 by Russell McMahon

face
flavicon
face
USB 3 due from about 2009 on
Amongst it's many other capabilities it should utterly transform data
transfer rates to/from Flash memory cards.

       http://www.eetimes.com/news/latest/showArticle.jhtml;jsessionid=BUHGCG0JZFOCYQSNDLRSKH0CJUNN2JVN?articleID=201807389&printable=true

   "SAN FRANCISCO, Calif. — Intel Corp. announced it is working with
a handful of companies on a specification that could push the USB's
theoretical throughput beyond 4 Gbits/second, ten times its current
rate. The USB 3.0 spec aims to deliver 300 Mbytes/second of usable
data at the applications level and add new quality of service
capabilities that could challenge the 1394 interconnect also known as
Firewire.  ...

much more.

       Russell

__________________

2007\09\19@213400 by Mujahidin

picon face
Does the USB 3.0 spec is the same with USB On The Go??

Regard

iddhien

-----Original Message-----
From: spam_OUTpiclist-bouncesTakeThisOuTspamMIT.EDU [.....piclist-bouncesKILLspamspam@spam@MIT.EDU] On Behalf Of
Russell McMahon
Sent: Thursday, September 20, 2007 5:33 AM
To: PIC List
Subject: [EE]:: USB 3 plans

USB 3 due from about 2009 on
Amongst it's many other capabilities it should utterly transform data
transfer rates to/from Flash memory cards.


www.eetimes.com/news/latest/showArticle.jhtml;jsessionid=BUHGCG0JZFOC
YQSNDLRSKH0CJUNN2JVN?articleID=201807389&printable=true

   "SAN FRANCISCO, Calif. - Intel Corp. announced it is working with
a handful of companies on a specification that could push the USB's
theoretical throughput beyond 4 Gbits/second, ten times its current
rate. The USB 3.0 spec aims to deliver 300 Mbytes/second of usable
data at the applications level and add new quality of service
capabilities that could challenge the 1394 interconnect also known as
Firewire.  ...

much more.

       Russell

__________________

2007\09\19@223651 by Xiaofan Chen

face picon face
On 9/20/07, Mujahidin <iddhienspamKILLspamgmail.com> wrote:
> Does the USB 3.0 spec is the same with USB On The Go??
>
Apparently not. USB OTG is a supplement to USB 2.0.

You can download USB standards from http://www.usb.org.

Regards,
Xiaofan

2007\09\20@170033 by Cedric Chang

flavicon
face
Intel does
good 'ole pre-announcement to create FUD.

Cedric

2007\09\20@190009 by Tony Smith

picon face
> Intel does
> good 'ole pre-announcement to create FUD.
>
> Cedric


Against what?

RS-232 making a comeback?

Tony

2007\09\20@203931 by David VanHorn

picon face
>
> Against what?
>
> RS-232 making a comeback?


firewire

2007\09\20@213149 by Xiaofan Chen

face picon face
On 9/21/07, David VanHorn <.....microbrixKILLspamspam.....gmail.com> wrote:
> >
> > Against what?
> > RS-232 making a comeback?
> firewire

I believe in Intel. Apple will switch to USB 3 once it is popular
in the Windows PC world. And it will once Intel comes out
chipsets with USB 3. AMD, Via and SIS will follow soon. And since
Apple is now using Intel CPUs and chipsets, Apple will have to
support USB 3 later.

Firewire might still survive in the consumer electronics
world.

Regards,
Xiaofan

2007\09\21@080205 by Tony Smith

picon face
> >
> > Against what?
> >
> > RS-232 making a comeback?
>
>
> firewire


Firewire is dead, Apple killed it when they changed to USB only on the iPod.
(FireWire never was a roaring success anyway (royalties))

Sure, it'll continue on in it's own little niche, maybe one day you be able
to connect your DAT to MiniDisc via Firewire.

Tony

2007\09\21@092918 by Howard Winter

face
flavicon
picon face
Tony,

On Fri, 21 Sep 2007 22:01:35 +1000, Tony Smith wrote:

{Quote hidden}

I'm trying to remember the timing - I think USB came out first, with FireWire following very shortly afterwards, and being 30-odd times faster it was the winner on
performance until USB2 came out.  I have some Iomega "Peerless" drives (20GB removeable cartridges) which have interchangeable bases with Firewire or USB1 in
them, so they were obviously designed in the interim time, perhaps around the turn of the millenium?  These drives fly using FireWire, and dawdle using USB,
obviously, and it's sad that the product line was discontinued without them producing a USB2 base.  I don't know why they discontinued them - the cartridges are
basically a 2.5" disk drive in a box, so there's no reason I can see why they couldn't have updated them by fitting larger capacity disks.  A 160GB removeable
cartridge system with USB2 would have been incredibly useful!

Firewire is a much "nicer" standard, IMHO, first for its performance, which is at least as good as USB2 despite the difference in headline speeds, and secondly
because it has daisy-chaining as standard so you only need one port on the machine and everything can hang from it.  And thirdly because there isn't this ludicrous
thing where Windows loads another set of drivers every time you plug it in in a way that's different from previously!  It just feels more like a proper standard,
rather than something where manufacturers (the cheap ones, anyway) get their product to the point where it just works on test, and stop developing there.  Hence
some products that only work when plugged straight into the machine, not via a hub, or draw more power than is allowed and hope that the machine is kind to
them, or the thing where you *must* load the manufacturer's drivers before plugging in the device, or else Windoze loads an incompatible driver which you then
have to jump through hoops to remove (ICD2, anyone? :-)

FireWire "just works"!

I wonder what USB3 would be used for?  External disks are already availble with SATA connectors, so I don't see the need for a faster interface there, and I can't
think of anything else that would need speeds above USB2 that doesn't already have a standard (Gigabit Ethernet, for example).

Cheers,


Howard Winter
St.Albans, England


2007\09\21@102622 by Xiaofan Chen

face picon face
On 9/21/07, Howard Winter <EraseMEHDRWspam_OUTspamTakeThisOuTh2org.demon.co.uk> wrote:
> FireWire "just works"!

This does not really matter. A lot of the better standard just died
because of cost. Firewire is still more expensive.

USB standard puts a lot of burden to the developers and there
are many "half-working" USB devices. Many of them might work
under Windows but will not work under Linux or Mac. However USB
is still very popular than other connectivity solutions for PC.

> I wonder what USB3 would be used for?  External disks are
> already availble with SATA connectors, so I don't see the need
> for a faster interface there, and I can't  think of anything else that
> would need speeds above USB2 that doesn't already have a
> standard (Gigabit Ethernet, for example).

USB is easier for the enduser than many other standards. And
wireless USB will again be popular. I have strong faith in USB ;-)
But I am not so happy that they do not accept my registration
at USB.org forum using my gmail email address. I am also
not so happy that they only retend one year of posts in the
usb.org forum...

Xiaofan

2007\09\21@103749 by Dario Greggio

face picon face
Xiaofan Chen wrote:

> But I am not so happy that they do not accept my registration
> at USB.org forum using my gmail email address. I am also
> not so happy that they only retend one year of posts in the
> usb.org forum...

crazy, huh?? :-)

Dario

2007\09\21@104558 by Herbert Graf

flavicon
face
On Fri, 2007-09-21 at 14:29 +0100, Howard Winter wrote:
> I wonder what USB3 would be used for?  External disks are already availble with SATA connectors, so I don't see the need for a faster interface there, and I can't
> think of anything else that would need speeds above USB2 that doesn't already have a standard (Gigabit Ethernet, for example).

I am always very worried when someone says "what will this new thing be
used for". If you build it...

As for right now, what would USB3 be good for? Tons of stuff IMHO.
External storage is the big one, yes SATA is there, but you forget that
backwards compatibility has been a LARGE part of the success of many
things in the computer world.

As a manufacturer, what would you choose to include on your external
harddrive: a SATA port, that a consumer MIGHT have, or a USB3 port,
which while a consumer might not have a USB3 port, they pretty much are
guaranteed to have a USB2 port.

Ignoring that, yes, SATA is good for external storage, but what about
other high bandwidth things? What about uncompressed HD streams from
video capture hardware, or HD camcorders? A 1080i uncompressed stream
needs about: 1920x1080x1.3x30x3 ~= 240MB/s, the 4Gbps being discussed
all of a sudden doesn't seem fast enough any more if you consider that
1080i is pretty low if you look a few years in the future. 1080p is
already here, deep colour is coming, and I'm sure a newer higher res HD
standard isn't far off.

You mention gigabit ethernet, what about wireless ethernet adapters?
Current wireless standards are already in the ~300Mbps range, in a few
years time?

What about remote displays? That a technology with SO many uses, but has
been held back by interface technologies. Imagine having a computer
monitor with a USB3.0 interface. Need a second (or third, or fourth)
display but your video card doesn't have it? Just plug in your USB3
monitor. Heck, for many people who only do desktop stuff, why even have
a video card? Just boot up with a USB3.0 monitor?

And then, there are all the other apps that you or I haven't thought of,
that will appear when higher bandwidth options come available. Just
remember, around 10 years ago external hard drives were extremely
expensive and rare (SCSI based) before USB became big.

TTYL

2007\09\21@104736 by Herbert Graf

flavicon
face
On Fri, 2007-09-21 at 22:01 +1000, Tony Smith wrote:
> Firewire is dead, Apple killed it when they changed to USB only on the iPod.
> (FireWire never was a roaring success anyway (royalties))

I don't know if you can label removing firewire from the iPod as the
death of firewire. Firewire IMHO is not dead, but it is dying. Many
digital camcorders still use firewire, and it remains the interface with
less problems when trying to transfer video from a camcorder to a
computer.

That said, I do agree that firewire has no future beyond that. TTYL

2007\09\21@105252 by Tony Smith

picon face
{Quote hidden}

USB just works too.  Well, almost usually for almost all people.  Almost
good enough, really.  One day I shall succeed in installing a USB2 card and
actually having it work, and not swear at the 'ya know, if you plug it into
you USB2 port, it'll work faster' message.

Yeah, Firewire is nicer, but USB is cheaper and more widespread.  Sometimes
it's easier to start at the bottom and work up.  After all, how many USB
keyboards & mice are there?  Doesn't help that Firewire has two different
plugs either.  I'd say Firewire is on about 1 in 3 machines these days.

Speed wasn't much of an issue, most devices (thumbdrives, cameras, MP3 etc)
didn't have much memory, so it didn't take long to transfer anyway.  Ok,
USB1 CD-ROMs & DVDs sucked, but hardly anyone (nerds don't count) did that.

USB3 will be used on the same stuff we have today (just faster).  Your
webcam will do 60FPS, and your 500GB iPod won't take 'like just ages' to
copy to like it does on USB2.  All of those things that are currently
Firewire (camcorders etc) will switch to USB3.

>From a consumer point of view, it's a no-brainer.  'You mean my old stuff
will still work?  Without a stupid dongle thing?  Sweet!'

Tony

2007\09\21@105924 by M. Adam Davis

face picon face
On 9/21/07, Howard Winter <HDRWspamspam_OUTh2org.demon.co.uk> wrote:
> I wonder what USB3 would be used for?  External disks are already availble with SATA connectors, so I don't see the need for a faster interface there, and I can't
> think of anything else that would need speeds above USB2 that doesn't already have a standard (Gigabit Ethernet, for example).

The other standards typically do not:
* Provide power to the device
* Truly support plug and play
* Support all the various transfer modes (isochronous, bulk, etc)
* Support On the Go functionality - you can't hook two SATA devices
together and make a backup.  If you do this with ethernet you need a
special cable or adaptor, and will probably have to configure the
network yourself.

Further, would you rather have 20 ports of 5 different types on your
computer (probably in the back) and have to match the right peripheral
to the right port?  What if the intended port is full, do you unplug
something, add another card, or drop another port expander (if the
interface supports that) onto your already crowded desk?

In fact, if USB can do the things those other interfaces accomplish,
wouldn't you simply rather have one type of port on the computer which
supports pretty much anything you could want to plug in?

The reasons for faster transfer speed are obvious for existing
applications.  I have a 4GB CF card for my camera, and I'd like it to
download all my pictures faster than it now does.

But there are lots of applications you may not have considered simply
because 1) it's not available now and 2) _you_ wouldn't need it...
until you saw it

For instance, I expect to see high performance video cards plugging
into such a port.  Right now peripheral manufacturers have to sell
them to OEMs to install, or convince the end user to perform the
install.  On this list that isn't a problem, but for 80% of the
computer owning population they don't ever want to open up their
computer.  There are many products in this category.

In fact, I expect to see monitors come out that only attach to this
connector, and include high performance graphics engines.  Need
another monitor?  Don't count the number of monitor ports, just buy
one and plug it in, including the latest huge, high resolution
displays.  People don't want to worry about whether their graphics
card is dual link capable, they just want to buy a 30" display and
plug it in.  This will become more of an issue as even the "small and
cheap" monitors become 1080p compatible and beyond.

High performance computing clusters come closer to home every year.
One might choose, for instance, to buy a computer and drop it on top
of their old one.  Plug the two together with such a fast connection
and you can treat it essentially like a single multi-core computer.
This is already something done for graphics cards in SLI
configurations, but imagine having a connection fast enough that
rather than upgrading your computer so you can do video work quickly,
you simply buy a video coprocessor to do your movie manipulation and
encoding at 10-100x realtime.  And you don't have to even move your
computer - just plug it in and go.  Although it'll require a separate
power supply - there's only so much you can do with 2.5W!

But I think that the ultimate useful reason is simply media transfer.
Soon 1080p camcorders will be on the prosumer market, and even with
MPEG4 you're looking at transferring 50GBytes for a clip.  No one
really wants to wait 13 minutes (best case USB2.0) to transfer that to
their computer.  When you start seeing that 1TB drives are common, and
10TB drives are readily available, you'll be wondering where this spec
is.

-Adam

--
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
- - - - - - - - - - - - - - - - - - - - - - - - -
Moving in southeast Michigan? Buy my house: http://ubasics.com/house/

Interested in electronics? Check out the projects at http://ubasics.com

Building your own house? Check out http://ubasics.com/home/

2007\09\21@113353 by Martin Klingensmith

face
flavicon
face
Herbert Graf wrote:
> On Fri, 2007-09-21 at 22:01 +1000, Tony Smith wrote:
>  
>> Firewire is dead, Apple killed it when they changed to USB only on the iPod.
>> (FireWire never was a roaring success anyway (royalties))
>>    
>
> I don't know if you can label removing firewire from the iPod as the
> death of firewire. Firewire IMHO is not dead, but it is dying. Many
> digital camcorders still use firewire, and it remains the interface with
> less problems when trying to transfer video from a camcorder to a
> computer.
>
> That said, I do agree that firewire has no future beyond that. TTYL
>
>  
And I *just* got a computer with a firewire port in 2007. It certainly
wasn't hyped as much after USB 2 came out. Firewire was never pushed by
normal PC hardware makers.
--
Martin K

2007\09\21@115053 by M. Adam Davis

face picon face
Firewire isn't suitable for common, cheap hardware - such as mice and
keyboards.  Primarily due to cost, I suspect.  There's a big
difference between 1 million royalty free keyboards and one million
royalty-to-apple-for-creating-firewire keyboards.  I don't think Apple
ever pushed it for that anyway - they use USB for keyboards now, did
they still use ADB when the first firewire ports came out standard on
Macs?

So no, firewire was never pushed.  It made great inroads into video
production, and early on external hard drives, but the cost difference
is still there.  The keyboard and mouse thing really shoved USB onto
every system everywhere.

One main reason firewire succeeds in video production is due to native
driver support for it.  You can put a firewire video port on a cable
box and just expect it to work - no drivers, or at least no
manufacturer specific drivers.

USB has a limited selection of standard protocols, and due to
extensions and conflicts with desired operation one almost always has
to load a driver.  Mass storage and most HID (keyboards, mice) are the
biggest exceptions.

So, in essence, firewire includes more pieces of the software stack
(some in hardware) as standard, while USB requires more driver
software.

Of course software is cheaper to mass produce than hardware, so USB
was ahead of the game on cost even without the royalty free advantage.

-Adam

On 9/21/07, Martin Klingensmith <@spam@martinKILLspamspamnnytech.net> wrote:
{Quote hidden}

> -

2007\09\21@115439 by Tony Smith

picon face
> > Firewire is dead, Apple killed it when they changed to USB
> only on the iPod.
> > (FireWire never was a roaring success anyway (royalties))
>
> I don't know if you can label removing firewire from the iPod
> as the death of firewire. Firewire IMHO is not dead, but it
> is dying. Many digital camcorders still use firewire, and it
> remains the interface with less problems when trying to
> transfer video from a camcorder to a computer.
>
> That said, I do agree that firewire has no future beyond that. TTYL


Don't push your own spec on your flagship product?  Sounds like dead to me.

If Apple couldn't be bothered to push Firewire, why would anyone else?  It
could have been like the old days, instead of 'buy a scanner & get a SCSI
card', you'd have 'buy an iPod and get a Firewire card'.

Firewire was always a 'high-end' product, USB was cheap & nasty.  Now USB is
cheap & nasty & fast.  

USB3 will take the camcorders, etc away from Firewire.  A lot of things have
an optical out these days, I wonder if USB3 will be compatible with that...

Tony

2007\09\21@130037 by William \Chops\ Westfield

face picon face

On Sep 21, 2007, at 8:53 AM, Tony Smith wrote:

>
> Don't push your own spec on your flagship product?  Sounds like  
> dead to me.
>
> USB3 will take the camcorders, etc away from Firewire.

I'll believe it when I see it.  You start talking video products like
camcorders and you're talking about an awful lot of legacy product in
areas where two-year product lifetimes are not acceptable.

Or, look at it this way: USB2 was supposed to obviate the need for
firewire, and yet today I see many more computers ship with built-in
firewire support than ever before (as part of their multimedia support.)
I can see USB3 replacing SATA for external drives, but I can't really
see it replacing firewire.

BillW

2007\09\21@130840 by Russell McMahon

face
flavicon
face
>> I wonder what USB3 would be used for?  ...

They note that it will handle parallel streams.
At present flash memory cards are getting painfully slow to download.
At present its the memory speed that dictates this but with "300X" SD
cards in the works it won't be long before its interface limited.

I presently use 2 GB and 4 GB cards for my cameras. 8 GB are dearer
per GB and I don't want to risk too many photos on one card and the
download time is too long.  But if they placed multiple eg 1 GB chips
on one card (or one chip with true multiple access) and accessed them
in parallel with multiple streams, then we could see a 16 GB card
downloading in a minute or two.

I am presently more concerned about write speeds than read speeds so
at present the IC write speed is a limiting factor.



           Russell


2007\09\21@144612 by peter green

flavicon
face

> did
> they still use ADB when the first firewire ports came out standard on
> Macs?
>
>  
Macs got USB as standard before they got firewire as standard.

At the time USB was a lowish speed perhiperal bus, firewire was a high
speed perhipheral/video/networking bus. However USB-2 took over for
perhipherals due to cost reasons and 1000BASE-T took over for high speed
networking (firewire was always a pain to use for networking because you
couldn't use hugely long homemade cables)


2007\09\21@144704 by peter green

flavicon
face

> For instance, I expect to see high performance video cards plugging
> into such a port.  
>  
The announcement I have seen says that USB 3 will be a tenfold increase
over USB 2. That would put it in the same ballpark as x2 pci express IF
it manages to acheive high efficiancy which is NOT something that USB
has been good at in the past.

I saw a review of a low end PCI express x1 graphics card and the
conclusion was it was pretty crippled so I would imagine a high end card
would be crippled by x2 like performance.



2007\09\21@145800 by M. Adam Davis

face picon face
You're right.  I expect to see low performance video cards plugging
into USB3 ports.  We've got _really_ low performance USB2.0 video
adaptors now, but at these speeds reasonable gaming and 3D would be
feasible.

But they only just announced it.  We won't see them as a regular
feature on most new computers for another 3-5 years.

-Adam

On 9/21/07, peter green <KILLspamplugwashKILLspamspamp10link.net> wrote:
{Quote hidden}

> -

2007\09\21@153520 by William \Chops\ Westfield

face picon face

On Sep 21, 2007, at 8:46 AM, Russell McMahon wrote:

> then we could see a 16 GB card downloading in a minute or two.

I dunno.  I thought the current USB2 was pretty far from being
the limiting factor.  A disk-to-samedisk copy of a 10GB photo
subdirectory took 13+ minutes on my old-but-not ancient Mac,
for instance (hmm.  about 6 minutes thereafter for a
disk-to-otherdisk copy, but who knows what was cached by then.)

SATA-2 disks and a faster CPU/Bus might improve that, but it's
pretty clear that MERELY a faster version of USB wouldn't give
me the sort of transfer times you are looking for.

BillW

2007\09\21@181121 by peter green

flavicon
face
M. Adam Davis wrote:
> You're right.  I expect to see low performance video cards plugging
> into USB3 ports.  We've got _really_ low performance USB2.0 video
> adaptors now, but at these speeds reasonable gaming and 3D would be
> feasible.
>
> But they only just announced it.  We won't see them as a regular
> feature on most new computers for another 3-5 years.
>
>  
Right, I can see such adaptors working for those who want an extra
monitor for desktop stuff and maybe as an upgrade for those with really
shitty integrated graphics but they will probablly get blown out of the
water by even the lowest end nvidia and ati PCIe x16/AGP 8x cards and
quite possiblly by integrated stuff too if intels graphics guys get
their act together.

Remember in the time it will take USB 3 to become a standard feature
games will have got more demanding.

P.S. I think the designers of PCI express made a real mistake in not
allowing card connectors to overhang the back of the slot.

2007\09\21@201828 by Herbert Graf

flavicon
face
On Fri, 2007-09-21 at 23:11 +0100, peter green wrote:
> Right, I can see such adaptors working for those who want an extra
> monitor for desktop stuff and maybe as an upgrade for those with really
> shitty integrated graphics but they will probablly get blown out of the
> water by even the lowest end nvidia and ati PCIe x16/AGP 8x cards and
> quite possiblly by integrated stuff too if intels graphics guys get
> their act together.
>
> Remember in the time it will take USB 3 to become a standard feature
> games will have got more demanding.

True, but how many PC users PLAY games? Frankly (and I work for a
company that relies on graphics being pushed up every year) games are
important for very few people. For most people, me included, the most
demanding thing I demand from my video cards are playing videos.

TTYL

2007\09\21@203932 by Xiaofan Chen

face picon face
On 9/22/07, Herbert Graf <RemoveMEmailinglist3TakeThisOuTspamfarcite.net> wrote:
> On Fri, 2007-09-21 at 23:11 +0100, peter green wrote:
> > Right, I can see such adaptors working for those who want an extra
> > monitor for desktop stuff and maybe as an upgrade for those with really
> > shitty integrated graphics but they will probablly get blown out of the
> > water by even the lowest end nvidia and ati PCIe x16/AGP 8x cards and
> > quite possiblly by integrated stuff too if intels graphics guys get
> > their act together.
> >
> > Remember in the time it will take USB 3 to become a standard feature
> > games will have got more demanding.
>
> True, but how many PC users PLAY games? Frankly (and I work for a
> company that relies on graphics being pushed up every year) games are
> important for very few people. For most people, me included, the most
> demanding thing I demand from my video cards are playing videos.

I do not play games either. But the voice of those people who
love gaming is quite loud and is what Nvidia/ATI(now AMD) likes to
hear. And the popular websites like Tom's Hardware and Anandtech
still compare graphics cards based on gaming performance. I
still follow their articles from time to time.

Soundcard is a different story, many gamers do not really care
about sound. So Creative is not doing well since the core sound
card business is quite small now.

For professional CAD users, graphics cards are important. Software
like ProE is really intensive. So does those in the creative industry
like digital media content creation.

Xiaofan

2007\09\21@215415 by Jake Anderson

flavicon
face
Herbert Graf wrote:
{Quote hidden}

And perhaps to run vista?
At least with beryl/compiz the price you pay in terms of system
requirements gets you something that actually is 3D. (And on that note I
have installed a perfictaly nifty beryl system using xubuntu on a
pentium 4 1.2ghz with a 32mb ati graphics card, ok so you had to turn
some effects off but you still got the "cube", fire and wobbly windows.)

2007\09\21@220925 by Xiaofan Chen

face picon face
On 9/22/07, Jake Anderson <spamBeGonejakespamBeGonespamvapourforge.com> wrote:
> >
> And perhaps to run vista?
> At least with beryl/compiz the price you pay in terms of system
> requirements gets you something that actually is 3D. (And on that note I
> have installed a perfictaly nifty beryl system using xubuntu on a
> pentium 4 1.2ghz with a 32mb ati graphics card, ok so you had to turn
> some effects off but you still got the "cube", fire and wobbly windows.)

I do not know but I think beryl/compiz is not useful for me and it is
not working for me anyway on Ubuntu 7.04 and 6.06. I have a 128M
AGP ATI 9800SE graphics card running with AMD 64 3000+, 1GB
DDRRAM and NForce 3 based motherboard (btw: quite bad USB
support for NForce 3, under Linux and Windows and FreeBSD).

Xiaofan

2007\09\22@001858 by Herbert Graf

flavicon
face
On Sat, 2007-09-22 at 08:39 +0800, Xiaofan Chen wrote:
> I do not play games either. But the voice of those people who
> love gaming is quite loud and is what Nvidia/ATI(now AMD) likes to
> hear. And the popular websites like Tom's Hardware and Anandtech
> still compare graphics cards based on gaming performance. I
> still follow their articles from time to time.

You misunderstand me, I'm not saying graphics performance isn't
important. I'd be out of a job if it wasn't. What I'm saying is that for
the majority of people it isn't actually important, making a "USB
monitor" a very useful thing, counter to the prior opinion.

TTYL

2007\09\22@014808 by Tony Smith

picon face
> > Don't push your own spec on your flagship product?  Sounds
> like dead
> > to me.
> >
> > USB3 will take the camcorders, etc away from Firewire.
>
> I'll believe it when I see it.  You start talking video
> products like camcorders and you're talking about an awful
> lot of legacy product in areas where two-year product
> lifetimes are not acceptable.
>
> Or, look at it this way: USB2 was supposed to obviate the
> need for firewire, and yet today I see many more computers
> ship with built-in firewire support than ever before (as part
> of their multimedia support.) I can see USB3 replacing SATA
> for external drives, but I can't really see it replacing firewire.


PCs might have Firewire support, but no-one uses it.  Same as parallel,
serial, PS2 etc.  I noticed a Dell ad the other day, you had to pay a few
dollars to get the Firewire adaptor (plugs into the MB, like USB).

As I said, Firewire will stay in its little niche, and that will probably be
the pro end of the camcorder stuff.  The rest of the world doesn't care,
they'll buy the USB version.  Firewhat?

Tony

2007\09\22@095944 by peter green

flavicon
face

> True, but how many PC users PLAY games? Frankly (and I work for a
> company that relies on graphics being pushed up every year) games are
> important for very few people. For most people, me included, the most
> demanding thing I demand from my video cards are playing videos.
>  
How many users upgrade thier video card for any reason other than games?
My guess is that the number of people who want more montiors than thier
system supports ouf of the box and are afraid to open up thier PC to get
it is small.

I'm sure someone will make a USB 3 graphics adaptor but I also beleive
it will remain a niche product.

2007\09\22@103333 by Tony Smith

picon face
> > True, but how many PC users PLAY games? Frankly (and I work for a
> > company that relies on graphics being pushed up every year)
> games are
> > important for very few people. For most people, me
> included, the most
> > demanding thing I demand from my video cards are playing videos.
> >
>  
> How many users upgrade thier video card for any reason other
> than games?
> My guess is that the number of people who want more montiors
> than thier system supports ouf of the box and are afraid to
> open up thier PC to get it is small.
>


Gamers, programmers & stock brokers for multi-monitor.  Possibly graphics
people, but I've never seen it.  They all seem to like bigger screens.

I'll take multi-monitor (I have 3) over a bigger screen any day.  I use the
on-board video, plus a dual-head card.  Screen 3 is also on a KVM, so I can
plug someone else's box in.

I don't need a cutting edge video card for my screens (an IDE doesn't have
all that many polygons) so even a USB one would work.  I've occasionally
used a TV as a secondary screen.

Tony

2007\09\22@134823 by Howard Winter

face
flavicon
picon face
Russell,

On Sat, 22 Sep 2007 03:46:53 +1200, Russell McMahon wrote:

> >> I wonder what USB3 would be used for?  ...
>
> They note that it will handle parallel streams.

There is obviously some meaning of the word "Serial" with which I've been previously unfamiliar!  :-)

Cheers,




Howard Winter
St.Albans, England


2007\09\22@143659 by William \Chops\ Westfield

face picon face

On Sep 22, 2007, at 6:59 AM, peter green wrote:

> How many users upgrade thier video card for any reason other than  
> games?
>
A fair number for bigger screens?  The 30inch LCDs require a pretty
special graphics card.

>
> I'm sure someone will make a USB 3 graphics adaptor but I also beleive
> it will remain a niche product.

There are already assorted such "niche" adaptors that allow a second
monitor to operate over ethernet, USB, and perhaps other media.  They're
apparently quite adequate for "moderate" graphics activity (KDE  
desktops,
Eclipse IDE, etc.)

BillW

2007\09\22@145200 by Martin Klingensmith

face
flavicon
face
Xiaofan the problem is that you have an ATI card and you're trying to
use it in Linux! The drivers are notoriously bad. AMD just announced
that they're going to open the spec. for the ATI cards, so it may get
better in 6 months - 1 year. I just bought a new nvidia card and I'm
amazed by what it can do in linux.
--
Martin K

Xiaofan Chen wrote:
{Quote hidden}

2007\09\22@153420 by Gerhard Fiedler

picon face
Howard Winter wrote:

>>>> I wonder what USB3 would be used for?  ...
>>
>> They note that it will handle parallel streams.
>
> There is obviously some meaning of the word "Serial" with which I've
> been previously unfamiliar!  :-)

While it sounds funny, I think there might be differences between several
serial busses in parallel and a parallel bus. One could be that the signal
lines on a parallel bus are synced with each other, whereas the several
parallel serial busses not necessarily are.

Gerhard

2007\09\22@175249 by Herbert Graf

flavicon
face
On Sat, 2007-09-22 at 18:48 +0100, Howard Winter wrote:
> Russell,
>
> On Sat, 22 Sep 2007 03:46:53 +1200, Russell McMahon wrote:
>
> > >> I wonder what USB3 would be used for?  ...
> >
> > They note that it will handle parallel streams.
>
> There is obviously some meaning of the word "Serial" with which I've been previously unfamiliar!  :-)

Actually a "parallel-serial" type of config is becoming very common
these days.

PCIE is basically this, each lane is a seperate serial channel, so
multiple lanes become parallel serial links.

This even extends to broadband internet, the most common methods these
days virtually split the spectrum into hundreds or even thousands of
seperate serial "channels" running in parallel.

TTYL

2007\09\22@193827 by Xiaofan Chen

face picon face
On 9/23/07, Martin Klingensmith <RemoveMEmartinspamTakeThisOuTnnytech.net> wrote:
> Xiaofan the problem is that you have an ATI card and you're trying to
> use it in Linux! The drivers are notoriously bad. AMD just announced
> that they're going to open the spec. for the ATI cards, so it may get
> better in 6 months - 1 year. I just bought a new nvidia card and I'm
> amazed by what it can do in linux.
> --

The ATI9800SE graphic card works fine for me. I do not use the
prorietory ATI driver under Linux. I use the open source ATI
driver as I do not need strong 3D capability. It is stable enough.

Actually it is the NForce 3 chipsets from NVidia are giving me
problems. PICkit 2 Serial Analyser does not work with the
motherboard under Linux/Windows.

And IMHO beryl/compiz are useless screen candy which makes the
computer slower. Gnome/KDE are already bloated enough
and Beryl/Compiz are just too much for me. So even if it
worked for me, I will turn it off. And I believe if I tried a bit
harder, it would work. I just do not need them.

I like to watch other people play games (not the dizzy 3D first
man shooter type, but those green color nice RPG type of games).
But I do not play myself. I only need Gameboy/NES level of
simple games...

Xiaofan

2007\09\22@194619 by Xiaofan Chen

face picon face
On 9/22/07, peter green <plugwashEraseMEspam.....p10link.net> wrote:
>
> > True, but how many PC users PLAY games? Frankly (and I work for a
> > company that relies on graphics being pushed up every year) games are
> > important for very few people. For most people, me included, the most
> > demanding thing I demand from my video cards are playing videos.
> >
> How many users upgrade thier video card for any reason other than games?

I think this is true. Most PC users who upgrade their video card are
for Gaming reason. And among them, a few people will also upgrade
the sound card for gaming reasons. The percentage is too low for
Creative Labs to prosper...

> I'm sure someone will make a USB 3 graphics adaptor but I also
> beleive it will remain a niche product.

I agree with you. I believe USB (even USB3) will not be the standard
for graphics card. I am not so sure if I understand this but I feel
that system latency would be a major problem with USB
in high speed applications. USB is kind of master/slave system
and the round trip takes time (Master --> Slave --> Master).

Xiaofan

2007\09\22@211728 by Jake Anderson

flavicon
face

> Actually it is the NForce 3 chipsets from NVidia are giving me
> problems. PICkit 2 Serial Analyser does not work with the
> motherboard under Linux/Windows.
>  
Sucky :-<
I tried to go 64 bit with my new computer.
Quad core q6600 and an nvidia 8800GTS
Problem i had was the default drivers included in ubuntu gutsy don't
support the 8800.
I got the ones from nvidia and installed them but the whole thing was so
unstable that i wiped it and went to 32 bit.
All seems fine now. There isn't any inherent reason for the 64 bit to be
unstable its just not tested as much yet so theres compatibility
problems i think.

> And IMHO beryl/compiz are useless screen candy which makes the
> computer slower. Gnome/KDE are already bloated enough
> and Beryl/Compiz are just too much for me. So even if it
> worked for me, I will turn it off. And I believe if I tried a bit
> harder, it would work. I just do not need them.
>  
Actually the performance impact seems to be pretty minimal if you don't
pick insane "animations" All the heavy lifting is offloaded to the GPU
which is otherwise just sitting there chewing power. In some cases it
seems to improve performance. On my friends P4 1.2Ghz it made it seem
much "snappier" and the response you get when other people see the cube
or a "burn" minimize (which really isn't CPU friendly ;->) makes it
worth it. If you want to impress a client with your geekary compiz is
the way to do it (incidentally flipping desktops while you get that
"Impress" sideshow ready).

> I like to watch other people play games (not the dizzy 3D first
> man shooter type, but those green color nice RPG type of games).
> But I do not play myself. I only need Gameboy/NES level of
> simple games...
>
> Xiaofan
>  

2007\09\22@220205 by Xiaofan Chen

face picon face
On 9/23/07, Jake Anderson <EraseMEjakespamvapourforge.com> wrote:
>
> > Actually it is the NForce 3 chipsets from NVidia are giving me
> > problems. PICkit 2 Serial Analyser does not work with the
> > motherboard under Linux/Windows.
> >
> Sucky :-<
> I tried to go 64 bit with my new computer.
> Quad core q6600 and an nvidia 8800GTS
> Problem i had was the default drivers included in ubuntu gutsy don't
> support the 8800.
> I got the ones from nvidia and installed them but the whole thing was so
> unstable that i wiped it and went to 32 bit.
> All seems fine now. There isn't any inherent reason for the 64 bit to be
> unstable its just not tested as much yet so theres compatibility
> problems i think.
>

Ah similar experience with Ubuntu 6.06 64bit. It was not too bad
overall but just too troublesome to get some of the things to work.
So I went back to Ubuntu 6.06 32bit and it is still my main Linux
distribution to use. I have installed 7.04 but not used it so extensively.
Now I am using 7.04 because I want to diagnose the problem with
PICDEM FS USB bootloader. It is not working with the stock
kernel.
http://sourceforge.net/mailarchive/forum.php?thread_name=1189786321l.13173l.12l%40mofo&forum_name=libusb-devel

So I build a vanilla kernel 2.6.22.7 with USB debugging on.
It is still not working. Probably I will ask in the linux-usb-user or
linux-usb-devel list. Alan Stern and Greg KH are quite helpful there.
Eg:
RemoveMElinux-usb-develEraseMEspamEraseMElists.sourceforge.net/7630804.html">http://www.opensubscriber.com/message/RemoveMElinux-usb-develspam_OUTspamKILLspamlists.sourceforge.net/7630804.html

By the way, I am not able to remember all those fancy names of
Ubuntu releases and I almost always use the 6.06/7.04 name. ;-)

Xiaofan

2007\09\22@220440 by Stephen R Phillips

picon face

--- Xiaofan Chen <RemoveMExiaofancTakeThisOuTspamspamgmail.com> wrote:

> On 9/22/07, peter green <EraseMEplugwashspamspamspamBeGonep10link.net> wrote:
> I think this is true. Most PC users who upgrade their video card are
> for Gaming reason. And among them, a few people will also upgrade
> the sound card for gaming reasons. The percentage is too low for
> Creative Labs to prosper...
>
Creative labs suffers from not being the solution to peoples problems.
It's quite simple, high end consumer grade audio systems are a non
market, and a moronic exercise in futility.  High end professional
solutions do exist, and have a high profitability just not a high
volume market.  Numerous other companies have now filled this space so
Creative is in the realm of useless. They can't even make a decent MP3
style audio player, let alone professional high end audio system
(single ended 24bit audio is silly, the noise margins alone make it
pointless 20bit is pushing the margins hard for this type of
interface), I feel no pity for them.

> I agree with you. I believe USB (even USB3) will not be the standard
> for graphics card. I am not so sure if I understand this but I feel
> that system latency would be a major problem with USB
> in high speed applications. USB is kind of master/slave system
> and the round trip takes time (Master --> Slave --> Master).

This is highly dependent on the API and underlying protocol used for
the graphics.  Using project Athena type systems it would work quite
nicely to be blunt (X is an existing standard but I doubt anyone would
have the forethought or the insite to use something that exists and
works quite well when you can reinvent a wheel and create 'new
technology' that you can sell).  If you don't know what I'm referring
to try this http://www.x.org/wiki/. The X protocol works well with low
or high bandwidth and with compression added on is quite impressive. It
worked fine on 10mbit heavily loaded Ethernet systems 15-18 years ago
when work stations were being connected to super computers and
'powerful' systems that current desktop systems now easily rival.

This just requires some mild effort in design to make a decent X server
for the other end of the USB cable.  In this case the Windows GDI calls
could be translated into X tokens and calls compressed and transmitted
across USB (heck full speed USB would work just fine instead of
480mbps). A lowly ARM9T processor with some DDRAM NAND flash and an LCD
interface would be quite sufficient to create X server attached
directly to a monitor. Need multiple monitors?  Have a bit of software
to split things up and the OS wouldn't be the wiser. X was SPECIFICALLY
designed for the application that's being discussed. However Microsoft
continues to reinvent wheels (think thin clients) when it comes to
graphics API's etc.  I find there use of technology rather pathetic
(Microsoft that is) considering these problems were considered in depth
almost 20 years ago.

There is an old saying "There is nothing new under the sun".

Some references for multiprocessing OpenGL API information.
http://chromium.sourceforge.net/

Tungsten Graphics
http://www.tungstengraphics.com/services.htm
http://www.tungstengraphics.com/chromium/siframes.htm

This company is rather interesting I think the person who made MESA
In summary most if not all these concerns about turn around time for 3d
and 2d applications across USB are a tad unfounded, as technology has
long ago (say 10 or 15 years ago) been developed to deal with this
problem. Say thank you to Silicon Graphics and MIT already having done
all the pioneering work in 2d and more recently in 3d chromium has been
addressing this.

See
www.tungstengraphics.com/chromium/chromium.html
for a very interesting discussion of 2d and 3d API parallelism across
multiple processing elements. This all in existence while USB 2.0 spec
was being hammered out.

I remember discussing with my college professor 20 years ago about
using hardware for 3d acceleration and his words were "You don't think
they were wanting to do the same when they came up with many of the
software algorithms for 3d graphics do you?" He said this with a grin.
Something to consider is that these ideas are definitely NOT NEW and
someone already has thought of the solution long before the problem
needed to be addressed today.

Adapt improvise and over come. I don't see any of these actually being
a real technical problem to solve considering software has already been
written to address these. However in reality we all want to make the
next wheel it's our nature to make something 'ours' (I've done it so I
should know a bit about reinventing a wheel, just think you repeat ALL
the same mistakes they did before you as well).

Stephen

Stephen R. Phillips was here
Please be advised what was said may be absolutely wrong, and hereby this disclaimer follows.  I reserve the right to be wrong and admit it in front of the entire world.


     
____________________________________________________________________________________
Boardwalk for $500? In 2007? Ha! Play Monopoly Here and Now (it's updated for today's economy) at Yahoo! Games.
http://get.games.yahoo.com/proddesc?gamekey=monopolyherenow  

2007\09\22@222953 by Jake Anderson

flavicon
face

> Ah similar experience with Ubuntu 6.06 64bit. It was not too bad
> overall but just too troublesome to get some of the things to work.
> So I went back to Ubuntu 6.06 32bit and it is still my main Linux
> distribution to use. I have installed 7.04 but not used it so extensively.
> Now I am using 7.04 because I want to diagnose the problem with
> PICDEM FS USB bootloader. It is not working with the stock
> kernel.
> sourceforge.net/mailarchive/forum.php?thread_name=1189786321l.13173l.12l%40mofo&forum_name=libusb-devel
>  
Gutsy is 7.10  it should be out of beta pretty soon (if its not now)
I moved to it a while ago to get the latest version of asterisk.
I've never had to compile a kernel for ubuntu, Is there much to worry
about with the ubuntu specific patches?

{Quote hidden}

2007\09\22@224558 by Xiaofan Chen

face picon face
On 9/23/07, Stephen R Phillips <spamBeGonecyberman_phillipsSTOPspamspamEraseMEyahoo.com> wrote:
>
> --- Xiaofan Chen <KILLspamxiaofancspamBeGonespamgmail.com> wrote:
>
> > On 9/22/07, peter green <EraseMEplugwashspamEraseMEp10link.net> wrote:
> > I think this is true. Most PC users who upgrade their video card are
> > for Gaming reason. And among them, a few people will also upgrade
> > the sound card for gaming reasons. The percentage is too low for
> > Creative Labs to prosper...
> >
> Creative labs suffers from not being the solution to peoples problems.
> It's quite simple, high end consumer grade audio systems are a non
> market, and a moronic exercise in futility.  High end professional
> solutions do exist, and have a high profitability just not a high
> volume market.  Numerous other companies have now filled this space so
> Creative is in the realm of useless. They can't even make a decent MP3
> style audio player, let alone professional high end audio system
> (single ended 24bit audio is silly, the noise margins alone make it
> pointless 20bit is pushing the margins hard for this type of
> interface), I feel no pity for them.

To me their major problem in MP3 is not getting the right
industrial design people and they were too ambitious to take on Apple.

On the soundcard front, they are still dominate but the market is
too small.

As a company, Creative is too conservative and lack of foresight. If they
bought Nvidia last time (Creative was one of the major graphics card
vendor last time), either Nvidia would die (probably the case) or
Creative would dominate the discrete graphics card market along
with ATI/AMD.

By the way, the latest Creative MP3 players are not bad but it
is already too little and too late.

{Quote hidden}

Thanks for the education. My maths is very good but my 3D imagination
is not good (the only course I got D in the university was mechanical
engineering graphics) so I am not that into 3D graphics. But you are
certainly right that X is nice. Open GL is nice as well. But I hear
from my brother Direct X is not bad either for graphics.

But I have reservations about think clients. PC is still better
than thin clients IMHO. And PCs can be "thinner" than "thin" clients
in terms of cost...

Regards,
Xiaofan

2007\09\22@224953 by Xiaofan Chen

face picon face
On 9/23/07, Jake Anderson <@spam@jake@spam@spamspam_OUTvapourforge.com> wrote:
> Gutsy is 7.10  it should be out of beta pretty soon (if its not now)
> I moved to it a while ago to get the latest version of asterisk.
> I've never had to compile a kernel for ubuntu, Is there much to worry
> about with the ubuntu specific patches?

I do not want to compile a kernel if I can. And in most cases I do
not need to do that.

However in this case, I intend to ask the question in the linux-usb-devel
list. If you want to seek help from the kernel USB developers, you should
build the latest vanilla kernel and turn on USB Debug as well as build
usbmon as a module. By default these options are not on.

> > So I build a vanilla kernel 2.6.22.7 with USB debugging on.
> > It is still not working. Probably I will ask in the linux-usb-user or
> > linux-usb-devel list. Alan Stern and Greg KH are quite helpful there.
>

2007\09\23@012454 by Peter Todd

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

On Sat, Sep 22, 2007 at 12:52:24AM +1000, Tony Smith wrote:
> Yeah, Firewire is nicer, but USB is cheaper and more widespread.  Sometimes
> it's easier to start at the bottom and work up.  After all, how many USB
> keyboards & mice are there?  Doesn't help that Firewire has two different
> plugs either.  I'd say Firewire is on about 1 in 3 machines these days.

One downside to firewire that no-one has mentioned is security. A
firewire port is essentially a peripheral controlled DMA port, so if you
want to you can design a firewire peripheral that when plugged into a
computer simply makes a copy of all available system ram.

There are law enforcement and intelligence targeted products available
in small form factors with the idea that you find the targets
computer/laptop, quickly plug the device in, wait until the leds stop
flashing and run. Good way to get encryption keys from memory.

- --
http://petertodd.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFG9YJv3bMhDbI9xWQRAq7UAJ9M8YiBXZ5Yn46qBzCy3HAgegz6cwCglyeV
lf3UEcV2Utbm6u2rdgmQGB0=
=cf9C
-----END PGP SIGNATURE-----

2007\09\23@050403 by SM Ling

picon face
>
> Creative labs suffers from not being the solution to peoples problems.


I have to reluctantly agree to the above point.

I am no representative, not the audio nor the gamer fan, but has always been
wanting to buy Creative stuff for my podcast listening.  Mainly, because the
CEO Sim has always been giving his own money to various charities and
individuals when they are in need, and regardless his own wealth has been
coming down.  But so far, creative product enhancement is skin deep, not
what they are actually capable of.

Ling SM

2007\09\23@092758 by Xiaofan Chen

face picon face
On 9/23/07, SM Ling <spamBeGonesm.ling11spamKILLspamgmail.com> wrote:
> >
> > Creative labs suffers from not being the solution to peoples problems.
>
> I have to reluctantly agree to the above point.
>
> I am no representative, not the audio nor the gamer fan, but has always been
> wanting to buy Creative stuff for my podcast listening.  Mainly, because the
> CEO Sim has always been giving his own money to various charities and
> individuals when they are in need, and regardless his own wealth has been
> coming down.  But so far, creative product enhancement is skin deep, not
> what they are actually capable of.

I do not know about Mr Sim's good deeds. But I know that many
engineers left Creative headquarter in Singapore last year and this year
due to several years of lack of promotion and salary increment. As a
direct result, I do not have much concerns of Creative now...

Back in year1998-2000, Creative employees enjoyed generous bonus and
stock options. After that, things became from bad to worse. Now they seem
to recover a bit. The bulk of the revenue is now from MP3 players.
I believe that this section is still losing money but I am not sure. I hear that
the speaker business is doing fine. The soundcard business is still
generating good returns but the revenue will continue to go down.
http://sg.finance.yahoo.com/q/bc?s=C76.SI&t=my&l=on&z=m&q=l&c=

Xiaofan

2007\09\23@093307 by Xiaofan Chen

face picon face
On 9/23/07, Peter Todd <.....petespam_OUTspampetertodd.ca> wrote:
> One downside to firewire that no-one has mentioned is security. A
> firewire port is essentially a peripheral controlled DMA port, so if you
> want to you can design a firewire peripheral that when plugged into a
> computer simply makes a copy of all available system ram.

IEEE1394 is a suppored kernel debug option for Windows along
with the good old serial port. Vista is said to support USB 2.0
kernel debugging.
http://www.microsoft.com/whdc/devtools/debugging/whatsnew.mspx

The thing is that there are still many PCs without IEEE1394 ports.

Xiaofan

2007\09\23@093642 by Xiaofan Chen

face picon face
On 9/23/07, Martin Klingensmith <TakeThisOuTmartin.....spamTakeThisOuTnnytech.net> wrote:
> I just bought a new nvidia card and I'm
> amazed by what it can do in linux.

Some examples please so that I can get some good reasons
to buy a new computer with Vista. I will then have it dual boot with Linux.
Maybe it will be a good time when Vista SP1 is out.

Xiaofan

2007\09\23@105547 by Herbert Graf

flavicon
face
On Sun, 2007-09-23 at 10:02 +0800, Xiaofan Chen wrote:
> Ah similar experience with Ubuntu 6.06 64bit. It was not too bad
> overall but just too troublesome to get some of the things to work.
> So I went back to Ubuntu 6.06 32bit and it is still my main Linux
> distribution to use. I have installed 7.04 but not used it so extensively.

Well, FWIW, I don't think 64 bit "issues" are in the domain of Ubuntu
alone, after 3 tries I gave up trying to install Fedora 6 64bit on my
main machine and just went and installed the 32bit version. When a
distro won't even install...

> By the way, I am not able to remember all those fancy names of
> Ubuntu releases and I almost always use the 6.06/7.04 name. ;-)

Hehe, I prefer the numbers too. FWIW 7.04 is what I'm running on my
laptop now, I believe it's called feisty?

TTYL

2007\09\23@194910 by Russell McMahon

face
flavicon
face
>>> They note that it will handle parallel streams.

>> There is obviously some meaning of the word "Serial" with which
>> I've
>> been previously unfamiliar!  :-)

> While it sounds funny, I think there might be differences between
> several
> serial busses in parallel and a parallel bus. One could be that the
> signal
> lines on a parallel bus are synced with each other, whereas the
> several
> parallel serial busses not necessarily are.

The concept that I (and probably they) was trying to convey was a
number of logical data channels interleaved on a single serial
physical channel. The point I raised was that a device, such as a
Flash memory. that could access various parts of it's anatomy in true
simultaneous "parallel" mode may then be able to make decent use of a
higher speed channel than is presently available from USB2 or
'firewire'. You could in fact (of course) also access multiple
parallel hardware objects with a single logical channel by doing the
interleaving prior to the channel access.


           Russell


2007\09\24@065934 by Alan B. Pearce

face picon face
>Gamers, programmers & stock brokers for multi-monitor.  Possibly
>graphics people, but I've never seen it.  They all seem to like
>bigger screens.

Add CAD users for large screen/multi-monitor.

>I'll take multi-monitor (I have 3) over a bigger screen any day.

Me too, I have a two monitor setup for doing PCB CAD, put the schematic on
one monitor, and the layout on the other. But I would still like larger
monitors, especially when doing any draughting type CAD, but larger is
better for doing PCB layout.

2007\09\24@074419 by Peter Todd

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

On Sun, Sep 23, 2007 at 09:33:02PM +0800, Xiaofan Chen wrote:
{Quote hidden}

Interesting! While debugging via IEEE1394 is more of a side-effect of
it's architecture, turns out that debugging via USB is actually an Intel
standardized feature, with it's own protocol and what not. Unfortunatly
it has to be enabled to work, and not many host controllers do according
to the following:

http://winprogger.com/blog/?p=1

- --
http://petertodd.org
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)

iD8DBQFG96Hv3bMhDbI9xWQRAsGpAKCjQ71iSf8k5wq14EOWH6nwrnTwLwCfYy8R
+yTpEa5WZEs9AzPMNtDOrJQ=
=ZoXK
-----END PGP SIGNATURE-----

2007\09\24@093954 by Russell McMahon

face
flavicon
face
I want more monitors ! :-) - for photograph processing in particular
and for everything in general.
I use 2 at present and tried to make 3 work with no success. I see I
need to try again.
4 would probably be about right :-).

Bigger is nice but desk space and head motion is also a factor.
Philips 19" CRTs of various specs seem about OK. (The lovely Philips
109P is hard to beat for spec in the LCD world. Mass and depth is
"rather high").


       Russell


{Quote hidden}

2007\09\25@084740 by Howard Winter

face
flavicon
picon face
Russell,

On Mon, 24 Sep 2007 11:20:23 +1200, Russell McMahon wrote:

{Quote hidden}

Right, but that isn't parallel anything, it's multiplexed serial, which we've had for decades in various implementations.  With only one pair of data lines it is serial,
even if they're used for a number of unrelated data streams in quick succession.  Even ADSL, which uses multiple frequencies at once to parallel-up the data, is fed
by Ethernet at either end, so the data itself is flowing serially through the end-to-end link even if it's temporarily parallel in one (slower) part of the journey - I
don't believe the data present on the line at any one moment is unrelated.

> The point I raised was that a device, such as a
> Flash memory. that could access various parts of it's anatomy in true
> simultaneous "parallel" mode may then be able to make decent use of a
> higher speed channel than is presently available from USB2 or
> 'firewire'. You could in fact (of course) also access multiple
> parallel hardware objects with a single logical channel by doing the
> interleaving prior to the channel access.

But this could happen now with USB (even USB1!) or FireWire, because multiplexing is part of their design.  The fact that nobody has done it suggests either that
developing the access to multiple areas of memory on the card isn't cheap enough to make it worth it, or that they don't think it's important enough to users to
make it a selling point.  As far as I've seen so far, the limiting speed for access to memory cards seems to be the cards themselves, not the link from the card reader
to the PC - I've put CompactFlash cards into PCMCIA adaptors and the time taken to transfer data is the same as using a USB2 card reader.  

Incidentally, do the connectors on memory cards carry the data in parallel?  I've never looked at the pinouts, but CF seems to have plenty of scope for parallel bits,
whereas SD looks a bit sparse.

Cheers,


Howard Winter
St.Albans, England


2007\09\25@100157 by Lee Jones

flavicon
face
> As far as I've seen so far, the limiting speed for access
> to memory cards seems to be the cards themselves, not the
> link from the card reader to the PC - I've put CompactFlash
> cards into PCMCIA adaptors and the time taken to transfer
> data is the same as using a USB2 card reader.

Depends on the cards and the readers.  Not true with the
newer, higher end "professional" grade CompactFlash cards.

I'd been using PCMCIA and Cardbus (32-bit PCMCIA) adapters for
photo upload for several years.  PCMCIA (16-bit) was dog slow.
With Cardbus adapter and various CF cards, I get about 9 MB/sec.

Several months ago, I switched to a SanDisk Extreme Firewire
CompactFlash reader.  With their Extreme IV CF cards, I get
22-26 MB/sec upload speed on a laptop.  With the Extreme III
CF cards, I get 11-13 MB/sec.  With older, slower speed CF
cards, it tops out at ~8-9 MB/sec.

Before my son left to study abroad, he got a camera that uses
SD cards.  I selected SanDisk's Extreme USB2 reader that does
both SD & CF cards.  Hooked up via a USB2 port during testing,
a SanDisk Extreme III SD card uploaded at 11 MB/sec -- about
the same as the Firewire reader with an Extreme III CF card.

However, using the same Extreme IV CF card (from above), USB
reader uploaded at 14 MB/sec -- limiting factor appears to be
USB2 link and/or USB reader.  Personally, I prefer Firewire.

                                               Lee Jones

P.S. I also see a speed difference in camera operation during
    burst mode with RAW files when my older, slower CF cards
    are inserted.

2007\09\25@100639 by Paul Hutchinson

picon face
EE Times has an article about USB 3.0 this week and they included a picture
of the new cable design (picture not included on web page, only in the
subscription version). The new cable has the traditional USB 2.0 four
conductors plus 6 additional conductors split into 2 channels for the USB 3
functionality. Essentially you end up with 3 parallel serial channels per
cable. So, existing cables and connectors won't do USB 3.0. IEEE-1394 on the
other hand can increase it's speed up the USB 3.0 targets with the existing
cables and connectors. Since there are 3 parallel serial channels maybe they
should rename it to UPSB 3.0 :-).

Paul

> {Original Message removed}

2007\09\25@102716 by Xiaofan Chen

face picon face
On 9/20/07, Russell McMahon <.....apptechspamRemoveMEparadise.net.nz> wrote:
> USB 3 due from about 2009 on
> Amongst it's many other capabilities it should utterly transform data
> transfer rates to/from Flash memory cards.

Some interesting discussions from usb.org forum.
http://www.usb.org/phpbb/viewtopic.php?t=12563
http://www.usb.org/phpbb/viewtopic.php?t=7094

2007\09\25@102836 by Russell McMahon

face
flavicon
face
>> The concept that I (and probably they) was trying to convey was a
>> number of logical data channels interleaved on a single serial
>> physical channel.

> Right, but that isn't parallel anything, it's multiplexed serial,
> which we've had for decades in various implementations.

It's good engineer speak in one sense to say it's not parallel iof
there's only one pair/fibre ... BUT unless one wishes to totally
dispense with the ISO (or any) multi layered model then we must forgo
the luxuries of such obviousness.

It's parallel anything you want at a high enough layer. ie it doesn't
matter what the physical layer implementation is (as long as it works
[tm]), it's how it can be made to look to the applications. I believe
that intel were making the point that theyw ere hoping to make the
pipe hydra headed so that it could transparently and easily [tm] be
used by multiple applications at once if desired.

A made up example might be that a camera could do direct to disk
writes of photos on a connected PC as the photos were taken (as one of
my Minoltas can do now) AND simultaneously a different PC application
could access the camera's onboard flash memory, which is not being
accessed by the camera for storage at this time in this mode. This
would require the camera to be able to handle the load and multiple
tasks but would be completely doable by (even) Windows with no mods
right now.

> With only one pair of data lines it is serial,

At the physical layer, yes.

>> The point I raised was that a device, such as a
>> Flash memory. that could access various parts of it's anatomy in
>> true
>> simultaneous "parallel" mode

> But this could happen now with USB (even USB1!) or FireWire, because
> multiplexing is part of their design.

Not sensible with USb1 as current flash cards are faster than it is
able to handle. USB1 is about "12X" using the standard flash card
speed rating system (which is the orriginal CD rating system). I have
a 1X external CD writer which I long ago paid good money for and which
is now very very much a museum piece :-).

> The fact that nobody has done it suggests either that
> developing the access to multiple areas of memory on the card isn't
> cheap enough to make it worth it, or that they don't think it's
> important enough to users to
> make it a selling point.

AFAIK none of the systems will presently do what Intel is talking
about - although layers could be added to do it. You can write
applications which merge and demerge data onto the single logical
pipe, but Intel are talking about having a multiple pipe at some
logical layer level.

> As far as I've seen so far, the limiting speed for access to memory
> cards seems to be the cards themselves, not the link from the card
> reader
> to the PC - I've put CompactFlash cards into PCMCIA adaptors and the
> time taken to transfer data is the same as using a USB2 card reader.

As I noted, the recently announced and yet to ne seen latest
generation of SD cards will push USb2 to or beyond its limit and the
next lot after that will be fully bottlenecked by USb2.

> Incidentally, do the connectors on memory cards carry the data in
> parallel?  I've never looked at the pinouts, but CF seems to have
> plenty of scope for parallel bits,
> whereas SD looks a bit sparse.

CF is a subset of ISA and is 8 bit parallel (AFAIR).

SD has grown out of MMC. MMC had a single bit serial mode, essentially
SPI, and a nibble mode and all SD cards are *meant* to be backwards
compatible with the serial mode, although I believe that some recent
cards do not support the serial mode. The recent large SD cards (2GB
to 32 GB) support a new access mode (SDHC)

This en.wikipedia.org/wiki/Secure_Digital_card
Says
SD supports at least three transfer modes:

 a.. One-bit SD mode (separate command and data channels and a
proprietary transfer format)
 b.. Four-bit SD mode (uses extra pins plus some reassigned pins)
 c.. SPI mode (basically, a simpler subset of the SD protocol for use
with microcontrollers)
All memory cards must support all three modes, except for microSD
where SPI is optional. The cards must also support clock frequencies
of up to 25 MHz for regular cards, and 50 MHz for high-speed cards.

The comments about speed in the above article seem to be well out of
date.

CF has been the card of choice in top end cameras until now but SD
seems to be taking over.
I now use only Transcend 133X CF cards and they return in excess of
that speed on read and write.
133X is 17.3 MB/sec and I get up to about 20 MB/second on read.
Even at that speed it takes 8 minutesto read an 8 GB card. USB2 is
fast enough for that but not the MUCH faster speeds that the latest SD
are promising.


       Russell

2007\09\25@124525 by Herbert Graf

flavicon
face
On Tue, 2007-09-25 at 13:47 +0100, Howard Winter wrote:
> Incidentally, do the connectors on memory cards carry the data in parallel?  I've never looked at the pinouts, but CF seems to have plenty of scope for parallel bits,
> whereas SD looks a bit sparse.

How "parallel"? All the major ones I can think of can carry data in
parallel form. CF is parallel (8bit). SD has a parallel mode (4bit). MS
can also do 4bit IIRC.

TTYL



2007\09\25@143050 by peter green

flavicon
face

> Incidentally, do the connectors on memory cards carry the data in parallel?  I've never looked at the pinouts, but CF seems to have plenty of scope for parallel bits,
> whereas SD looks a bit sparse.
>
>  
The following is all from memory, i'm pretty sure it's correct but there
may be some minor errors

A CF card can act as either IDE or PCMCIA (with a mode select pin to
switch between them) which are in turn I belive both based on ISA which
is a 16 bit paralell bus.

MMC cards use a SPI based serial interface

SD cards can operate in either the same SPI based serial mode as MMC or
a 4 bit paralell mode.

Not sure about other card types.

2007\09\25@150317 by Howard Winter

face
flavicon
picon face
Adam,

On Fri, 21 Sep 2007 10:59:19 -0400, M. Adam Davis wrote:

> On 9/21/07, Howard Winter <RemoveMEHDRWspamspamBeGoneh2org.demon.co.uk> wrote:
> > I wonder what USB3 would be used for?  External disks are already availble with SATA connectors, so I don't see the need for a faster interface there, and I
can't
> > think of anything else that would need speeds above USB2 that doesn't already have a standard (Gigabit Ethernet, for example).
>
> The other standards typically do not:
>  * Provide power to the device

Power Over Ethernet?

>  * Truly support plug and play

Neither does USB - just connect an ICD2 before loading the software and you'll be in for an hour or so's playing that you weren't planning for!  :-)

>  * Support all the various transfer modes (isochronous, bulk, etc)

These are only there because USB is designed with them - Ethernet, for example, has nothing of the sort but still seems to work - they are really only a scheduling
scheme for the multiplexing.

>  * Support On the Go functionality - you can't hook two SATA devices
> together and make a backup.  

I'm not sure you could do this whatever the connection type with just the bare connection - it needs some logic and control as well as the actual interface, and
adding that does not mandate any particular interface.  USB as designed was always a master-slave scheme, until some people added "USB Network" cables which
drive a coach and horses through the standards.  On The Go is a kludge to make the standard do something it wasn't designed for.  Any other standard can be
similarly kludged.

> If you do this with ethernet you need a
> special cable or adaptor, and will probably have to configure the
> network yourself.

Why?  Most hubs nowadays cope with the Tx/Rx reversal or not as needed - no reason why devices couldn't do that too.  And what's to stop two devices
negotiating addresses between themselves?  DHCP is long-established, I can't imagine it's hard to tweak it to work in a situation where neither device is designated
as master.  Some DHCP servers will already stand down if they find that another one is already active, you just need to sort out which is client and which server at
the time.

> Further, would you rather have 20 ports of 5 different types on your
> computer (probably in the back) and have to match the right peripheral
> to the right port?

Well 20 is an exaggeration - I just had a look at a "Legacy" PC, and it has 12 connectors, 3 of which are audio (I hope you're not proposing to make headphones &
microphones into USB devices - that would be daft) and the line-in socket will always be needed for connecting HiFi or portable devices.  I don't see why it's
necessary to have all one type of connector when the existing ones are fine - I have no problem plugging a mouse and keyboard into PS/2 sockets - once they're
there they tend to stay for a very long time, and I use a KVM switch anyway, so it's actually what I want.  

>  What if the intended port is full, do you unplug
> something, add another card, or drop another port expander (if the
> interface supports that) onto your already crowded desk?

I'm not sure what you're saying here - I was in favour of FireWire which is daisy-chained - you can have just one port on the PC and plug all the peripherals into
each other.  USB needs a Hub to have more than one device per PC socket, so I really don't get the point you are making.

> In fact, if USB can do the things those other interfaces accomplish,
> wouldn't you simply rather have one type of port on the computer which
> supports pretty much anything you could want to plug in?

Since every non-laptop PC has a basic set of devices permanently connected - monitor, mouse, keyboard - I see no reason to have these on general-purpose
sockets.  When it comes to other, high speed interfaces, there is a case for standardisation but manufacturers will always scrimp where they can, so I think you'll
never see 20 USB sockets on a PC - the most I've seen on a PC is six, and it wasn't for want of space.  So you are likely to need a hub at some point whatever type
of connection you use.

> The reasons for faster transfer speed are obvious for existing
> applications.  I have a 4GB CF card for my camera, and I'd like it to
> download all my pictures faster than it now does.

You may find the card can't do that - I've seen some *dreadfully* slow memory devices.  I have a USB "stick" that takes 20 minutes to download a gigabyte, and it's
not the interface that's the limit.

> But there are lots of applications you may not have considered simply
> because 1) it's not available now and 2) _you_ wouldn't need it...
> until you saw it

Oh, there are quite a few things I don't need even after I've seen them!  :-)

> For instance, I expect to see high performance video cards plugging
> into such a port.  Right now peripheral manufacturers have to sell
> them to OEMs to install, or convince the end user to perform the
> install.  On this list that isn't a problem, but for 80% of the
> computer owning population they don't ever want to open up their
> computer.  There are many products in this category.

I think your 80% is a vast underestimate!  But there have been external devices in the video area for some time, such as the "Dazzle" from Iomega (I think) which
allows video cameras (digital and analog) and other A/V equipment to be connected easily, with just a single PC connection.  They aren't selling like hot cakes...  I
think the market for really high performance PC video that you can plug-&-play is rather small.  I have a terestrial digital PVR (twin tuners, records to HDD) and it
has a USB2 connection.  But it's designed for geeks, and is one of only two that have any sort of PC connection - all the others expect you to watch the material on
television and then delete it - very few people seem to want to pay extra for something like this that communicates with a PC.

> In fact, I expect to see monitors come out that only attach to this
> connector, and include high performance graphics engines.  Need
> another monitor?

Most people I know would never dream of having more than one monitor - that's definitely high-end geek!  :-)

> Don't count the number of monitor ports, just buy
> one and plug it in, including the latest huge, high resolution
> displays.  People don't want to worry about whether their graphics
> card is dual link capable, they just want to buy a 30" display and
> plug it in.  This will become more of an issue as even the "small and
> cheap" monitors become 1080p compatible and beyond.

But by moving the image-generation logic to the monitor you're making it much more expensive.  People change monitors more often than graphics cards, and this
would just increase the cost of doing that.  And where on *earth* do you put a 30" monitor?  Most offices would have no way to get the person far enough from
the screen to make it usable, and most home users have their computer desk against a wall, so they'd have to move it back into the room to get a decent view of
the screen without doing an impression of a Wimbledon crowd!  I think the largest screens we have now (22" widescreen) are likely to be close to the limit of what
you'd have on a desk - and I think the market for a general-purpose computer used for watching TV and video on large screens is not very large.  Special purpose
computers ("set top boxes") may actually be PCs inside, but since they are specialised to home entertainment I don't see the need for the multiple general purpose
interfaces there either.  

> High performance computing clusters come closer to home every year.

Only for some values of "home"!  :-)  For every home that has a high-performance cluster, there will be 10 to 100 that have just a single computer that's only used
for email, IM, web surfing and the occasional bit of printing.  IMH, of course!

> One might choose, for instance, to buy a computer and drop it on top
> of their old one.  Plug the two together with such a fast connection
> and you can treat it essentially like a single multi-core computer.

But even firms with a close control of their market - Apple for example - keep producing new machines that are incompatible with earlier ones.  They want you to
buy a whole load of new stuff, not just something to upgrade what you bought last year.  And the rest of the PC market is even worse!

> This is already something done for graphics cards in SLI
> configurations, but imagine having a connection fast enough that
> rather than upgrading your computer so you can do video work quickly,
> you simply buy a video coprocessor to do your movie manipulation and
> encoding at 10-100x realtime.  And you don't have to even move your
> computer - just plug it in and go.  Although it'll require a separate
> power supply - there's only so much you can do with 2.5W!

While this would be great, I think it's just not mainstream-enough to drive the technology.

> But I think that the ultimate useful reason is simply media transfer.
> Soon 1080p camcorders will be on the prosumer market, and even with
> MPEG4 you're looking at transferring 50GBytes for a clip.  No one
> really wants to wait 13 minutes (best case USB2.0) to transfer that to
> their computer.  When you start seeing that 1TB drives are common, and
> 10TB drives are readily available, you'll be wondering where this spec
> is.

I just don't think that being able to see wobbly hand-held home videos in beautiful 1080p resolution justifies the technology.

Now, I must find my wooden clog and wind up my throwing arm... :-)))

Cheers,


Howard Winter
St.Albans, England



'[EE]:: USB 3 plans'
2007\10\01@123520 by Martin Klingensmith
face
flavicon
face
Xiaofan Chen wrote:
> On 9/23/07, Martin Klingensmith <spamBeGonemartin@spam@spamspam_OUTnnytech.net> wrote:
>  
>> I just bought a new nvidia card and I'm
>> amazed by what it can do in linux.
>>    
>
> Some examples please so that I can get some good reasons
> to buy a new computer with Vista. I will then have it dual boot with Linux.
> Maybe it will be a good time when Vista SP1 is out.
>
> Xiaofan
>  
My amazement is based on the fact that I had an ATI card in linux for 3
years and getting it to do anything other than accelerate web-page
scrolling was inherently futile. I've played a few free 3D games based
on Quake III with the quality settings maxed out and it runs at 85Hz
screen refresh rate. If you don't play any games you can be assured that
any graphical things you want to do will work well. Often times cheap /
on-board graphics can't do the special resolutions for some wide-screen
monitors. This card also has 2 ports so that a dual-headed setup is simple.
--
Martin K

More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...