STB Velocity 128 is out!
kba...@halo.aeromet.com
I saw the STB Velocity 128 in it's 8 meg version on the pricewatch
specials section for $269.00. I called to order it and they had it in
stock. Unfortunately it's an OEM version with no bundled software! Rip
Off! Anyone else who doesn't care about a decent bundle should go get it.
Later,
Kevin
P.S. For those who don't know http://www.pricewatch.com
-------------------==== Posted via Deja News ====-----------------------
http://www.dejanews.com/ Search, Read, Post to Usenet
Randall Bagwell
I couldn't find the Velocity 128 on Price Watch. I am sure what you saw
was the Velocity 3D which is based on the Virge/VX. The Riva 128
chipset can't support more than 4MB of onboard RAM.
Randall Bagwell
rbag...@hiwaay.net
Volker Oth
kba...@halo.aeromet.com schrieb:
>
> I saw the STB Velocity 128 in it's 8 meg version on the pricewatch
> specials section for $269.00. I called to order it and they had it in
> stock. Unfortunately it's an OEM version with no bundled software! Rip
> Off! Anyone else who doesn't care about a decent bundle should go get it.
Sounds like you mixed the STB Velocity 128 with the Stingray 128 3D.
The Riva128 chipset doesn't support 8MB at all and the Velocity
should be cheaper, too.
Volker
Randall Bagwell
Mark Pickavance wrote:
>
> I found it, I'd look a little harder.
> And... the Riva can support more than 4Mb's, so I suggest you go to the STB
> site and get some information.
>
> --
> Mark Pickavance
I have information from both STB and nVidia stating that this chipset
only supports 4MB onboard memory. But I went to STB's site, anyway, and
I could not find where they list an 8MB version or even an option to
upgrade to 8MB.
Randall Bagwell
rbag...@hiwaay.net
Robert Schlabbach
Mark Pickavance <ma...@picknmix.demon.co.uk> wrote in article
<01bcae09$50ea43c0$c8646464@it01>...
> I found it, I'd look a little harder.
> And... the Riva can support more than 4Mb's, so I suggest you go to the
STB
> site and get some information.
Why don't you try that yourself and give an exact URL?
The RIVA128 is restricted to 4MB frame buffer.
Regards,
--
Robert Schlabbach
e-mail: rob...@powerstation.isdn.cs.TU-Berlin.DE
Technical University of Berlin, Germany
Kagetora
In article <8721137...@dejanews.com>, kba...@halo.aeromet.com wrote:
#I saw the STB Velocity 128 in it's 8 meg version on the pricewatch
#specials section for $269.00. I called to order it and they had it in
#stock. Unfortunately it's an OEM version with no bundled software! Rip
#Off! Anyone else who doesn't care about a decent bundle should go get it.
Ummmm....the Velocity 128 doesn't come w/ 8 megs...and it likely never will.
nVidia's chip, reportedly, won't deal w/ more than 4 megs of onboard RAM.
What you likely saw was the Velocity 3D, which IIRC has a ViRGE chip (shitty).
"Power is in tearing human minds to | Jon Hickman
pieces and putting them together again | Oni Dragon -=={UDIC}==-
in forms of our own choosing." - _1984_ | Kage...@pernet.net
(Note: Because of Sensless Acts of Spamming (SAS), I am forced to
disable the header's e-mail...just use the one on the .sig if you
really want to send me something)
Matthew Doughty
Mark Pickavance (ma...@picknmix.demon.co.uk) wrote:
: I found it, I'd look a little harder.
: And... the Riva can support more than 4Mb's, so I suggest you go to the STB
: site and get some information.
: --
: Mark Pickavance
Your very confident for somone so completely wrong.
I. the STB velocity 128 is not out yet.
II. there is no listing for it on pricewatch.
III. When it does come out it will not have an 8 MB version.
IV. the riva chipset only supports 4 MB that is one of the complaints
about it
--
Matt Doughty
Alf
kba...@halo.aeromet.com wrote:
>I saw the STB Velocity 128 in it's 8 meg version on the pricewatch
>specials section for $269.00. I called to order it and they had it in
>stock. Unfortunately it's an OEM version with no bundled software! Rip
>Off! Anyone else who doesn't care about a decent bundle should go get it.
> Later,
> Kevin
>
>P.S. For those who don't know http://www.pricewatch.com
Thought I'd jump in here and clear this up. That is patently
impossible... the V128 has not even started shipping, so they couldn't
have it, it is a 4MB only card, and the RETAIL version costs less.
This is a definite case of mislabeling. Perhaps it is a Velocity 3D,
but it is almost certainly NOT a Velocity 128.
Alf
STB Tech Support
[remove the * from the address to contact me via e-mail]
________________
STB Systems Inc.
Web site: http://www.stb.com/
FTP site: ftp.stb.com
BBS: (972) 437-9615
Mark Pickavance
I found it, I'd look a little harder.
And... the Riva can support more than 4Mb's, so I suggest you go to the STB
site and get some information.
--
Mark Pickavance
Randall Bagwell <rbag...@hiwaay.net> wrote in article
<33FBB9F6...@hiwaay.net>...
Rob Rodgers
>Mark Pickavance wrote:
> I found it, I'd look a little harder.
Where? Nowhere I've checked has it in stock.
Matt Conway
Rob Rodgers (rsro...@wam.umd.edu) wrote:
: >Mark Pickavance wrote:
: > I found it, I'd look a little harder.
: Where? Nowhere I've checked has it in stock.
I'm not sure exactly where I saw it, but I've seen it advertised in a couple places.
However, none of them that I called have them in stock yet because they haven't been
shipped. So sure you might see them advertised, but that doesn't mean they have them.
--
Matt Conway
Georgia Tech - Computer Engineer
gt2...@prism.gatech.edu
"All the world's indeed a stage" - Rush "Limelight"
James Taylor
I talked to the people at PCConnections and they said the expected
shipping date is August 29 for the Velocity 128.
james
ASSFUNGUS wrote:
> Velocity 128 is expected to be out around the 2nd week of September.
> Hercules Thriller (V2200) is scheduled to be out by August 29.
Alf
youn...@hawaii.edu (James Young) wrote:
>I don't understand it either.. the Velocity 128 is listed on the last
>page of the STB cards. It says Velocity 128 8 megs SGRAM 1.6GB, so it
>sure sounds like the Riva card. I called up their 1-888 number and
>they said it did use the Riva chipset.
The facts are that STB has not shipped the Velocity 128 to ANY vendor
and there are no plans for an 8MB version. That tends to preclude any
vendor actually having a Velocity 128, let alone an 8MB version.
Sorry for the confusion and thank you very much for bringing it to our
attention.
Jedi
Seems like end users want an 8mb capable board, why not give us the option
to expand the board to 8mb.
Regards
Alf wrote in article <33fde93a...@STB.COM>...
Kagetora
In article <5tm777$kek$1...@columbine.singnet.com.sg>, "Jedi" <je...@hotmail.com> wrote:
# Seems like end users want an 8mb capable board, why not give us the option
#to expand the board to 8mb.
Because it's not STB's option to give...it's nVidia's chip design. From all
reports, including nVidia's, the Riva 128 *will not* support more than 4 Megs
of VRAM (whatever flavor) on-board. Sure, STB could slap all the RAM on the
damn thing it wants, but it isn't going to do anything but drive up the
cost...for no benefit whatsoever.
jlw...@{fightspam}mindspring.com
>Because it's not STB's option to give...it's nVidia's chip design. From all
>reports, including nVidia's, the Riva 128 *will not* support more than 4 Megs
>of VRAM (whatever flavor) on-board. Sure, STB could slap all the RAM on the
>damn thing it wants, but it isn't going to do anything but drive up the
>cost...for no benefit whatsoever.
....like most of the people bitching would even notice.
Rik
swanlee
On 21 Aug 1997 17:38:12 GMT, dou...@arches.uga.edu (Matthew Doughty)
wrote:
>Mark Pickavance (ma...@picknmix.demon.co.uk) wrote:
>: I found it, I'd look a little harder.
>: And... the Riva can support more than 4Mb's, so I suggest you go to the STB
>: site and get some information.
>:
>: --
>: Mark Pickavance
>Your very confident for somone so completely wrong.
>I. the STB velocity 128 is not out yet.
>II. there is no listing for it on pricewatch.
>III. When it does come out it will not have an 8 MB version.
>IV. the riva chipset only supports 4 MB that is one of the complaints
>about it
Man whatever I saw it today at computer city here in atlanta for 299$
jlw...@mindspring.com
On Sun, 24 Aug 1997 02:51:56 GMT, swa...@avana.net (swanlee) wrote:
>
>Man whatever I saw it today at computer city here in atlanta for 299$
"Man whatever..." you either saw the Velocity 3D or a very over priced
Lightspeed 128. STB screwed up by naming this "Velocity".
Rik
Kagetora
In article <3400acc3...@news.mindspring.com>, jlw...@mindspring.com wrote:
#"Man whatever..." you either saw the Velocity 3D or a very over priced
#Lightspeed 128. STB screwed up by naming this "Velocity".
They've done this before...there was a Lightspeed before there was a
Lightspeed 128, which confused a friend of mine when he went up to Best Buy
and tried to get a LS128 and ended up with the lesser version.
Condor
"Jedi" <je...@hotmail.com> wrote:
> Seems like end users want an 8mb capable board, why not give us the option
>to expand the board to 8mb.
Yeah. That's why I'm so interested in the V2200. The first boards will
come with the option of 8MB and are upgradable to 16MB. The Riva can't
even come close to competing with that.
William J. Ener
Condor wrote:
Anyone seen any numbers on the V2200 yet?
TormatoJES
>I saw the STB Velocity 128 in it's 8 meg version on the pricewatch
>specials section for $269.00. I called to order it and they had it in
>stock. Unfortunately it's an OEM version with no bundled software! Rip
>Off! Anyone else who doesn't care about a decent bundle should go get it.
This is not the Velocity 128! It's the Velocity 3D. The Velocity 128 is
not out yet and it will only be available in a 4MB card. Check
www.stb.com for more details.
TormatoJES
>>I. the STB velocity 128 is not out yet.
>>II. there is no listing for it on pricewatch.
>>III. When it does come out it will not have an 8 MB version.
>>IV. the riva chipset only supports 4 MB that is one of the complaints
>>about it
>
The guy speaks the truth...it is not out yet and it will only support 4MB
of memory. You are seeing the Velocity 3D.
Dwight Diercks
Condor wrote:
>
> "Jedi" <je...@hotmail.com> wrote:
>
> > Seems like end users want an 8mb capable board, why not give us the option
> >to expand the board to 8mb.
>
> Yeah. That's why I'm so interested in the V2200. The first boards will
> come with the option of 8MB and are upgradable to 16MB. The Riva can't
> even come close to competing with that.
PCMag just did a review with systems having the RIVA128
and those from ATI(8MB) and #9(8MB). By the way, I run
GPolice here no problem at 1024x768 on my RIVA128. Notice
that even though the ATI and #9 used onboard memory, the RIVA128
texture access was still FASTER. Don't make broad statements
about competing till you seen the results.
An excerpt from the review below:
ZD 3D WinMark 97 AGP Once again, machines running graphics cards based
on the nVidia Riva 128 chip produced significantly higher scores than
others. There is a 61 percent difference between the high-scoring Micron
Millennia and the low-scoring NEC Direction. The AST Bravo's ATI Rage
and the NEC Direction's Number Nine Revolution 3D graphics boards both
had 8MB of on-board frame-buffer memory, compared with all others, which
had only 4MB. The extra memory allowed these boards to contain all
textures from our AGP test in the frame buffer (our test uses up to 5MB
of textures). Therefore, in the AST and NEC machines, processing of
these textures was left entirely to the graphics processor and local
frame buffer.
http://www8.zdnet.com/pcmag/features/300p2/_open.htm
Dwight
Terry Harris
Dwight Diercks <dwi...@nvidia.com> wrote:
>> > Seems like end users want an 8mb capable board, why not give us the option
>> >to expand the board to 8mb.
>>
>> Yeah. That's why I'm so interested in the V2200. The first boards will
>> come with the option of 8MB and are upgradable to 16MB. The Riva can't
>> even come close to competing with that.
>PCMag just did a review with systems having the RIVA128
>and those from ATI(8MB) and #9(8MB). By the way, I run
>GPolice here no problem at 1024x768 on my RIVA128. Notice
>that even though the ATI and #9 used onboard memory, the RIVA128
>texture access was still FASTER. Don't make broad statements
>about competing till you seen the results.
And what exactly is anyone supposed to understand from the results?
That the AGP version of Riva with 4MB of RAM works better than the
competition on a 300MHz PII system with LX chipset (only officially
announced today) and SDRAM on the motherboard?
Wonder how the comparisons would be on PCI versions.
Cheers Terry...
steven joseph chmura
In article <3401F7...@prodigy.net>,
ASSFUNGUS <ASSF...@prodigy.net> wrote:
>Alf wrote:
>>
>> con...@cryogen.com (Condor) wrote:
>>
>> >Yeah. That's why I'm so interested in the V2200. The first boards will
>> >come with the option of 8MB and are upgradable to 16MB. The Riva can't
>> >even come close to competing with that.
>>
>> is, why do you need it? What applications are you going to be running
>> that require 8MB? And as for 16MB... can you name a single
>> application that would require it? Extra memory is great, but if
>> you're not going to be using it, it's just extra money.
>>
>> Alf
>> STB Tech Support
>
>There's only one game that I know of that uses that extra textured
>memory - Quake2. Is it worth spending an extra $80 for that extra
1. All games based on QUake2 will poll for texutre memory
2. ZOID has hinted at an update for QW2.x
3. ALL direct3D games can use it - the driver polls for whne the
page swap is needed
Thus iF22, Flying Corps etc. will ALL use the memory. You could run
iF22 or Flying corps in 1024x768
--
________________________________________________________________________________Steven Chmura MuDPhud4
1st Zoloft Lawndart Member "... no frame rate too high"
Alf
criple...@lookinmysig.net (Kagetora) wrote:
>In article <3400acc3...@news.mindspring.com>, jlw...@mindspring.com wrote:
>#"Man whatever..." you either saw the Velocity 3D or a very over priced
>#Lightspeed 128. STB screwed up by naming this "Velocity".
>They've done this before...there was a Lightspeed before there was a
>Lightspeed 128, which confused a friend of mine when he went up to Best Buy
>and tried to get a LS128 and ended up with the lesser version.
We apologize for any confusion our naming convention may have caused.
For customers who are uncertain (ie. Lightspeed/Lighspeed 128), feel
free to visit our web site and jot down some information on the card.
Before you leave the store, just double-check what you purchased
against what you wanted to purchase and make sure they match up.
However, a vendor advertising an STB product that does not exist is is
quite a different matter. Vendors are responsible for correctly
advertising products they sell (for example, the Lightspeed and
Lighspeed 128 should be named and advertised accordingly and if they
are not, it is the vendor that is responsible for the mistake) and
there is no margin for error. That would be like "Joe Bob's Ford"
advertising a Mustang GT to find out they actually carry only a plain
Mustang, and then blaming Ford for the confusion. Vendors should not
advertise what they do not have and what is not available.
Alf
sjch...@midway.uchicago.edu (steven joseph chmura) wrote:
>Alf - what the hell are you smoking? Every designer is BEGGING for more
>memory and you are defending a flawed design decision in this newsgroup?
>Please :)
In what sense is it flawed?
>Seriously, I can't believe someone as knowledgeable as yoursefl would look
>like such an idiot in a public forum:
I really don't see how I'm looking like an idiot here.
>1. Sierra: They have canned 3D support for Red Baron II and ProPilot due
>to only 4 meg cards. They have publically PLEADED for more memory
AGP is right around the corner, which should give game designers all
the memory they care to work with without having to charge the
consumers extra for cards with more memory. At this time, however,
4MB is plenty for virtually any game there is.
>2. Look at the "Texture" load problems of iF22, Flying Corps, GLQuake. All
>of this the designers have stated are due to SWAPPING in and out texture
>memory.
The main problem there is actually bandwidth restriction. The
Velocity 128 is a bus mastering card with wider bandwidth capability
and an internal cache, which can make up for a lot of that even in PCI
configuration.
>So for there here and now (assuming GLQUAKE 2.01 has this patched as
>promised) a 8 meg card would allow:
>800x600
>Z-Buffer
>50% less Texture thrashing
>if22 woulds use ut (direct3D all does)
>FlyingCorps Direct3D (again, Direct3D will automagically use the texxture
>memory)
>BOtht of the above sims are CODED ALREADY for 1024x768 3D hardware is
>anyone makes it
>F22:ADF will be DirectX5 and use it (also 3DFX specific)
>I cannot give more examples without busting and NDA. But thesis should
>suffice.
For good realtime performance at greater than 640x480 resolutions with
true color, a wider bandwidth is needed. Basically, this means AGP
because PCI is too limited. Consequently, AGP makes extra memory
meaningless because it uses system memory for these functions.
>________________________________________________________________________________Steven Chmura MuDPhud4
>1st Zoloft Lawndart Member "... no frame rate too high"
There's a reason game designers are looking forward to AGP. In order
to take full advantage of these applications when they become
prevelant (which is not the case currently), AGP architecture is
almost a must for data throughput. For PCI and current applications,
4MB should be ample. For AGP and future applications, it is not even
an issue.
Terry Harris
ACOVEY@*STB.COM (Alf) wrote:
>For good realtime performance at greater than 640x480 resolutions with
>true color, a wider bandwidth is needed. Basically, this means AGP
>because PCI is too limited.
Perhaps you could explain that, I really don't see how a 3D
accelerator needs any additional data from the system to render an
image at 800x600 than it does to render the same image at 640x480, and
so it doesn't need any additional bandwidth on the system bus. Except
perhaps if it uses larger more detailed textures but then this whole
argument is about the need for additional on card texture memory.
I don't know why you don't come clean and say that the Riva's 4MB
memory limit will be a sever limitation on the PCI version instead of
spinning some crap about AGP being required for high performance.
AGP is not required for performance, memory is required for
performance. AGP is required for cheapness by using system memory
instead of on card memory. Not that system memory or system memory
bandwidth is free (a point AGP proponents conveniently forget).
While AGP might be cheap making do with less on board memory the rest
of the system required to support AGP is a horrendously expensive
upgrade.
Hmmm, I think I will scrap my motherboard, and PPro 200 and 128MB of
simms and current video card(s) so I can buy a new Pentium II 266 and
a new motherboard and 128MB of SDRAM to enable me save $50 on an new
AGP video card because it only needs 4MB of memory - yeah sounds like
a bargain.
Cheers Terry...
ARCHivisT
<snip>
> There's a reason game designers are looking forward to AGP. In order
> to take full advantage of these applications when they become
> prevelant (which is not the case currently), AGP architecture is
> almost a must for data throughput. For PCI and current applications,
> 4MB should be ample. For AGP and future applications, it is not even
> an issue.
This is gone too far!
I'm sick and tired of hearing hype statements about how good AGP will be
for 3D cards. This is what INTEL wants you to think. In fact that is
about 90% false (marketing) and 10% true. AGP will accelerate mostly 2D.
It will accelerate 3D only IF the card has a lot of display memory too,
to store textures in local memory. I will explain...
As you know a computer with an accelerated 2D/3D card functions as a
multiprocessor system. The two processors (CPU and 2D/3D graphics
processor) have their own local memory and are connected to each other
by the PCI or AGP bus. What happens on the local memory of each
processor and the communication that takes place between the two is very
different deppending on the type of application: 2D or 3D.
In 2D:
The local processing of the graphics card (2D) is not very intensive,
consisting mainly in bit-blit operations, line draw and fill operations.
Most 2D graphic intensive operations consist in transferring large
bitmaps from the CPU main memory to the graphics board local memory.
Motion video displaying consists of this fast bitmap transfers from main
memory to graphics memory, even if in some cases the MPEG decoding is
being made inside the 2D chip. These transfers are the ones that benefit
mostly from AGP wider bandwidth.
3D:
The local processing of the graphics card (3D) is very intensive. So is
the processing of the main CPU. Both memory buses (main and graphics)
will be very busy feeding both processors with data. If the 3D card has
all the textures in its onw memory, it doesn't have to interrupt the
main CPU processing to access the texture in main memory, thus freeing
the main memory bus for the main CPU use. Plus, the graphics local
memory bus is faster than the main memory bus, and even faster than PCI
or AGP.
So a graphics card with local texture memory will be always faster than
a card that has to access the textures from main memory. Processing data
(textures) on the graphics local bus WILL ALWAYS BE FASTER, be they AGP
or PCI cards. This is the way to maximize each processor throughput.
The only REAL benefit from AGP for 3D cards will be in the download of
the textures and 3D commands from main memory to graphics memory. But
that's only a small percentage of the 3D work...
I hope you understand why the use of texture memory in maim memory is a
false advantage presented to you by Intel. Using main memory for
textures will be like having the UMA model all over again... And
remember: UMA had direct access to memory, AGP will have a bridge (an
AGP bridge, like a PCI bridge) and a comunication protocol overhead...
ARCHivisT
steven joseph chmura
In article <34066fbc...@news.alt.net>,
Hardball <h/a/r/d/b/a/l/l/@/a/l/t/./d/o/t/./n/e/t> wrote:
>On Wed, 27 Aug 1997 18:56:25 GMT, ACOVEY@*STB.COM (Alf) wrote:
>
>>AGP is right around the corner, which should give game designers all
>>the memory they care to work with without having to charge the
>>consumers extra for cards with more memory. At this time, however,
>>4MB is plenty for virtually any game there is.
>
Alf,
You know this is bullshit.
GLQuake
iF22
Flying Corps
Hexen2
Need I mention MORE games that the designers are crying for
more texture ram?
Red Barron II
Pro Pilot
Both Canceled Due to lack of texture memory (4 megs)
Tell me, How the HELL can you get 800x600 Z-buffer with 2 megs
of frame buffer memory?
Ah, dynamically take it away from the texture memory. So now
GLQuake will thrash with 2 people in a room.
--
Jeff Atwood
Yeah, but texture thrashing is already a problem on regular 3dfx cards. And
I hate to limit developers to 2 or 4 mb of texture ram; it'd be nice if we
could just use that as a texture CACHE (ie 90% of time the texture will be
on the card) and then the other 10% of the time AGP could be used so texture
thrashing wouldn't be as bad.
Jeff
ARCHivisT wrote in article <3405406E...@tp.telepac.pt>...>
>This is gone too far!
>
>I'm sick and tired of hearing hype statements about how good AGP will be
>for 3D cards. This is what INTEL wants you to think. In fact that is
>about 90% false (marketing) and 10% true. AGP will accelerate mostly 2D.
>It will accelerate 3D only IF the card has a lot of display memory too,
>to store textures in local memory. I will explain...
>
[snip]
Matthew Doughty
Alf (ACOVEY@*STB.COM) wrote:
: con...@cryogen.com (Condor) wrote:
:
:
: It's true that the Riva doesn't have the extra memory. The question
: is, why do you need it? What applications are you going to be running
: that require 8MB? And as for 16MB... can you name a single
: application that would require it? Extra memory is great, but if
: you're not going to be using it, it's just extra money.
This raises an intteresting question. Obviously the man reason people want
the extra memory is so they have it when they do need it. I know I talked
to a few developers at E3 who were unsatisfied with the amount of space
they had to work with texture maps in regards to the Voodoo boards. I
also hear rumors fo games supporting 16bit textures (QII) which would seem
to require suggest that more memory might become useful in the near
future. Now it is possible that the chipset will be obsolete before the
memory becomes an issue. but all of this remains to be seen. In short
sure we don't need the memory now but then we don't need the riva chipset
now. Most of us are trying to hedge against the future and hopefully
extend the life of our purchase.
--
+-=-=-=-=-=-=-=-=-=-=-=--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
Matt Doughty - Computer support, Armentrout, Roebuck, & Co.
Email: dou...@arches.uga.edu
+-=-=-=-=-=-=-=-=-=-=-=--=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-+
LAW
Hey Does the Velocity128 support 800x600x16bit in 3d, think Falcon4?
LAW
Alf wrote in article <34059d41...@STB.COM>...
>bhay...@gte.net (Robert Hayward) wrote:
>
>>On Fri, 22 Aug 1997 08:39:55 -0500, ASSFUNGUS <ASSF...@prodigy.net>
>>wrote:
>
>>>Velocity 128 is expected to be out around the 2nd week of September.
>
>> I bought my Velocity 128 this morning at CompUSA and just finished
>>installing it. Looks good so far.
>
>Yes, the Velocity 128 has started shipping! Check with your local
>CompUSA and Best Buy to see which ones already have them in stock.
Carl Mueller
"Alex Brin" <ajb...@cs.rit.edu> wrote:
>[discussion about the 4MB local memory limit]
>
>Good points, of course, except that unlike the Voodoo, the Riva 128 does not
>rely on dedicated local texture memory. One of the basic premises behind
>the Riva 128 design (according to nVidia) is to use system memory for much
>greater texture storage, limited only by system memory; it's supposed to
>stream those textures over the bus instead of relying on dedicated local
>texture storage like the Voodoo.
>
>It was specifically designed this way with the AGP spec that calls for
>texture storage in local memory, although it can do this over the PCI bus as
>well.
>
>But the big question is how well would this work? Wouldn't it saturate the
>PCI bus?
Yes, it would. In theory, at least half of the PCI bus bandwidth should
be taken up by the triangle data being sent to the board. Also, there
is that fact that PCI reads are much slower than PCI writes (they are
not pipelined), so having the card read in textures this way while
rendering does not appear to be practical.
>Some speculation: since the Riva 128 has the capability of doing full
>geometry setup on board without the need to rely on the host CPU's FTP for
>triangle setup (even the Voodoo relies on the host FPU partially for
>geometry setup). Since other cards saturate the bus with triangle and
>vertex data, the Riva in theory shouldn't; this should free the bus for
>texture streaming.
>
>Of course, this is speculation on my part. Take it with a grain of salt, or
>a whole shaker of Morton's.
Your speculation is not well stated, though there might be a few
grains of truth to it. Use of the FPU has nothing to do with the
amount of data sent. [And in either case, the FPU is still being
used to do the geometry transformation work, which neither card
does yet.] However, sending the raw triangle data might result in
a slight data reduction over sending the setup triangle data.
In any case, there's still a significant amount of data to send
per triangle, and this is going to require a fair amount of PCI
bandwidth, leaving fairly little for texture paging.
In short, 4 MB on the V128 might be adequate for today's needs
(or perhaps yesterday's), but it's inadequate for future needs.
Compared to the coming competition, I'd count this as a heavy
strike against this card.
-Carl (mue...@cs.unc.edu)
Carl Mueller
"LAW" <L...@sprynet.com> wrote:
> Hey Does the Velocity128 support 800x600x16bit in 3d, think Falcon4?
In theory, it ought to be able to. That resolution would require
about 2.75 MB of memory for a 16 bit front buffer, back buffer,
and Z-buffer. Of course, it only leaves 1.25 MB for textures, so
that might not be enough for the game's needs.
-Carl (mueller@cs!unc!edu)
Jean Salvati
Terry Harris a écrit dans l'article <3406a339...@news.dial.pipex.com>
...
>ACOVEY@*STB.COM (Alf) wrote:
>
>>For good realtime performance at greater than 640x480 resolutions with
>>true color, a wider bandwidth is needed. Basically, this means AGP
>>because PCI is too limited.
>
>Perhaps you could explain that, I really don't see how a 3D
>accelerator needs any additional data from the system to render an
>image at 800x600 than it does to render the same image at 640x480, and
>so it doesn't need any additional bandwidth on the system bus.
That's true. But 800x600 still means more pixels to fill than 640x480. And
I would think that performing texture mapping from main memory could free
up some local graphics memory bandwidth for other tasks.
Jean Salvati
Alf
terry....@dial.pipex.com (Terry Harris) wrote:
>No they just have to charge them for Pentium II's, new motherboards,
>new SDRAM system memory, and new AGP cards - bargain so much cheaper
>than putting an extra 4MB on a video card.
>Cheers Terry...
You might want to double-check my post where I explain how memory is
not the only issue.
Alf
terry....@dial.pipex.com (Terry Harris) wrote:
>ACOVEY@*STB.COM (Alf) wrote:
>
>>For good realtime performance at greater than 640x480 resolutions with
>>true color, a wider bandwidth is needed. Basically, this means AGP
>>because PCI is too limited.
>Perhaps you could explain that, I really don't see how a 3D
>accelerator needs any additional data from the system to render an
>image at 800x600 than it does to render the same image at 640x480, and
>so it doesn't need any additional bandwidth on the system bus. Except
>perhaps if it uses larger more detailed textures but then this whole
>argument is about the need for additional on card texture memory.
Both the AGP and PCI versions of the Velocity 128 can store textures
in system memory. That's why our 4MB Velocity 128 card is faster than
8MB cards in 3D performance.
>I don't know why you don't come clean and say that the Riva's 4MB
>memory limit will be a sever limitation on the PCI version instead of
>spinning some crap about AGP being required for high performance.
As I said before, for this chip, 4MB is not a limitation in that sense
even in the PCI version.
>AGP is not required for performance, memory is required for
>performance. AGP is required for cheapness by using system memory
>instead of on card memory. Not that system memory or system memory
>bandwidth is free (a point AGP proponents conveniently forget).
If it was all memory and no bandwidth, we'd still be using XT bus
video cards with lots of memory and we'd all be happy doing it. This
illustration should also answer the above two questions I skipped
(sorry for not getting to them in order!).
>While AGP might be cheap making do with less on board memory the rest
>of the system required to support AGP is a horrendously expensive
>upgrade.
>Hmmm, I think I will scrap my motherboard, and PPro 200 and 128MB of
>simms and current video card(s) so I can buy a new Pentium II 266 and
>a new motherboard and 128MB of SDRAM to enable me save $50 on an new
>AGP video card because it only needs 4MB of memory - yeah sounds like
>a bargain.
You might want to re-read the rest of my post where I explain how and
why memory is not the only issue here. Once AGP is taken advantage of
by programmers, it won't be, "I think I'll upgrade my entire system to
save on some video RAM," it will be, "I think I'll upgrade my entire
system because that's the way applications are written." (remember
when PCI came out?) In other words: Currently, 4MB should be plenty
for virtually every game.
>Cheers Terry...
Alf
mue...@cs.unc.edu (Carl Mueller) wrote:
[snip]
>In short, 4 MB on the V128 might be adequate for today's needs
>(or perhaps yesterday's), but it's inadequate for future needs.
The same might be said for the bus architecture, processor, system
memory, operating system, current APIs and any other piece of
equipment in the computer.
>Compared to the coming competition, I'd count this as a heavy
>strike against this card.
So the "big strike" against the Velocity 128 is adequacy for future
needs? I have to point out, in that case, that the competition has
just as big a strike against them. In fact, if we can take what we've
seen of the Riva 128 and "upcoming" technology, the strike against the
competition would be even larger. You might want to check out
http://www8.zdnet.com/pcmag/features/300p2/p2test.htm
>-Carl (mue...@cs.unc.edu)
Al
ACOVEY@*STB.COM (Alf) wrote:
> Both the AGP and PCI versions of the Velocity 128 can store
> textures in system memory. That's why our 4MB Velocity 128
> card is faster than 8MB cards in 3D performance.
Faster based on what? A much better goal would be to see how much
faster the Velocity would be with games experiencing a good
degree of texture thrashing (as in GLQuake and it's siblings and
a few flight sim texture hungry titles). Now this would be a
good comparison between the Velocity's 2MB of local texture RAM
and the 3DFX and Rendition boards with 4 or more MB of texture RAM,
rather than using the titles that have very low resolution textures
(e.g. MotoRacer).
> Once AGP is taken advantage of by programmers, it won't be,
> "I think I'll upgrade my entire system to save on some video
> RAM," it will be, "I think I'll upgrade my entire system
> because that's the way applications are written." (remember
> when PCI came out?) In other words: Currently, 4MB should be
> plenty for virtually every game.
AGP is a one slot solution, not comparable to PCI's
multi-purpose/multi-slot usability. That is, the first
statement would indeed be true: "I think I'll upgrade my
entire system to save on some video RAM."
Carl Mueller
ACOVEY@*STB.COM (Alf) wrote:
>mue...@cs.unc.edu (Carl Mueller) wrote:
>[snip]
>>In short, 4 MB on the V128 might be adequate for today's needs
>>(or perhaps yesterday's), but it's inadequate for future needs.
>
>The same might be said for the bus architecture, processor, system
>memory, operating system, current APIs and any other piece of
>equipment in the computer.
No, that analogy doesn't hold. When I say "future needs", I'm
talking about what's right around the corner, not what's coming
several years from now.
>>Compared to the coming competition, I'd count this as a heavy
>>strike against this card.
>
>So the "big strike" against the Velocity 128 is adequacy for future
>needs? I have to point out, in that case, that the competition has
>just as big a strike against them.
I said "inadequate". And I said "coming competition", such as the
Verite 2200 and Permedia 2 cards that have 8MB available (or even
the PowerVR, which has 4MB of dedicated texture memory, or the
announced 8MB Rush cards).
Alf, I think you're wasting your time with all this flag-waving.
If I were you I'd just answer peoples' questions and let the card's
performance speak for itself. I expect the card will perform very
well for certain tasks, so that should be enough for some people.
It's my considered opinion that the 4MB limitation is an undeniable
shortfall for the RIVA chip. It limits BOTH the maximum resolutions
the chip can display and the local texture storage. PCI bus
bandwidth is not adequate to both send polygons AND large amounts of
texture data to the card. A little math will demonstrate this fact.
Finally, if 4MB were adequate, why is it that nVidia is planning to
raise this limit for the next spin of the chip (from reports I've read)?
The RIVA will be fine for the games that are out or will be coming
in the next few months. But the limit reduces cards based on it to
"just a gaming card", where they could have been so much more...
-Carl (mue...@cs.unc.edu)
Alf
bhay...@gte.net (Robert Hayward) wrote:
> I bought my Velocity 128 at CompUSA in Tampa the morning of
>8/26/97. While viewing the readme.txt file on the CD I came across the
>list of known driver problems and a statement that a more recent
>driver available at www.stb.com would fix all of these problems. So
>before I began the installation I went hunting for the newer driver
>and couldn't find any mention of a driver for a Velocity 128 at STB's
>web site. Upon calling STB tech support I was instantly asked where I
>got the Velocity 128. After I gave the store and location to the tech
>I was told the card should not have been released for retail sale
>until 9/1/97 and STB wasn't prepared for tech calls at that time.
There were advanced shipments of the Velocity 128 and STB Tech Support
was not made aware of this until later that week. At that time, we
did not have documentation or driver disks for reference. This was
made available to us before the week was out and we apologize if it
made for any inconvenience for those first couple of days.
> The tech even tossed out a comment about the lack of good
>documentation not being in the retail box as yet. Made me wonder if I
>will be eligible for a proper manual if and when STB produces one.
Because we did not have documentation available, we were unaware at
that time that the manual would be included on the CD. If you have
the on-line documentation, you already have the proper manual. Again,
sorry for the confusion.
>--
>Bob
>bhay...@gte.net
Alf
bhay...@gte.net (Robert Hayward) wrote:
> This situation caused no problems whatsoever. I was merely telling
>"my" story of obtaining a Velocity 128. I like the card, it suits my
>needs perfectly. If it is any consolation, I never even got put on
>hold. The tech answered so quickly I couldn't even think of what I
>wanted to say. :)
Glad to see you're enjoying the card! I couldn't tell from the
original post if the technician was able to help you, so I just wanted
to make sure. I'm glad to see it wasn't an issue! :)
> I'll buy that story. I did know of the CD documentation. It's just
>that it is so hard to lay in bed at night and read the manual with a
>21" monitor on my stomach. :) I was hoping for your usual good printed
>manual.
Yeah, a monitor is a little heavier than a book! Along these lines,
I'm making note of the suggestions for the printed manual as opposed
to an online one.
James Taylor
Hey Alf
enter my vote for printed manuals. Definitely worth an extra few bucks!
james
Ed Brown
> I'll buy that story. I did know of the CD documentation. It's just
>that it is so hard to lay in bed at night and read the manual with a
>21" monitor on my stomach. :) I was hoping for your usual good printed
>manual.
>--
>Bob
>bhay...@gte.net
Right on, Bob. I like an on-line manual as a supplement but not
as the ONLY manual we get. I like the spiral bound docs best
becasuse they lay flat and allow me to use both hands for mouse
and keyboard during tutorials. It's also difficult to take the
on-line manual into the rest room, onto the shuttle bus to and from
work, and I really hate using them on airplanes! ;-)
-=- Ed
--
Anti-Spam Mode On: Please Reply to: ebr...@pacifier.com Thanks.
Alf
James Taylor <jts...@foto.infi.net> wrote:
>
>Hey Alf
>enter my vote for printed manuals. Definitely worth an extra few bucks!
>james
Gotcha, James, Robert and Ed!
Bongo
here here!
enter my vote as online guides as supplements, and not the "only"
version of the manual.
kairyl
Gary Higgins
Alf wrote:
> Alf
> STB Tech Support
> [remove the * from the address to contact me via e-mail]
> ________________
> STB Systems Inc.
> Web site: http://www.stb.com/
> FTP site: ftp.stb.com
> BBS: (972) 437-9615
I am going to buy a new Dell PC with Intel FX400 chipset with AGP. Dell
says that if you want the unit equipped with the STB Velocity AGP card, I
will have to wait until October 15th before they can ship. Isn't this card
already on the market? From what I understand, Gateway and Micron are
already shipping PCs with the Velocity AGP card. Do you have any idea why
Dell would be delayed on this?
--
Gary Higgins
Chicago, IL USA <The Windy City>
ga...@mcs.net
http://www.mcs.net/~garyh
+++ Don't Dream It, Be It +++
Jeff Atwood
Probably a volume and production issue.. IE they need to get so many
thousand of these AGP cards to Dell. Certainly AGP cards won't be found in
stores until christmas at the earliest, nobody has the right motherboard to
buy one, and those buying a new P2 system already have an AGP video card.
Jeff
Gary Higgins wrote in article <341393F6...@mcs.net>...