nVidia nForce (Crush) chipset and MSI
Yousuf Khan
the name, everything else that we know about the chipset is exactly the
same. There's even an official announcement from MSI about their motherboard
that uses this chipset:
http://www.msi.com.tw/newsrelease/2001/0601_6367.htm
Giles Hjort-Tyson
board that lets you play decent 3d games useing the onboard video chipset
alone.
"Yousuf Khan" <yk...@nospam.home.com.spam> wrote in message
news:6M8S6.118488$eK2.27...@news4.rdc1.on.home.com...
Paul Tiseo
gile...@hotmail.com says...
> Dual-channel DDR. Can't wait to see the benchmarks. This might be the
> board that lets you play decent 3d games useing the onboard video chipset
> alone.
As I posted in another thread this might be your usual marketing
fluffing up of facts. I believe the way it works is that there are two
channels, but only one is for system memory. Thus, while the *total*
bandwidth is greater than current DDR chipsets, when compared to other
system the bandwidth *seen* by the CPU to main memory is still the
same...
Anyone feel free to correct me if I am wrong, or is this MSI board
actually dedicating both channels to system memory? (is that even
possible?)
--
(Any opinions expressed are stricly mine only and not my employer's)
--------------------------------------------------------------------
Paul Tiseo, Intermediate Systems Programmer
Birdsall 3, Mayo Clinic Jacksonville
4500 San Pablo Rd, FL, 32224
tiseo...@mayo.edu -- (904) 953-8254
Paul Tiseo
tiseo...@mayo.edu says...
> Anyone feel free to correct me if I am wrong, or is this MSI board
> actually dedicating both channels to system memory? (is that even
> possible?)
Replying to one's self is a sure sign of madness. :)
Upon re-reading, seems like if you populate the alternate slots of
the 128-mbit interface, you actually do get to use both channels for
system memory.
This will obviously help latency somewhat, making it even better
than the current advantage it holds over RDRAM.
Will it help bandwidth though? 4.2GB/sec over 128-bit interface
going into 64-bit CPU interface? If so, by how much will it help...oh
well, I guess we'll have to be patient...
Rune
wrote:
check out this link
http://www.penstarsys.com/previews/nvidia/nforce/nforce_2.htm
more than 900 mb/s i Stream. That's pretty decent. Though the numbers
was supplied by Nvidia
Rob Barris
<tiseo...@mayo.edu> wrote:
> In article <MPG.158557373...@mayonews.mayo.edu>,
> tiseo...@mayo.edu says...
> > Anyone feel free to correct me if I am wrong, or is this MSI board
> > actually dedicating both channels to system memory? (is that even
> > possible?)
>
> Replying to one's self is a sure sign of madness. :)
>
> Upon re-reading, seems like if you populate the alternate slots of
> the 128-mbit interface, you actually do get to use both channels for
> system memory.
>
> This will obviously help latency somewhat, making it even better
> than the current advantage it holds over RDRAM.
So if RAM on bank A takes, say 50ns to respond to a random fetch, and
RAM on bank B takes the same amount of time, how fast can the
combination of both banks respond to a random fetch?
I see the obvious bandwidth boost but not where the better
critical-word latency comes from. If interleaving is done then the
total time to fetch a cache line might decrease of course, but "time to
first word" is harder to improve.
Rob
Paul Tiseo
rba...@quicksilver.com says...
> RAM on bank B takes the same amount of time, how fast can the
> combination of both banks respond to a random fetch?
>
> I see the obvious bandwidth boost but not where the better
> critical-word latency comes from. If interleaving is done then the
> total time to fetch a cache line might decrease of course, but "time to
> first word" is harder to improve.
http://www.rambusite.com/dramreview/technology/tech_latency.html
The only thing I can point you to that has little variance between
platforms is this old DRAMReview article, now archved at RambusSite. In
it, you will find a graphic that shows overall latency between one- and
two-channel forms of RDRAM. You'll notice that at 0 transactions
pending, the latency of the response is bettered by about 20ns in this
type of test, which is about 20%. Of course, the actual benefit relies
on so many variables that lie somewhere between hardware to software
that a generalization will be useless. However, the article does point
out that increasing channels can have a good effect on latency. I assume
that similar benefits can be realized with DDR.
From my understanding and in simple terms (and I am no expert so
anyone correct me please), critical-word latency is the latency to the
first access in a stream. Now, what multiple channels allow you to do is
handle off a request to one channel and, if you are lucky in that the
next request is in another channel, a proper chipset can hand that off
to the other channel and while it receives from the former. Thus, you
can realize benefits, but it really depends on things like a good
chipset, proper drivers, application code that maximally ends up in
different channels, etc. You won't get 2x, but you'll get something out
of it.
So, in your example, if the time to response is 50ns, the chipset
can either wait, and suffer, or if it has another channel and the next
item it is interested in lies there, it takes those 50ns to issue the
second request, then goes back to receive from the first. The staggering
of requests can overcome inherent limitations in one channel.
NewsBrowser
wrote:
>In article <MPG.158557373...@mayonews.mayo.edu>,
If I understood right, the Nvidia chipset will have the appearance of
reduced latency because it has some sort of hardware pre-fetch built
in.
Here is a quote from the article
http://www.penstarsys.com/previews/nvidia/nforce/nforce_3.htm
"There is no doubt the GeForce 2 core as well as the Twinbank
architecture are quite impressive, the new DASP portion of the core is
another innovative product. DASP stands for Dynamic Adaptive
Speculative Pre-processor, and it is essentially a hardware data
prefetch processor. This was designed around the current Thunderbird
processor to give it hardware prefetch capabilities. It can almost be
seen as an 8 way L3 data cache that the processor can access at high
speeds (instead of going all the way to main memory to get the data).
NVIDIA is quoting a 20% gain in CPU performance due to the DASP."
This is great news for thunderbird owners. You have fairly good
latency from the DDR SDRAM plus the additional performance boost of
hardware pre-fetch when everything jives. This chipset is really
sweet.
I would like to see how the performance will really work with
Palomino now.
-Ralph Watson
My real E-Mail Address is "Ralphwat-dot-home-at-excite dot com"
<searching for a .sig >
Notorious G.I.B.
"Yousuf Khan" <yk...@nospam.home.com.spam> wrote in message
news:6M8S6.118488$eK2.27...@news4.rdc1.on.home.com...
Courageous
>If I understood right, the Nvidia chipset will have the appearance of
>reduced latency because it has some sort of hardware pre-fetch built
>in.
You understood this right, although I think there's a concurrency effect
as well.
C//
Giles Hjort-Tyson
According to some uk website (either the reg or inq) nvidia will have a
geforce 3 based chipset about 18-24 months after the release of the geforce
three.
"Notorious G.I.B." <mo...@nospam.com> wrote in message
news:9fh4v5$7lp0$1...@news.swt.edu...
Yousuf Khan
news:MPG.1585a70ae...@mayonews.mayo.edu...
There's also some word that the nVidia chipset can do internal prefetching:
http://www.wired.com/news/business/0,1367,44256,00.html
That's prefetching ontop of the Athlon/Duron's own software prefecting, and
ontop of the Palo/Morg's own hardware prefetching.
Yousuf Khan
Ray
wrote:
>
>If I understood right, the Nvidia chipset will have the appearance of
>reduced latency because it has some sort of hardware pre-fetch built
>in.
>
>Here is a quote from the article
>
>http://www.penstarsys.com/previews/nvidia/nforce/nforce_3.htm
>
> "There is no doubt the GeForce 2 core as well as the Twinbank
>architecture are quite impressive, the new DASP portion of the core is
>another innovative product. DASP stands for Dynamic Adaptive
>Speculative Pre-processor, and it is essentially a hardware data
>prefetch processor. This was designed around the current Thunderbird
>processor to give it hardware prefetch capabilities. It can almost be
>seen as an 8 way L3 data cache that the processor can access at high
>speeds (instead of going all the way to main memory to get the data).
>NVIDIA is quoting a 20% gain in CPU performance due to the DASP."
Sounds great but won't adding another layer of hardware between the cpu and
memory make the latency worse in some cases? Also where is this prefetched
data being stored, a nice fat cache?, if so how big?
--
Ray
Carlo Razzeto
Or maybe I just didn't hear about it untill late. First I hear of it
couldn't have been more than 3 months ago.
Carlo
"Yousuf Khan" <yk...@nospam.home.com.spam> wrote in message
news:6M8S6.118488$eK2.27...@news4.rdc1.on.home.com...
chrisv
>Wow, it kind of amazes me how fast it seems nVidia developed that chipset...
>Or maybe I just didn't hear about it untill late. First I hear of it
>couldn't have been more than 3 months ago.
It wasn't that fast. It's based on the Xbox design, which they've
been working on for some time.
Paul Tiseo
@san.rr.com says...
Exactly. I the time I wrote that post, I hadn't read the Pentarsys
article which explains there is an additional benefit possibly derived
from chipset prefetch on top of the possibility of using two channels
for DDR instead of one.
Carlo Razzeto
arch. to Athlon arch and everything...
Carlo
"chrisv" <chr...@chrisv.com> wrote in message
news:hamphtguk4h2ann8o...@4ax.com...
Brian Stretch
HyperTransport bus), but Intel did their FUD dance on Bill Gates and
thus nVidia had to go thru the irritation of adapting their work to the
outdated P3. 'Twas a big deal when it happened.
--
Brian Stretch http://www.mindspring.com/~bstretch Cert. Technojunkie
"Wherever there is a jackboot stomping on a human face there will be a
well-heeled Western liberal to explain that the face does, after all,
enjoy free health care and 100 percent literacy."--John Derbyshire, NR
Yousuf Khan
news:gf6T6.32358$DG1.5...@news1.rdc1.mi.home.com...
> yeah, I guess that must have helped out... But still, they moved from PIII
> arch. to Athlon arch and everything...
The Xbox was originally supposed to be on an AMD processor anyways, until
Intel strongarmed Microsoft. It's looking like as if this chipset will even
beat the Xbox chipset to market, so likely Nvidia is having trouble adapting
it to the P3 architecture. The last Xbox demo was apparently run on an Intel
i820 chipset with RDRAM and a GeForce2 card added-in.
If Microsoft hadn't succumbed to politics, it might actually have the Xbox
out in time.
Yousuf Khan
Yousuf Khan
> Wow, it kind of amazes me how fast it seems nVidia developed that
chipset...
> Or maybe I just didn't hear about it untill late. First I hear of it
> couldn't have been more than 3 months ago.
>
> Carlo
Been hearing about for close to six months I'd say, so it would seem that it
was being developed even before that, probably nine months ago. The other
one that we keep hearing about is Micron's chipset, first Samurai, and then
Mamba.
Yousuf Khan
Carlo Razzeto
chipset might have been nice... But I guess at some point they decided they
didn't want to deal with the chipset bussines. I guess with all the
interesting politics going on behind the scenes I don't blame them, though I
don't know if that was the reason...
Carlo
"Yousuf Khan" <yk...@nospam.home.com.spam> wrote in message
news:83fT6.140916$eK2.33...@news4.rdc1.on.home.com...
Yousuf Khan
> Yeah, they sounded really exciting at first... Up to 8MB L3 cache in the
> chipset might have been nice... But I guess at some point they decided
they
> didn't want to deal with the chipset bussines. I guess with all the
> interesting politics going on behind the scenes I don't blame them, though
I
> don't know if that was the reason...
Strangely enough, Micron seems to be getting a boost from dealing with AMD,
even though it's not supplying any chipsets for it. I don't know how valid
the reasons for this story are, but a story is a story:
http://cbs.marketwatch.com/news/story.asp?guid={6BB1C228-3BC1-42C9-9598-370D
BE6B4E74}&siteid=mktw
It seems to be going up simply on the hint that more DDR is going to be sold
because of the dual-processor AMD chipset. This and the nVidia nForce
announcement yesterday. With so many manufacturers lining up behind AMD,
despite the fact that AMD is only selling to a little over 20% of the
market, goes to show how much Intel has cut the rest of them out of their
businesses. Despite the smaller sized marketplace, at least the AMD chipset
business is a marketplace, unlike the Intel chipset business; better to
compete in a small pond than in a pond where you don't stand a chance.
Yousuf Khan
Ed
100-150 watts and it's not anywhere near 1GHz yet.
Anthony Hill
<gile...@hotmail.com> wrote:
>Dual-channel DDR. Can't wait to see the benchmarks. This might be the
>board that lets you play decent 3d games useing the onboard video chipset
>alone.
Early numbers I've seen put it as being around 5% faster then an
AMD760 based board, and occasionally a bit better. The integrated
video should also provide a reasonable gaming solution, and it will
certainly beat the pants off anything Intel or VIA have brought out so
far. Generally speaking, the integrated video should be more or less
on-par with a GeForce2MX.
I must say, I'm really quite impressed with the technology that nVidia
has brought to the table here. For their first chipset this thing
looks REALLY impressive on paper at least, and early reports seem
quite positive too. There is, of course, the very important matter of
stability/reliability to sort out, but even there things seem to be
off to a good start (though we'll definitely have to wait to final
products hit the shelves before we can know for sure).
Another thing that looks pretty decent with this chipset is the
built-in sound. Rumor has it that a bunch of the old Aureal team
ended up going to nVidia after the whole infamous legal battle between
Creative and Aureal (where Aureal "won", but then subsequently went
out of bankrupt due to high legal costs). Generally speaking this
built-in sound looks as impressive as just about any add-in card out
there today. All in all, I have to tip my hat to the nVidia folk, it
looks like they've got a winning design on their hands. If they
implement this well, they could start becoming a noticeable factor in
the motherboard chipset industry in a big hurry.
-----------------------
Tony Hill
hi...@uoguelph.ca
Anthony Hill
wrote:
>In article <rbarris-FAAFB5...@news.newsguy.com>,
>rba...@quicksilver.com says...
>> So if RAM on bank A takes, say 50ns to respond to a random fetch, and
>> RAM on bank B takes the same amount of time, how fast can the
>> combination of both banks respond to a random fetch?
>>
>> I see the obvious bandwidth boost but not where the better
>> critical-word latency comes from. If interleaving is done then the
>> total time to fetch a cache line might decrease of course, but "time to
>> first word" is harder to improve.
>
> http://www.rambusite.com/dramreview/technology/tech_latency.html
>
> The only thing I can point you to that has little variance between
>platforms is this old DRAMReview article, now archved at RambusSite. In
>it, you will find a graphic that shows overall latency between one- and
>two-channel forms of RDRAM. You'll notice that at 0 transactions
>pending, the latency of the response is bettered by about 20ns in this
>type of test, which is about 20%. Of course, the actual benefit relies
>on so many variables that lie somewhere between hardware to software
>that a generalization will be useless. However, the article does point
>out that increasing channels can have a good effect on latency. I assume
>that similar benefits can be realized with DDR.
One downside to this article is that it doesn't really compare 1
channel vs. two channel RDRAM, but rather the i840 with 2 channel
RDRAM vs. the i820 with only a single RDRAM channel. The i840 has
built in hardware prefetching, which is precisely designed to reduce
latency. As a side note, the nVidia nForce chipset also has built in
hardware prefetching.
-----------------------
Tony Hill
hi...@uoguelph.ca
Anthony Hill
Unfortunately this benchmark (Copy64) is only one of the 6 sub-tests
of STREAM, presumably chosen because it shows the largest improvement
as compared to the AMD 760 chipset. In fact, after looking at the
STREAM scores on this page:
http://www.aceshardware.com/Spades/read.php?article_id=25000216
it looks to me like the AMD 760 might simply be rather weak on this
particular sub-bench, turning in scores that are only just a bit above
the VIA KT133A and well behind the VIA KT266 in this test. On the
other hand, a score of 970 in Copy64 isn't too far off the 1163 that a
Pentium4 on an i850 scores, and it's well above anything else we've
seen for the Athlon so far.
-----------------------
Tony Hill
hi...@uoguelph.ca
Anthony Hill
<craz...@hotmail.com> wrote:
>yeah, I guess that must have helped out... But still, they moved from PIII
>arch. to Athlon arch and everything...
Actually, if the rumors are to believed, it's more the other way
around. Apparently AMD had the XBox contract in hand and all hardware
was being designed around the Athlon. Then, at the last minute, Intel
offered Microsoft a really good deal, and literally the weekend before
the official announcement, Microsoft changed switched the specs from
an Athlon to a PIII. Now, I don't know how true these rumors are, but
they were fairly well supported at the time.
So, long story short, nVidia probably already had the groundwork laid
for an Athlon core logic chipset, and I guess they figured that they
might as well continue with it rather then scrap the design. nVidia
has been looking at getting into the motherboard chipset market for a
little while now apparently, and given that Intel won't give them a
license to their bus, designing chipsets for the Athlon was the
obvious choice.
On a side note, it looks like Intel's reluctance to license their bus
to chipset vendors might once again hurt their processor sales. Their
strong opposition to VIA getting a GTL+ (P6) bus license back in the
i820 days is often quoted as being a major reasons as to why the
Athlon managed to get such a strong foothold early on, and now it
looks like Intel's lack of third-party P4 chips from either VIA or
nVidia might hurt their P4 sales. Mind you, it looks like VIA will
once again save Intel's skin by producing P4 chipsets against Intel's
will... but I digress.
-----------------------
Tony Hill
hi...@uoguelph.ca
Paul Tiseo
hi...@uoguelph.ca says...
> channel vs. two channel RDRAM, but rather the i840 with 2 channel
> RDRAM vs. the i820 with only a single RDRAM channel. The i840 has
> built in hardware prefetching, which is precisely designed to reduce
Hum. I thoguht the i840 didn't have prefetching. Oh well, there
goes my only possible insight into the benefits of dual channels on
latency... :)
--
chrisv
>Early numbers I've seen put it as being around 5% faster then an
>AMD760 based board, and occasionally a bit better. The integrated
>video should also provide a reasonable gaming solution, and it will
>certainly beat the pants off anything Intel or VIA have brought out so
>far. Generally speaking, the integrated video should be more or less
>on-par with a GeForce2MX.
In other words, on par with a $50 video card. Today's $50 video card.
By Christmas, virtually worthless for anyone who games, except as
maybe a stop-gap measure until you get a REAL video card in there to
go with your 1.5GHz+ CPU.
>I must say, I'm really quite impressed with the technology that nVidia
>has brought to the table here. For their first chipset this thing
>looks REALLY impressive on paper at least, and early reports seem
>quite positive too. There is, of course, the very important matter of
>stability/reliability to sort out, but even there things seem to be
>off to a good start (though we'll definitely have to wait to final
>products hit the shelves before we can know for sure).
It does appear to be a true advancement. Too bad about the video....
>Another thing that looks pretty decent with this chipset is the
>built-in sound.
This I do like. Sound should be free, and I like that fact it's not
Creative.
chrisv
>On a side note, it looks like Intel's reluctance to license their bus
>to chipset vendors might once again hurt their processor sales. Their
>strong opposition to VIA getting a GTL+ (P6) bus license back in the
>i820 days is often quoted as being a major reasons as to why the
>Athlon managed to get such a strong foothold early on, and now it
>looks like Intel's lack of third-party P4 chips from either VIA or
>nVidia might hurt their P4 sales. Mind you, it looks like VIA will
>once again save Intel's skin by producing P4 chipsets against Intel's
>will... but I digress.
It will be interesting to see Intel's response to this. Maybe they
will license Kryo II technology or something for their chipsets...
Anthony Hill
>Anthony Hill <hi...@uoguelph.ca> wrote:
>
>>Early numbers I've seen put it as being around 5% faster then an
>>AMD760 based board, and occasionally a bit better. The integrated
>>video should also provide a reasonable gaming solution, and it will
>>certainly beat the pants off anything Intel or VIA have brought out so
>>far. Generally speaking, the integrated video should be more or less
>>on-par with a GeForce2MX.
>
>In other words, on par with a $50 video card. Today's $50 video card.
>By Christmas, virtually worthless for anyone who games, except as
>maybe a stop-gap measure until you get a REAL video card in there to
>go with your 1.5GHz+ CPU.
Despite popular belief, expensive/good video cards are definitely the
exception in the computer world, not the rule. The Intel i810 and
i815 chipsets, using their integrated video, are very big sellers in
the world today. VIA and SiS integrated video also sell a lot. For
those that are buying external video, it's primarily ATI Rage128 or
nVidia GeForce2MX cards at best (there are still a number of PCs sold
using the nVidia Vanta/M64 chipset). Sure, the gamers will want
something better, but that's really a VERY small portion of the market
(consider first of all that home use in general is less then 50% of
the market, and most of those home users aren't serious gamers).
-----------------------
Tony Hill
hi...@uoguelph.ca
chrisv
>chrisv <chr...@chrisv.com> opened the coffin lid and croaked...
>>
>>In other words, on par with a $50 video card. Today's $50 video card.
>>By Christmas, virtually worthless for anyone who games, except as
>>maybe a stop-gap measure until you get a REAL video card in there to
>>go with your 1.5GHz+ CPU.
>
>Better than the current card I'm using (TNT2/M64), and plenty good enough
>for my needs for at least a year, maybe 2.
Is that TNT2 well-matched to your CPU, or is it the bottleneck for
games? Remember, this nVidia chipset will typically be matched to
1.5GHz+ CPU's. At that point, the low-end video is not well matched
at all.
>And I'll of course get a MB with an AGP slot, so when I'm ready to upgrade,
>it will be easy.
Courageous
>Is that TNT2 well-matched to your CPU, or is it the bottleneck for
>games? Remember, this nVidia chipset will typically be matched to
>1.5GHz+ CPU's. At that point, the low-end video is not well matched
>at all.
You're speculating on the use of a fairly limited category of games.
For most of the rest of all games, it won't matter one bit. But as to
your specific point, at what point now do games being played on
Geforce2 become CPU-constrained?
C//
Slash
scribbled:
Depends entirely on what setting you run the GF2 at. If you're playing
in 1600x1200@32bpp in a game that makes heavy use of multi-pass
rendering, you could max out the GPU with a Celeron 300. On the flip
side, you'll need upwards of a 1.7GHz P4 to max out the framerate of a
GF2 running 640x480@16bpp doing low detail Quake 3 looping demos.
Higher than that, actually.
-Slash
--
"Ebert Victorious" - The Onion
Yousuf Khan
news:p1jthts33jfv8c9tl...@4ax.com...
> Another thing that looks pretty decent with this chipset is the
> built-in sound. Rumor has it that a bunch of the old Aureal team
> ended up going to nVidia after the whole infamous legal battle between
> Creative and Aureal (where Aureal "won", but then subsequently went
> out of bankrupt due to high legal costs). Generally speaking this
> built-in sound looks as impressive as just about any add-in card out
> there today. All in all, I have to tip my hat to the nVidia folk, it
> looks like they've got a winning design on their hands. If they
> implement this well, they could start becoming a noticeable factor in
> the motherboard chipset industry in a big hurry.
It might be short lived, there's talk about the next generation processors
from AMD having a built-in memory controller, thus taking a lot of
functionality out of the chipset.
But then again, this might be exactly what is needed by the chipset
industry, it will take some of the most complicated functions out of their
chips and put them directly in the processor, so they can concentrate on
getting their chipsets to offer such things as sound and video and ethernet
and other things instead. So the nVidia chipset with all of this
functionality on it might be the step in the right direction for the future.
Yousuf Khan
Carlo Razzeto
twards system on a chip... And they died trying to push it...
Carlo
"Yousuf Khan" <yk...@nospam.home.com.spam> wrote in message
news:BquU6.177182$eK2.40...@news4.rdc1.on.home.com...
Yousuf Khan
> Ironically it's the direction Cyrix was pushing twards... More and more
> twards system on a chip... And they died trying to push it...
Yeah, I did think about that, the Cyrix MediaGX was probably a little too
early and it was the direction they were pushing towards. But I think this
undertaking will require some heft and might from whoever tries it, and some
support from friends. I'd say even if Cyrix couldn't do it, somebody like
VIA or AMD might be able to do it, possibly even Transmeta. It requires some
support from the software community (obviously not Microsoft), and from the
hardware community (obviously not Intel). Both Intel and Microsoft have some
investment in trying to keep things the way they are, but if you can get
some backing from people who together are just as big as these two, you
might be able to pull it off. The virtual gorilla strategy.
Yousuf Khan
Carlo Razzeto
thier product that really killed their push. Other wise they might have
pulled it off. Problem was they were interested in pushing a fairly well
preforming computer that perhaps wasn't too fancy but could get the job done
in a time when people were really starting to want faster computer with all
the gizmoss that make playing games and digital movies possible. Two things
the MediaGX (dispite it's name) wasn't necessarily the best at.
Carlo
"Yousuf Khan" <yk...@nospam.home.com.spam> wrote in message
news:d0EU6.180987$eK2.41...@news4.rdc1.on.home.com...
Yousuf Khan
> As far as Cyrix is concerned, I think it was more the poor preformance of
> thier product that really killed their push. Other wise they might have
> pulled it off. Problem was they were interested in pushing a fairly well
> preforming computer that perhaps wasn't too fancy but could get the job
done
> in a time when people were really starting to want faster computer with
all
> the gizmoss that make playing games and digital movies possible. Two
things
> the MediaGX (dispite it's name) wasn't necessarily the best at.
Well, it just goes to show that timing is everything, in whatever field you
talk about.
Yousuf Khan
Anthony Hill
>Never anonymous Bud <the...@san.rxyzr.com> wrote:
>
>>chrisv <chr...@chrisv.com> opened the coffin lid and croaked...
>>>
>>>In other words, on par with a $50 video card. Today's $50 video card.
>>>By Christmas, virtually worthless for anyone who games, except as
>>>maybe a stop-gap measure until you get a REAL video card in there to
>>>go with your 1.5GHz+ CPU.
>>
>>Better than the current card I'm using (TNT2/M64), and plenty good enough
>>for my needs for at least a year, maybe 2.
>
>Is that TNT2 well-matched to your CPU, or is it the bottleneck for
>games? Remember, this nVidia chipset will typically be matched to
>1.5GHz+ CPU's. At that point, the low-end video is not well matched
>at all.
Err, I don't think that 1.5GHz+ CPUs (which AMD doesn't currently sell
even) are the market that nVidia is really aiming for use of this
system. In fact, in their marketing plan they clearly state combining
the really high-end PCs (which would currently be 1.33GHz and higher
models) with an external video card. For the systems where they're
targeting integrated graphics, the nForce 420 is looking to be paired
with the $900-$1200 PC range (right now that would be 850 or 900MHz
Durons and 1.0-1.2GHz Athlons). The nForce 220 is looking at even
lower-end systems.
Also, while a few gamers might think that EVERYONE plays 3D games ALL
the time and buys hardware JUST for that, please remember that the
vast majority of computer buyers (probably about 80-90%) NEVER play a
single 3D game with their PC, and most others only play them on very
rare occasions and would be quite happy with the performance that a
GeForce2 MX has to offer.
-----------------------
Tony Hill
hi...@uoguelph.ca
Anthony Hill
<craz...@hotmail.com> wrote:
>Ironically it's the direction Cyrix was pushing twards... More and more
>twards system on a chip... And they died trying to push it...
Cyrix's problem wasn't the integrated memory controller, and in fact,
the MediaGX was about their only processor that was making them any
money. The problem they ran into was that they also integrated
piss-poor graphics and sound into their processor, along with what was
only an average CPU core with no L2 cache.
Integrating the memory controller onto the CPU just plain old makes
sense, it's something that people have been looking at for some time
now. Look at the ALpha EV7 for example. This chip is identical to
the Alpha EV6 except that it has an integrated memory controller (that
is, of course, assuming that Compaq can ever get this chip to market).
I expect Intel to start integrating the memory controller onto their
CPU down the road as well. Now, integrating sound and video into the
chip? Well that's a whole other ball-park.
-----------------------
Tony Hill
hi...@uoguelph.ca
Anthony Hill
<yk...@nospam.home.com.spam> wrote:
>"Carlo Razzeto" <craz...@hotmail.com> wrote in message
>news:RuBU6.56713$DG1.9...@news1.rdc1.mi.home.com...
>> Ironically it's the direction Cyrix was pushing twards... More and more
>> twards system on a chip... And they died trying to push it...
>
>Yeah, I did think about that, the Cyrix MediaGX was probably a little too
>early and it was the direction they were pushing towards. But I think this
>undertaking will require some heft and might from whoever tries it, and some
>support from friends. I'd say even if Cyrix couldn't do it, somebody like
>VIA or AMD might be able to do it, possibly even Transmeta.
It's rather ironic that you mention Transmeta, seeing as their current
Crusoe processors already have an integrated memory controller :>
-----------------------
Tony Hill
hi...@uoguelph.ca
Kev Ford
>Also, while a few gamers might think that EVERYONE plays 3D games ALL
>the time and buys hardware JUST for that, please remember that the
>vast majority of computer buyers (probably about 80-90%) NEVER play a
>single 3D game with their PC, and most others only play them on very
>rare occasions and would be quite happy with the performance that a
>GeForce2 MX has to offer.
>
I'd expect a Matrox if I was never going to use 3d.
--
Kev Ford
E-Mail: k...@frod.demon.co.uk
Web: http://frod.demon.co.uk
Mobile: 07740-302995
ICQ: 19382011
chrisv
>On Fri, 08 Jun 2001 13:19:23 GMT, chrisv <chr...@chrisv.com> wrote:
>>Is that TNT2 well-matched to your CPU, or is it the bottleneck for
>>games? Remember, this nVidia chipset will typically be matched to
>>1.5GHz+ CPU's. At that point, the low-end video is not well matched
>>at all.
>
>Err, I don't think that 1.5GHz+ CPUs (which AMD doesn't currently sell
>even) are the market that nVidia is really aiming for use of this
>system. In fact, in their marketing plan they clearly state combining
>the really high-end PCs (which would currently be 1.33GHz and higher
>models) with an external video card.
I think this is precisely the problem. This is a chipset with an
identity problem. It only really makes sense at the low end, but the
low end doesn't make sense. (Clear enough for you? 8)
I say the low end doesn't make sense because "high end" CPU's are now
available for less than $200. I stand by my opinion that MANY of
these chsipsets will be matched with (by then, available and cheap)
1.5GHz+ CPU's.
>Also, while a few gamers might think that EVERYONE plays 3D games ALL
>the time and buys hardware JUST for that, please remember that the
>vast majority of computer buyers (probably about 80-90%) NEVER play a
>single 3D game with their PC, and most others only play them on very
>rare occasions and would be quite happy with the performance that a
>GeForce2 MX has to offer.
I'm well aware of that. But who besides gaming enthusiasts get really
excited about the latest cool hardware like dual 64-bit DDR channels,
Geforce2 graphics, and 5.1 channel surround sound?
Martin Høyer Kristiansen
> I think this is precisely the problem. This is a chipset with an
> identity problem. It only really makes sense at the low end, but the
> low end doesn't make sense. (Clear enough for you? 8)
The nv11 is a low end chipset. I think the nv12 makes alot of sense (ie.
I would buy it) as a high end product. Of course the built in GF2MX will
sit idle as I'd want a REAL graphics adaptor in there. The performance
of a high TB og Palomino with dual channel DDR will KILL the competition
performance-wise (ie. not qualified with cost).
Cheers
Martin
Pete Ulrich
> Err, I don't think that 1.5GHz+ CPUs (which AMD doesn't currently sell
> even) are the market that nVidia is really aiming for use of this
> system. In fact, in their marketing plan they clearly state combining
> the really high-end PCs (which would currently be 1.33GHz and higher
> models) with an external video card. For the systems where they're
> targeting integrated graphics, the nForce 420 is looking to be paired
> with the $900-$1200 PC range (right now that would be 850 or 900MHz
> Durons and 1.0-1.2GHz Athlons). The nForce 220 is looking at even
> lower-end systems.
>
Carlo Razzeto
Carlo
"Pete Ulrich" <pmul...@netnitco.net> wrote in message
news:3b24d256$1...@news.netnitco.net...
Gav
wrote:
>Also, while a few gamers might think that EVERYONE plays 3D games ALL
>the time and buys hardware JUST for that, please remember that the
>vast majority of computer buyers (probably about 80-90%) NEVER play a
>single 3D game with their PC, and most others only play them on very
>rare occasions and would be quite happy with the performance that a
>GeForce2 MX has to offer.
And despite the slagging off that it has taken here, the GF2MX is
no slouch either. It may not be top of the range but it gives decent
performance for games at 1024x768 or below.
Its certainly the best gfx chipset yet built into a motherboard and
frankly I don't think there would be much enthusiasm for an 'all in
one' solution that was overpriced because of the need to build the
best of everything into the chipset. As long as theres an AGP slot to
upgrade the gfx card, this is just about the 'perfect' onboard gfx
solution at this time.
Gav
--
My _real_ email address is:
gavan(hypen)martin(dot)moran(at)ubs(dot)com
All opinions purely my own
Kev Ford
It's OK at 800x600 and below; when the bandwidth doesn't affect. I would say
that 1024x768 sees a marked dropoff. The point is is that NVidia themselves
have already superceded the GeForce 2 Family, and I believe it won't be long
before we see a budget version of the GeForce 3. Coupled with the fact that
the PC market may possibly see a large influx of GPU intensive conversions
from the X-box and you've got a recipe for disappointment. I don't understand
why NVidia would release an already obseleted graphics core with what looks
like an otherwise exciting chipset. GeForce 3 core please NVidia!
Courageous
>from the X-box and you've got a recipe for disappointment. I don't understand
>why NVidia would release an already obseleted graphics core with what looks
>like an otherwise exciting chipset. GeForce 3 core please NVidia!
*shrug*. For the desktop, why not just use all that extra silicon for something
else? For example, persuade AMD to up EV6 bus speed, and put a large
L3 on the chipset, ala "Mamba". Woo hoo. It would be especially nice in an
SMP solution. Can you say 8MB L3? :-)
C//
Anthony Hill
wrote:
>It's OK at 800x600 and below; when the bandwidth doesn't affect. I would say
>that 1024x768 sees a marked dropoff. The point is is that NVidia themselves
>have already superceded the GeForce 2 Family, and I believe it won't be long
>before we see a budget version of the GeForce 3. Coupled with the fact that
>the PC market may possibly see a large influx of GPU intensive conversions
>from the X-box and you've got a recipe for disappointment. I don't understand
>why NVidia would release an already obseleted graphics core with what looks
>like an otherwise exciting chipset. GeForce 3 core please NVidia!
I really don't think you understand the market at all here. Take a
look sometime at what video cards are offered on computers. Do you
see any GeForce3 cards? The top of the line consumer systems from
Compaq, Dell, Gateway, etc. mostly come with GeForce2 MX cards, with
an option to upgrade to a GeForce2 GTS or GeForce3. In other words,
this chipset's integrated graphics is already on-par with the BEST of
what is being offered in default configurations from the OEMs!
Take a look at what video cards are being used out there sometime,
generally speaking they're all crap. You know why? Because no one
can convince consumers to pay for better video, even if it would
benefit them. nVidia would be REALLY REALLY STUPID to put a GeForce3
core in the nForce chipset because it would gain them NOTHING in terms
of sales but would greatly increase their cost. It would also do
nothing for performance, since the GeForce2 MX is bandwidth limited
anyway, and they're already working to get the bandwidth up to the
highest levels they can with the dual-channel DDR SDRAM. Putting a
GeForce3 core in there would mean that the core would just have even
more time to sit around and twiddle it's fingers with the extra few
million transistors wasting power, causing system design to be more
difficult due to the higher power and grounding requirements, extra
cooling, and not to mention the extra cost of this all.
-----------------------
Tony Hill
hi...@uoguelph.ca
Anthony Hill
>Anthony Hill <hi...@uoguelph.ca> wrote:
>
>>On Fri, 08 Jun 2001 13:19:23 GMT, chrisv <chr...@chrisv.com> wrote:
>>>Is that TNT2 well-matched to your CPU, or is it the bottleneck for
>>>games? Remember, this nVidia chipset will typically be matched to
>>>1.5GHz+ CPU's. At that point, the low-end video is not well matched
>>>at all.
>>
>>Err, I don't think that 1.5GHz+ CPUs (which AMD doesn't currently sell
>>even) are the market that nVidia is really aiming for use of this
>>system. In fact, in their marketing plan they clearly state combining
>>the really high-end PCs (which would currently be 1.33GHz and higher
>>models) with an external video card.
>
>I think this is precisely the problem. This is a chipset with an
>identity problem. It only really makes sense at the low end, but the
>low end doesn't make sense. (Clear enough for you? 8)
I strongly disagree. This chipset makes PERFECT sense for the
mid-range to high-end of PCs, which is where the most sales are going.
Yeah, it's a bit too expensive for the low-end, though the nForce 220
is targeted for this market, which should be a sufficiently low-cost
solution for all but the very bottom of the barrel. On the mid-range,
this integrated video is better then anything else (integrated or
otherwise) that is offered in any default configuration out there. On
the high-end, the integrated video is essentially the same as what is
typically offered in default configurations (take a look at what
Compaq and Dell use for video cards on their top-of-the-line consumer
systems... I'll give you a hint, there's an "MX" at the end of the
name). Combine this with an option to upgrade to a higher-end card,
and it makes great sense here as well.
>I say the low end doesn't make sense because "high end" CPU's are now
>available for less than $200. I stand by my opinion that MANY of
>these chsipsets will be matched with (by then, available and cheap)
>1.5GHz+ CPU's.
Ok, let's take me for example. I play the odd 3D game, very rarely.
I want something that will work reasonably well in 3D games (which is
why I ditched my old Matrox card), but I also want something that
works well in 2D stuff and that isn't going to break the bank. What
does a GeForce2 GTS buy ME over a GeForce2 MX? A few more f/s in
Quake III? Whoopie.. for the handful of times I play that game (or
any other 3D game) in a month, I'll get by with "only" 50 or 60f/s
instead of the 80-90f/s.
On the other hand, there are a number of things that I do which could
benefit from a faster processor (and there are even more things that
benefit from a faster HD, but that's another story).
>>Also, while a few gamers might think that EVERYONE plays 3D games ALL
>>the time and buys hardware JUST for that, please remember that the
>>vast majority of computer buyers (probably about 80-90%) NEVER play a
>>single 3D game with their PC, and most others only play them on very
>>rare occasions and would be quite happy with the performance that a
>>GeForce2 MX has to offer.
>
>I'm well aware of that. But who besides gaming enthusiasts get really
>excited about the latest cool hardware like dual 64-bit DDR channels,
>Geforce2 graphics, and 5.1 channel surround sound?
I certainly do! :>
-----------------------
Tony Hill
hi...@uoguelph.ca
Yousuf Khan
there. :-)
Yousuf Khan
"Pete Ulrich" <pmul...@netnitco.net> wrote in message
news:3b24d256$1...@news.netnitco.net...
>
Keith R. Williams
Ford) wrote:
> On Mon, 11 Jun 2001 01:14:35 -0400, Anthony Hill <hi...@uoguelph.ca> wrote:
> >Also, while a few gamers might think that EVERYONE plays 3D games ALL
> >the time and buys hardware JUST for that, please remember that the
> >vast majority of computer buyers (probably about 80-90%) NEVER play a
> >single 3D game with their PC, and most others only play them on very
> >rare occasions and would be quite happy with the performance that a
> >GeForce2 MX has to offer.
> >
>
> I'd expect a Matrox if I was never going to use 3d.
Bingo! For professional graphics, there is no other! I
just wish their dual-PCI was a 6.5" card. ...though I'm Ok
with a single - today.
----
Keith
Pete Ulrich
happened to them?????
"Yousuf Khan" <yk...@nospam.home.com.spam> wrote in message
news:cgdV6.194446$eK2.44...@news4.rdc1.on.home.com...
Yousuf Khan
world, so it's easy to remember their name. :-)
Yousuf Khan
"Pete Ulrich" <pmul...@netnitco.net> wrote in message
news:3b25...@news.netnitco.net...
Kev Ford
Who buys GeForce3's? I wouldn't mind one, but the price is highly
excessive for a single card. I, like many other people, do 'one big
upgrade' every 18 months or so (OK so there are people who buy a little
bit here and there, but they should just learn to manage their money
better :) ).
I won't be buying an nForce, I already have a GF2 GTS, and consider
the integrated graphics a downgrade and not worth the extra
expenditure that the board is going to require because 2MX is
integrated. I do, however, only(!) have an old ASUS A7V (200 FSB)
board that I wouldn't mind replacing. Now if nForce included GF3 and
onboard Ram for the video, so that it could deliver I'd pay an extra
GBP200 to get that board, as I have no problem with integration so
long as the integrated products are current, but the 2MX is _not_.
Why are manufacturers so scared of integrating any form of memory
on motherboards anymore? You're going to need some memory on the board
anyway for graphics, why make it a DDR SDRAM slot, so that the minimum
is going to be 64MB anyway, I can't see the GF2 core making meaningful
use of any more than 128MB anyway
Putting the 2MX core on that board is a clear sign of intent at
creating an integrated gaming machine (it is the largest market for
3D applications by a _long_ way). Nvidia haven't had a problem
targetting the higher end before now, so why are they just utterly
ignoring it now? I'm not saying the current nForce is a bad idea,
but it is a little late and I fear the target market have already
just bought 2MX's instead.
nVidia's strategy is clear though. Get 2MX's on the desktop so people
can see 3D game show they might be, then sit back and wait for the
next gen GF3 and X-box ports to come along, and suddenly your now
switched on to 3D audience wants GeForce 3.
Gav
wrote:
>> Its certainly the best gfx chipset yet built into a motherboard and
>>frankly I don't think there would be much enthusiasm for an 'all in
>>one' solution that was overpriced because of the need to build the
>>best of everything into the chipset. As long as theres an AGP slot to
>>upgrade the gfx card, this is just about the 'perfect' onboard gfx
>>solution at this time.
>>
>
>It's OK at 800x600 and below; when the bandwidth doesn't affect. I would say
>that 1024x768 sees a marked dropoff. The point is is that NVidia themselves
>have already superceded the GeForce 2 Family, and I believe it won't be long
>before we see a budget version of the GeForce 3. Coupled with the fact that
>the PC market may possibly see a large influx of GPU intensive conversions
>from the X-box and you've got a recipe for disappointment. I don't understand
>why NVidia would release an already obseleted graphics core with what looks
>like an otherwise exciting chipset. GeForce 3 core please NVidia!
Unless they could put some sort of budget GF3 in there for the
same price as a GF2MX core they'd be plain nuts to do so. The GF2MX
core is pretty much standard in most entry level machines. Making the
chipset more expensive by adding better gfx would price it out of the
market. Few (of the minority market of )serious gamers will stick
with on-board gfx of any sort as they'll simply buy the latest and
greatest card regardless.
The GF2MX is an extremly sensible choice for onboard gfx.
Pete Ulrich
> nVidia's strategy is clear though. Get 2MX's on the desktop so people
> can see 3D game show they might be, then sit back and wait for the
> next gen GF3 and X-box ports to come along, and suddenly your now
> switched on to 3D audience wants GeForce 3.
>
> --
>
> Kev Ford
>
> E-Mail: k...@frod.demon.co.uk
> Web: http://frod.demon.co.uk
> Mobile: 07740-302995
> ICQ: 19382011
I think the consideration should be, at least in terms of gaming, 'How much
of a cards potential is being utilized by programmers?' And at the moment,
it seems even GeForce 2 is overkill. Seems to me it will be years before
programmers come anywhere close to exploring the full potential and
programmable nature of GeForce 3 and by then there'll be something much
better and probably cheaper. There is no doubt that GeForce 3 is a
revolutionary step up but this is a business and the idea is to sell. And
so many are swept up by the hype.....
John King
"Pete Ulrich" <pmul...@netnitco.net> wrote in message
> it seems even GeForce 2 is overkill. Seems to me it will be years before
> programmers come anywhere close to exploring the full potential and
> programmable nature of GeForce 3 and by then there'll be something much
> better and probably cheaper. There is no doubt that GeForce 3 is a
> revolutionary step up but this is a business and the idea is to sell. And
The important point for programmer's is to utilize *enough* features of the
GF3 so that people *notice*.
We rarely fully utilize things for the reason you mentioned.
Kev Ford
>
>I think the consideration should be, at least in terms of gaming, 'How much
>of a cards potential is being utilized by programmers?' And at the moment,
>it seems even GeForce 2 is overkill. Seems to me it will be years before
>programmers come anywhere close to exploring the full potential and
>programmable nature of GeForce 3 and by then there'll be something much
>better and probably cheaper. There is no doubt that GeForce 3 is a
>revolutionary step up but this is a business and the idea is to sell. And
>so many are swept up by the hype.....
>
>
They're not used because not enough people own one to make the extra
programming worthwhile..... until Doom 3 comes out that is!!!!!!!
Gav
wrote:
[.]
>I won't be buying an nForce, I already have a GF2 GTS, and consider
>the integrated graphics a downgrade and not worth the extra
>expenditure that the board is going to require because 2MX is
>integrated. I do, however, only(!) have an old ASUS A7V (200 FSB)
>board that I wouldn't mind replacing.
So replace it with a good board (without gfx), keep your GF2GTS and
upgade to a GF3 when the card becomes affordable.
> Now if nForce included GF3 and
>onboard Ram for the video, so that it could deliver I'd pay an extra
>GBP200 to get that board, as I have no problem with integration so
>long as the integrated products are current, but the 2MX is _not_.
The mass market wouldn't pay £200 ($300) for a motherboard let
alone an EXTRA £200 over the cost of a regular motherboard.
>Why are manufacturers so scared of integrating any form of memory
>on motherboards anymore?
Because they can be left high and dry by fluctuating RAM prices.
Its one more variable for them to be worried about.
From a consumer point of view its bad since you can't take the
on-mobo memory with you to your next board (its not always possible to
do that anyway, but with onboard memory its just impossible).
> You're going to need some memory on the board
>anyway for graphics, why make it a DDR SDRAM slot, so that the minimum
>is going to be 64MB anyway, I can't see the GF2 core making meaningful
>use of any more than 128MB anyway
Call me old fashioned, but maybe the user has APPLICATIONS that
would benefit from expandable memory? Just a thought.....
>Putting the 2MX core on that board is a clear sign of intent at
>creating an integrated gaming machine (it is the largest market for
>3D applications by a _long_ way).
If you want an integrated gaming machine, buy a console.
> Nvidia haven't had a problem
>targetting the higher end before now, so why are they just utterly
>ignoring it now? I'm not saying the current nForce is a bad idea,
>but it is a little late and I fear the target market have already
>just bought 2MX's instead.
Try to understand - people DON'T buy motherboards for the onboard
gfx capability. If they want good gfx, they buy the appropriate
card. With the 'state of the art' changing every 9 months, you'd be
chucking away an awful lot of money on motherboards by paying extra to
get a state-of-the-art gfx chip on it.
>nVidia's strategy is clear though. Get 2MX's on the desktop so people
>can see 3D game show they might be, then sit back and wait for the
>next gen GF3 and X-box ports to come along, and suddenly your now
>switched on to 3D audience wants GeForce 3.
nVidia simply built an integrated chipset because they thought they
could break into a new market and make money from it - not because
they thought it would spur sales of GF3s.
Anthony Hill
<pmul...@netnitco.net> wrote:
>Yea, I was thinking it might have been those 'WinChip' guys.;-) Whatever
>happened to them?????
VIA Cyrix = Winchip. Cyrix is essentially dead other then the name,
the VIA Cyrix processors are the continuation of the Winchip line.
-----------------------
Tony Hill
hi...@uoguelph.ca
Anthony Hill
gavan-martin.moran@remove_this_to_reply.ubs.com.invalid (Gav) wrote:
>On Mon, 11 Jun 2001 19:54:45 +0100, k...@frod.demon.co.uk (Kev Ford)
>wrote:
>same price as a GF2MX core they'd be plain nuts to do so.
Actually, they'd be pretty nuts to go for a budget GeForce3 core in
this chipset even if it was the same price, simply because it wouldn't
buy any performance. The GeForce2 MX is already memory bandwidth
limited, so getting a more powerful GPU just means that you'll have a
GPU that wastes even more of it's time waiting for data from memory.
> The GF2MX
>core is pretty much standard in most entry level machines.
Actually, take a look at what's in "entry level" machines from the
likes of Compaq, Dell, etc. It ain't a GeForce2 MX. Entry level
machines are typically shipped with Intel i810/i815 integrated
graphics or VIA KM133/PM133 integrated graphics. Some of the lower
cost systems actually advertise that they use an nVidia TNT2 M64
chipset because this is a higher-end solution. GeForce2 MX is pretty
much standard for top-of-the-line consumer systems from the big OEMs.
GeForce2 GTS and GeForce3 cards are pretty much ONLY offered as an
upgrade option.
-----------------------
Tony Hill
hi...@uoguelph.ca
nobody
duron 800 KT7A.
Obviously my next upgrade will have to be significantly better, and the move
to a DDR althon 1.7 or something sounds about right.
I'd love to be able to upgrade both sound and graphics at the same time. I
have an original soundblaster live and an elsa TNT2 M64 AGP.
A geforce 2MX is hugely faster than this.
My monitor refreshes at 80hz, so any frame rate faster is simply wasted, and
i don't believe a framrate above about 45 is necessary. Movies only run at
25 and we get persistance of vision. I do not believe there are any games
that need much more power than a 2MX at the moment or will be for some
years. Anyway this is better than a standard gf 2MX add in card because it
is integrated with the memory and so runs at AGP 6X. So while it may not
come up to gf 3 standards i think it will be close and games are not that
demanding really.
I thought my last soundcard would be futureproof but nothing ever is but
this one will certainly last a good few years. I thought it would be years
before there was anything that could do AC3 realtime.
I think that pcs should be going towards console type systems, as long as we
are not reliant on TV resolution which is pathetic.
I'd actually quite like to see AMD/nVidia capture the whole market but i
think this unlikely.
If more things are integrated they will always be compatible and things like
ACPI power management, hibernation etc will work more of the time. This will
be a big improvement.
I may even move to microATX, there are not so many expansion slots available
but you won't need as many with hard discs this big you only need one and
then just a DVD + CD-R will do.