On Tue, 20 Apr 2004 19:43:01 -0700, Peter Duniho wrote:

"Greg Copeland" wrote in message
news
On Tue, 20 Apr 2004 14:26:13 -0700, Peter Duniho wrote:
If your CPU and memory bandwidth is your bottleneck, a faster video card
will produce NO increase in speed whatsoever.

I have to call you on that one.

Really? I'm still waiting to see the post where you do.


Sadly, you've missed the boat... It already happened.

Remember, newer video cards are able to
offload a lot of work from CPUs these day.

Older cards can too. That's the whole point of a 3D acceleration card. So?

Obviously. They just can't offload the same amount of work. Furthermore,
newer cards are able to offload more kinds of work, especially as it
relates to additional detail. This is especially true for newer games that
take advantage of newer dx features or opengl extension on newer
drivers which push data to newer cards. So yes, it's the whole point of
acceleration. It's just that not all cards accelerate equally; especially
if they are an older card with older drivers running newer 3d software.
Since the title has 2004 in it, I think it's safe to say it qualifies as
newer software. Shesh.

Thankfully we can pretend you never replied here and move on.


If your video card is only a
generation or two back, it *can* make a big difference. Why? Well,
drivers are now able move the data the card and let the card compute
and figure out how to draw things.

Again, whole point, so?

Hmm. I think I made it. I guess it went over your head. Is there
specifically something there that you want to admit you didn't understand.
If so, I'm happy to try to explain a second time. In a nut shell, newer
cards can offload more work, more kinds of work, faster. Since it can
offload additional work that older cards couldn't do, this means that in
some cases, time the old computer spent computing and then shoving to the
card can now be spent shoving to the card to let it compute. Is it really
that hard to understand?

If we try to show how a workload can shift, we imagine a vague concept
such as this:
computer old card
x+y+z a

vs

computer new card
x y+z,a

In the situations where this is able to happen, suddenly, the picture
really turns into:

computer new card
x,x,x y+z,a,y+z,a,y+z,a

Whereby, each letter represents some 3d specific operation.

As you can see, suddenly the old computer is doing more work because the
GPU is doing more of what used to be the CPU's load. Now granted, you are
not going to get miracles, as I originally said. Just the same, it can
OFTEN be enough to put new life into an older computer; especially so for
newer 3d applications. Where, new life can mean more detail, higher
resolutions and same detail & resolutions with higher frame rates...or
some combination. Again, that also depends on the 3d app too.


This means, you now have more CPU available.

I guess I really did already state it. See. I said it again. It's up to
you to comprehend.


Only if the CPU is capable of preparing the data in time, and only if
the pathway from the data to the video card is not already running at
maximum speed. Are you sure you know what the word "bottleneck" means?
You're acting like you don't.

You missed the boat again. Just because you're maxed with an old card
does not mean you will be maxed with a new card. Things have gotten lots
better, even in the last couple of years. Best of all, they've gotten
better in ways other than just faster GPUs.


The net effect is that in some cases, it's like getting a faster
computer. Furthermore, if memory bandwidth is a bottleneck, it might be
because your computer is having to juggle large amounts of textures
bewteen its self and the video card.

So you're talking about more texture memory, not a faster video card.

Again, you've missed the boat. If your app needs to shuffle more textures
that you can fit onboard, this takes up a large amount of CPU, memory and
bus bandwidth. If you offload this to your video card, you suddenly have
a lot more CPU, memory and bus bandwidth to allocate to other takes. Is
this really that hard? More textures and especially higher resolution
textures directly translates to more visible detail. Again, the specifics
are going to depend on the app in question and the pre-existing hardware
in use.

Please go back and read what I wrote. My comment was specifically about
the card's processing speed, not its memory capacity. But even if it
was, a video card that's on par with a system only a year or two old is
not going to be running low on RAM for textures, not yet.

Memory capacity, depending on the application's texture requirements,
can be a significant performance factor. It can and does effect the
entire system's performance.


Again, a newer video card,
may greatly alleviate this. Why? Because a video card that is a year
or two old, especially if it's a commodity board, may only have 16M or
32M on it.

You meant to write ONLY if it's a commodity board. No serious 3D
accelerator card has had only 32MB of video RAM for several years (5+).
And if it's a commodity board, then by definition it's not on par with
the rest of your year or two old system.

That's not true. Commodity boards are often low end and very low end for
any day's standard. I think I clearly made the point. Ignore it if you
want, but it doesn't change facts. Remember, some commodity hardware
which offered minor 3d acceleration used the AGP bus and system memory for
texture cache. Ouch. Since the original post did not state what hardware
he currently has, one has to offer a broader range of options and
associated possibilities. I'm sorry that knowledge and possibilities
scares you. Hopefully it won't scare or confuse Jim.


But boy, your straw man did sure fall over nice for you, I'll give you
that much.

LOL. I think it just went over your head.


Maybe 64M if it's a fairly nice one. These days, you can get a nice
mid-range card which is several generations more advanced, which have
128M, 256M and even 512M on them.

Name one mid-range card with 512MB of video memory.

Feel free to look. It's not hard. Does it really matter at this point?
If you have an older commodity card with even 64M and you step up to 256M,
the point remains. Or are you going to be so foolishly stubborn to ignore
common sense and facts of reality? Are you in such a hurry to ignore the
reality of how computers work?

This means all those textures which
were saturating your memory bandwidth and bus can now be loaded, ONCE,
onto your video card. That also means more main memory may suddenly be
available. If you were paging before and offloading the textures
prevents this, it can make a **HUGE** performance difference (of
course, adding memory would probably be recommended too). Again, this
can result in new life in a slightly older computer.

You sure are making a lot of new assumptions about the computer in
question. No decent game PC built in the last year or two is going to be
running into ANY paging issues playing games. Besides, if you ARE
running into problems like that, no simple video card upgrade is going
to produce any significant improvement in frame rates.

Of course I'm making assumptions. You are too. A broad and generalized
question was asked. Only a fool can assume anything other than broad and
generalized answers. My answers attempted to cover the broadest range of
possibilities and offered facts to support them. Your answers made far
too many specific assumptions. I'm simply correcting/expanding/offering
additional detail.


Because computers, video cards, drivers, and the 3d software which is
running greatly differs, it's impossible to answer in absoluetes what
type of return you'll get by moving up to a new card.

That depends on your absolute. I specifically limited my comment to the
situation where CPU and memory bandwidth are already the bottleneck. You
know, "bottleneck". As in, the place where performance is most limited,
leaving the other components at less than 100% utilization.

And I specifically limited my answers to how a newer card can address your
specifically poor assumptions about modern hardware. Again, I think it
went over your head. Care to point out these imagined absolutes that I
supposedly offered...other than hardware and software facts?


Sure, if you try to broaden your assumptions, you can't make an absolute
statement. But I didn't do that. You did.

Where did I make absolutes? My posting clearly went over your head. I
constantly said things like, "could", "maybe", etc...and stated specifics
conditions when there were exceptions. Get real.


Just the same, If you
are thinking of getting a new system, try a nice card first. You may
find that it gives you the extra life that you was wanting. If it
falls short, then you already have your video card for your new system.
Nothing is lost.

Of course something is lost. If you are considering high-end hardware
(and if you aren't, why are we talking about this at all?), then a new
card is going to put you out somewhere in the $200-400 range. But just
because you can afford a new card, that doesn't mean you can afford a
whole new system. So now you've just wasted $200-400 in sunk capital.
Capital that's useless to you until you've saved up the $1500-2500
you'll need for the current fast hardware.

Oh brother. Back to reality. If you get a mid-range card
($175-$250), nothing is lost. I stated the exceptions. Obviously, he
only has a couple of options if he wants more power. One, get a new
system. Two, get a new system and card. Three, get a new card for his old
system. Four, do nothing and live with the fact that he won't have more
power.

Is comprehension really this problematic for you?


Tell you what. How about you send me four $100 bills. I will keep them
cozy for you, and I'll send them right back to you in six months.
They'll work just as well then as they do today. Nothing is lost.
Right? That's what you said.

I said no such thing. Is comprehension really this problematic for you?

The original question was, "HI pilots, Is it worth getting the expensive
new boards like the Radeon 9800 or gforce FX 5900 ie. is there sufficient
detail in FS 2004 to show off these boards?.......Comments welcome......"

I offered comments and expanded on your narrow and, IMO, overly specific
assumptions. I broadened them by offering additional details to a
generalized question. I offered hardware and software facts to back up my
assertions. No matter what, they are still based on assumptions. It's
just the my assumptions address a broader range of possibilities. Most
people prefer to make decisions based on knowledge rather than narrow
answers which may or may not address their specific situation. While all
of the possibilities I offered, may not address his situation, at least
some should.

Is comprehension really this problematic for you or are you upset that you
clearly don't know as much as you're trying to pretend you do?

If you want to specifically and in technical details explain why I'm
wrong, please feel free to do so. I must warn you that I am a programmer
and have a fair knowledge of what's going on under the covers. Feel free
to fire back if you insist.