Attn: Paul

  • Thread starter Thread starter Mary
  • Start date Start date
M

Mary

Hi Paul:

I posted a message for you early tonight. I don't know if you saw it or not
as you didn't comment. I thought you would be interested in the results of
the suggestions you gave me and I was interested in any possible comments
you might have. In case you didn't see my message, here it is again below.

Mary
 
"Mary" said:
Hi Paul:

I posted a message for you early tonight. I don't know if you saw it or not
as you didn't comment. I thought you would be interested in the results of
the suggestions you gave me and I was interested in any possible comments
you might have. In case you didn't see my message, here it is again below.

Mary

(I replied in the original thread. Congratulations on getting it
to work a bit!)

Paul
 
Paul said:
(I replied in the original thread. Congratulations on getting it
to work a bit!)

Well, I never got your reply in the NG, and I was watching for it. The last
message on the "AGP texture acceleration not being available" thread that I
saw at least, was the message with my updates that I wrote to you yesterday,
March 8 at 5.30 pm and which was the one which I asked if you saw when I
wrote the Attn Paul message. Do you think you could possibly repost the
message? I would like to see what you said. I was going to look in Google
groups for it, but for some reason I can't get google for the last day or
so. Google is usually a day or two behind with NG messages anyway.

Mary
 
Mary said:
Well, I never got your reply in the NG, and I was watching for it. The last
message on the "AGP texture acceleration not being available" thread that I
saw at least, was the message with my updates that I wrote to you yesterday,
March 8 at 5.30 pm and which was the one which I asked if you saw when I
wrote the Attn Paul message. Do you think you could possibly repost the
message? I would like to see what you said. I was going to look in Google
groups for it, but for some reason I can't get google for the last day or
so. Google is usually a day or two behind with NG messages anyway.

Mary

From the original thread:

******
I think my response to all of this is, congratulations! AGP texture
acceleration is the most important thing, because many games check for
it, and won't load if it is missing. (For example, now you should
be able to run 3DMark2001SE if you want.) The speed of AGP transfer
is a secondary issue - you'll still get some use out of the card, as
there is diminishing return with higher AGP transfer rate. In fact,
in some cases, AGP 4X on an AGP 2.0 compliant card (i.e. uses all
protocols available to AGP 2.0) can be faster than an AGP 8X
card on an AGP 3.0 compliant card. This is because AGP 3.0 removed
some features that presumably wouldn't have scaled well to 8X
rates. So, raw speed isn't everything.

If SmartGART offers you the 4X rate, but sets it back to 2X when
the Windows desktop reappears after a reboot, that means there
must have been trouble detected when SmartGART did its testing
after the reboot. At least this eliminates the BIOS as the
sole agent of the problem, as I've noticed that SmartGART doesn't
pay attention to the BIOS setting anyway. SmartGART must be
examining the same info that Powerstrip gave you, and picks the
lower of the motherboard/video card status register limits to
figure out what to offer in the control panel. (What I cannot tell
you, is if the ATI Catalyst drivers have any hard-coded motherboard
or video card info - like special case code for certain combinations
of hardware. This should only be necessary for cases where the
driver test at boot time would crash the machine, as if the bus
recovery mechanism works properly, the test should be able to
be carried out for any combination of hardware.)

Perhaps the new BIOS did something to change the device ID of the
AGP interface, and then the next time you used the Via 4in1
installer, a different AGP driver got installed ? Maybe you used
an older 4in1 driver, which had a different AGP driver in it ?
I think the significant difference is the "VIA CPU to AGP" thing.
Somehow you got a different driver, and it is working better.

At this point, if you benchmark it and are happy with the results,
you could keep your system as is. If not, then you could swap
some hardware, on the off chance that things could get better.
I'd suggest contacting tech support at ATI and/or Asus, to find out
if the lack of 4X is a known problem, but the odds of getting
an answer are pretty limited.

Now that the Via end of things is "fixed", maybe running through
the series of ATI Catalyst driver versions again might help.
Only if you have the stomach for more reboots :-)

Paul
******
 
Paul,

Thanks for forwarding your message. My ISP didn't seem to get it or didn't
show it. Makes you wonder how many other messages are missing. Please scroll
down for my updates.

Paul said:
From the original thread:

******
I think my response to all of this is, congratulations! AGP texture
acceleration is the most important thing, because many games check for
it, and won't load if it is missing. (For example, now you should
be able to run 3DMark2001SE if you want.)

I downloaded 3Dmark 2001 today. Its a pretty good program. I tried it at
640X480 -the result was 8516, then 800X600 and it was 6021, then 1024X768
and it was 7241. Is that no good? The games played fine. I play mainly
adventure games like Syberia, Gabriel Knight, etc., so wouldn't be as
demanding as those games in 3dMark, but I understand the games in 2Dmark are
needed for the tests. In 3Dmark my card is identified as 2X AGP..

Yesterday, I went to the computer store, showed the guy printout tests
showing the card only
running at 2X and he said he would exchange the card for the same one as he
thought it was the video card. I installed it, but its exactly the same as
the other one -2X only. I said to the guy, what if its the motherboard. He
said "I don't think it would be that. We never have problems with Asus
boards ". He said maybe the card only runs at 2X but I said it should work
at 4X, then he mentions about the BIOS has to be set to this and that. I say
yes, I know about that. I've done all my own upgrading including
motherboards for the last 10 years, and he's probably used to a lot of
people who know nothing about computers. I've never had a Radeon card
though. The board and video card have a warranty with the store for one
year, but the guy said they don't test hardware in the store. they send it
to their warehouse and they do testing there. so who knows how long they
take. They only do repairs in the store if you pay. They are all very nice
when you are buying stuff, and yes the hardware has one year warranty, but
its a different story if you need free service.

The guy also said I notice you use Win 98. Your problem might be helped by
installing Win XP. First of all, I don't believe that, and secondly I don't
have the money for Win XP and I don't want Win XP. I like win 98SE just fine
and can play older windows and DOS games I want to replay now and then. So
either I have to take out the motherboard and take it to them to test, and
have no computer again, or just leave things as they are and forget it. I
don't know if its worth it to pursue it any further, but on the other hand,
if its the board and I want to update my video card later, a new video card
may only work at 2X.

I downloaded that CPU-Z program tonight that someone mentioned in this NG
(thanks to them) and there is a tab for the motherboard and it identifies
the AGP transfer rate as 2X. Doesn't that mean for sure, the motherboard
will only do 2X, no matter which card I would put in? or does it not mean
that necessarily?

The speed of AGP transfer
is a secondary issue - you'll still get some use out of the card, as
there is diminishing return with higher AGP transfer rate. In fact,
in some cases, AGP 4X on an AGP 2.0 compliant card (i.e. uses all
protocols available to AGP 2.0) can be faster than an AGP 8X
card on an AGP 3.0 compliant card. This is because AGP 3.0 removed
some features that presumably wouldn't have scaled well to 8X
rates. So, raw speed isn't everything.

But in your example above though, you are talking about differences between
a 4X card and an 8X card, not a 2X like I have which is even slower than 4X.
Do you think there is much difference between a 2X and a 4X card?

So are you saying that AGP transfer acceleration is more important
(especially when playing games) than whether a card is 2X or 4X. Would a
video card with 2X and 128MB be faster than a 4X with 64 MB?
If SmartGART offers you the 4X rate, but sets it back to 2X when
the Windows desktop reappears after a reboot, that means there
must have been trouble detected when SmartGART did its testing
after the reboot. At least this eliminates the BIOS as the
sole agent of the problem, as I've noticed that SmartGART doesn't
pay attention to the BIOS setting anyway.

How does it eliminate the BIOS as a sole agent of the problem? you mean
because Smart gart ignoes the BIOS as you have mentioned? It looks like
Smartgart controls the safe speed of the video card regardless of what the
BIOS is set at. When I first downloaded Catalyst drivers, Smartgart slider
only went up to 2X, but when I installed different Catalyst drivers, the
slider went up to 4X and I thought great! the card can be set at 4X, but
that didn't work out because Smart puts the slider back to 2X. You can see
the screen blinking at bootup where the AGP rate is being changed by
Smartgart.

SmartGART must be
examining the same info that Powerstrip gave you, and picks the
lower of the motherboard/video card status register limits to
figure out what to offer in the control panel.

Thats what it looks like. Powerstrip gives you good information.

(What I cannot tell
you, is if the ATI Catalyst drivers have any hard-coded motherboard
or video card info - like special case code for certain combinations
of hardware. This should only be necessary for cases where the
driver test at boot time would crash the machine, as if the bus
recovery mechanism works properly, the test should be able to
be carried out for any combination of hardware.)

I don't know that either.
Perhaps the new BIOS did something to change the device ID of the
AGP interface, and then the next time you used the Via 4in1
installer, a different AGP driver got installed ? Maybe you used
an older 4in1 driver, which had a different AGP driver in it ?
I think the significant difference is the "VIA CPU to AGP" thing.
Somehow you got a different driver, and it is working better.

It might be working better. I don't really know, but it has not change the
AGP 2X problem.
At this point, if you benchmark it and are happy with the results,
you could keep your system as is. If not, then you could swap
some hardware, on the off chance that things could get better.

What hardware could I swap? I already swapped the video card. I would have
to take the motherboard in and leave it with them for testing. How do I know
they will give me the correct diagnosis? This is the first time dealing with
this store, though they have 3 other stores in Toronto.
I am wondering if I will be any better off in the end. Its a lot of
inconvenience and I've spent hours
and days on this thing already.
I'd suggest contacting tech support at ATI and/or Asus, to find out
if the lack of 4X is a known problem, but the odds of getting
an answer are pretty limited.

I could ask both of them but I would think that if the 9000 pro and Asus
A7V8X-X motherboard combination made the card run at 2X that it would have
been widely known long ago. The Radeon 9000 Pro is an oldish card now.
Now that the Via end of things is "fixed", maybe running through
the series of ATI Catalyst driver versions again might help.
Only if you have the stomach for more reboots :-)

Well, the AGP texture acceleration is working now. And my computer seems
stable enough so the Catalyst drivers are pk, but I doubt that changing the
Catalyst drivers are going to allow the video card to go at 2X. Or is it
possible different Catalyst drivers could change the card to 4X?
From what I have told you in this message, in your opinion, do you think the
motherboard is the problem, or do you think that the motherboard could be
ok, but does not like the Radeon 9000 pro and won't let it run any faster
than 2X.?

Mary
 
"Mary" said:
Paul,

Thanks for forwarding your message. My ISP didn't seem to get it or didn't
show it. Makes you wonder how many other messages are missing. Please scroll
down for my updates.

Paul said:
I downloaded 3Dmark 2001 today. Its a pretty good program. I tried it at
640X480 -the result was 8516, then 800X600 and it was 6021, then 1024X768
and it was 7241. Is that no good? The games played fine. I play mainly
adventure games like Syberia, Gabriel Knight, etc., so wouldn't be as
demanding as those games in 3dMark, but I understand the games in 2Dmark are
needed for the tests. In 3Dmark my card is identified as 2X AGP..

I happened to run 3DMark today on my P4B 1.8GHz with a Radeon 8500, and
it gave about 7200 at 1024x768. I think the 9000pro, in terms of hardware
capability, is somewhere in the same performance ballpark, so your card
seems to be working.

You may want to leave the 3DMark demo looping overnight. If the
computer is frozen the next morning, you are still having AGP
problems - considering your symptoms, I would tear out the mobo
and take it back to the store if that happens at AGP 2X.
Yesterday, I went to the computer store, showed the guy printout tests
showing the card only
running at 2X and he said he would exchange the card for the same one as he
thought it was the video card. I installed it, but its exactly the same as
the other one -2X only. I said to the guy, what if its the motherboard. He
said "I don't think it would be that. We never have problems with Asus
boards ". He said maybe the card only runs at 2X but I said it should work
at 4X, then he mentions about the BIOS has to be set to this and that. I say
yes, I know about that. I've done all my own upgrading including
motherboards for the last 10 years, and he's probably used to a lot of
people who know nothing about computers. I've never had a Radeon card
though. The board and video card have a warranty with the store for one
year, but the guy said they don't test hardware in the store. they send it
to their warehouse and they do testing there. so who knows how long they
take. They only do repairs in the store if you pay. They are all very nice
when you are buying stuff, and yes the hardware has one year warranty, but
its a different story if you need free service.

I don't see a particular reason for this to be an OS limit. This
sounds like a hardware/BIOS type issue.
The guy also said I notice you use Win 98. Your problem might be helped by
installing Win XP. First of all, I don't believe that, and secondly I don't
have the money for Win XP and I don't want Win XP. I like win 98SE just fine
and can play older windows and DOS games I want to replay now and then. So
either I have to take out the motherboard and take it to them to test, and
have no computer again, or just leave things as they are and forget it. I
don't know if its worth it to pursue it any further, but on the other hand,
if its the board and I want to update my video card later, a new video card
may only work at 2X.

I downloaded that CPU-Z program tonight that someone mentioned in this NG
(thanks to them) and there is a tab for the motherboard and it identifies
the AGP transfer rate as 2X. Doesn't that mean for sure, the motherboard
will only do 2X, no matter which card I would put in? or does it not mean
that necessarily?

The fact that SmartGART offered you 4X via slider, means that
SmartGART thinks the lowest common denominator of the motherboard
and video hardware is 4X now. Thanks to the different VIA driver
you managed to load. If you set the slider, and after a reboot
the slider ends up at 2X, that means the AGP channel faulted during
the AGP 4X test. Now that you've tried another video card, it is
time to swap the motherboard, or try something else...

I notice in the A7V8X-x manual, there is an AGP drive strength
setting in the BIOS. You could try setting it to manual, and then
adjust the P and N channel drive strengths. What this does, is
effectively changes the output impedance of the motherboard AGP
bus drivers. The manual says the defaults are "E" and "F", and
you might try "F" and "F" for example. (A similar hack, is to
adjust the AGP I/O voltage - some Asus motherboards allow changing
the 1.5V to 1.6 or 1.7, and sometimes this helps I/O speed as well.
Your particular motherboard model doesn't have this option.)

You could, for example, boot into Windows, set the SmartGART slider
to 4X, reboot, enter BIOS, change AGP drive strength by no more than
1 from the previous setting, save, exit, boot, and see whether 4X
"takes" or not.
The speed of AGP transfer

But in your example above though, you are talking about differences between
a 4X card and an 8X card, not a 2X like I have which is even slower than 4X.
Do you think there is much difference between a 2X and a 4X card?

So are you saying that AGP transfer acceleration is more important
(especially when playing games) than whether a card is 2X or 4X. Would a
video card with 2X and 128MB be faster than a 4X with 64 MB?

The transition from 2X to 4X will offer a little more performance,
but I don't have any numbers for you. I was trying to make the point
that it is a diminishing return.

For example, there is currently an article on tomshardware.com about
what to expect from PCI Express video cards. They tested existing
cards at 4X and 8X AGP, and there was virtually no difference between
them. There will be a small difference between 2X and 4X, but is it
enough difference to tear out your motherboard and replace it ?
Video card performance is limited more by processor speed than
anything else.

What you want are stability (pass the overnight AGP test), and
to have as many features enabled in the AGP driver as possible
(texture acceleration, without which some games won't load).
4X is gravy if you can get it.
How does it eliminate the BIOS as a sole agent of the problem? you mean
because Smart gart ignoes the BIOS as you have mentioned? It looks like
Smartgart controls the safe speed of the video card regardless of what the
BIOS is set at. When I first downloaded Catalyst drivers, Smartgart slider
only went up to 2X, but when I installed different Catalyst drivers, the
slider went up to 4X and I thought great! the card can be set at 4X, but
that didn't work out because Smart puts the slider back to 2X. You can see
the screen blinking at bootup where the AGP rate is being changed by
Smartgart.

SmartGART must be

Thats what it looks like. Powerstrip gives you good information.

(What I cannot tell

I don't know that either.


It might be working better. I don't really know, but it has not change the
AGP 2X problem.


What hardware could I swap? I already swapped the video card. I would have
to take the motherboard in and leave it with them for testing. How do I know
they will give me the correct diagnosis? This is the first time dealing with
this store, though they have 3 other stores in Toronto.
I am wondering if I will be any better off in the end. Its a lot of
inconvenience and I've spent hours
and days on this thing already.

(This is the time you want a store with a tech and workbench on site.
That way, you can talk to the tech directly, so all the testing you've
done won't get lost in the translation. If the store regularly builds
systems with your exact hardware configuration, they may know a great
deal about the "kinks" in the system. If, on the other hand, they
don't know what they are selling, you might only end up with more
unnecessary grief. Ask at the store, if you can phone the people
at the warehouse directly - maybe based on the kind of response you
get from them, you'll be able to tell what kind of techs they are.
If they insist you leave the whole system with them, make sure you
have a backup of any files on the hard drive - many computer stores
are known for reformatting systems without your permission.)
I could ask both of them but I would think that if the 9000 pro and Asus
A7V8X-X motherboard combination made the card run at 2X that it would have
been widely known long ago. The Radeon 9000 Pro is an oldish card now.


Well, the AGP texture acceleration is working now. And my computer seems
stable enough so the Catalyst drivers are pk, but I doubt that changing the
Catalyst drivers are going to allow the video card to go at 2X. Or is it
possible different Catalyst drivers could change the card to 4X?
From what I have told you in this message, in your opinion, do you think the
motherboard is the problem, or do you think that the motherboard could be
ok, but does not like the Radeon 9000 pro and won't let it run any faster
than 2X.?

Mary

All I can tell you about Catalyst, is one version used to crash on
me when the Windows desktop appeared. The next version of Catalyst
didn't do that. So, there are differences between versions. ATI has
had enough time, though, to work the bugs out of the SmartGART code,
which was being introduced around the time I was installing. Also,
for older cards, the code base should be stable - the latest drivers
normally would be concentrating on improving performance for the
latest cards, so if someone told you today that there was a new
version for download on the ATI site, I wouldn't expect any bug fixes
for your hardware combo.

HTH,
Paul
 
To show you how little you are missing, I reran 3DMark2001SE on
my P4B 1.8GHz Radeon 8500 at 1024x768, and got:

AGP 2X - 7271 pts
AGP 4X - 7255 pts

Now, I take it at face value that the hardware was actually at
the stated values. When I look in my BIOS, for some reason I
am only offered 1X and 4X AGP rates. Yet in SmartGART, I can choose
1X,2X,4X and after a reboot, both 2X and 4X were successfully
set according to SmartGART. Also, the AGP setting was verified
within 3DMark's info window as being set.

(Looking at the P4B user manual, it almost looks like the 1X
choice is there to cause PCI only transfers, while 4X is for
AGP burst transfers, so I guess 4X implies 4X/2X/1X AGP if
required.)

I don't know what to make of this, other than that there isn't
any texture transfer over the AGP bus, during the frame rate
tests. Note that in a lot of games, the "loading" interval
before a level starts in a game, is used to precompute and load
certain textures into the video card memory, so this can reduce
the bus mastering traffic required by the video card while the
game level is running.

Also, my motherboard uses SDRAM and not DDR, so if the video card
ever had to get a texture from main memory, the bandwidth would
be limited by the memory controller and not the AGP bus. On
a lot of the dual channel boards out now, the memory controller
is no longer the limit.

HTH,
Paul
 
Hi Paul: More updates.
I happened to run 3DMark today on my P4B 1.8GHz with a Radeon 8500, and
it gave about 7200 at 1024x768. I think the 9000pro, in terms of hardware
capability, is somewhere in the same performance ballpark, so your card
seems to be working.

I read in a few places on the net that ATI 8500 is a very good card (in the
lower end video card arena) and does better in tests than 9000 Pro, though
not that much faster and that was why I decided when I saw the 9000 Pro to
get it. I also read that 9000 Pro was built on the same design as the 8500.
You may want to leave the 3DMark demo looping overnight. If the
computer is frozen the next morning, you are still having AGP
problems - considering your symptoms, I would tear out the mobo
and take it back to the store if that happens at AGP 2X.

I did as you suggested above and left my computer on all night, and all
tests were still going on when I got up the next day. I did the same tests
for 6 hours that afternoon as well at the default which seems to be 1024X769
and the tests ran fine. By the way, do you know what game the Nature test
come from? The graphics and scenery were quite beautiful. Anyway, all tests
at low and high detail looping seemed to run fine. No freezes after 8 hours
constant running.
I don't see a particular reason for this to be an OS limit. This
sounds like a hardware/BIOS type issue.

In my view, me not using Win XP shouldn't affect the video card speed.
The fact that SmartGART offered you 4X via slider, means that
SmartGART thinks the lowest common denominator of the motherboard
and video hardware is 4X now. Thanks to the different VIA driver
you managed to load. If you set the slider, and after a reboot
the slider ends up at 2X, that means the AGP channel faulted during
the AGP 4X test. Now that you've tried another video card, it is
time to swap the motherboard, or try something else...

I am using Catalyst drivers version 3.8 so whether it would make any
difference if I installed version 4.2, I don't know. I think I installed it
before, when I was experimenting, but I don't think it made any difference
in my card speed. How can I know which version of 4 in 1 VIA drivers I am
using?
I notice in the A7V8X-x manual, there is an AGP drive strength
setting in the BIOS. You could try setting it to manual, and then
adjust the P and N channel drive strengths. What this does, is
effectively changes the output impedance of the motherboard AGP
bus drivers. The manual says the defaults are "E" and "F", and
you might try "F" and "F" for example.

I read the entire manual before and checked it again just now and I know
almost all settings in the BIOS and the peculiar thing is, it DOES mention
in the manual about AGP drive strength settings in the BIOS, but its not in
the BIOS on my board, and the board is identified as A7V8X-X on bootup and I
updated the BIOS so it shows BIOS version 1007. There was never any setting
for Drive Strength the BIOS. There are only three lines in the BIOS
regarding AGP settings and they are "AGP capability" -1X/2X4X/8X" "AGP
Performance Control (enabled or disabled") - I've tried both before. "AGP
Fast Write" - enabled/disabled - I've tried both. Smartgart disables Fast
write. AGP Drive Strength is nowhere in the BIOS. Maybe the manual was
printed before the BIOS was made for the board.

(A similar hack, is to
adjust the AGP I/O voltage - some Asus motherboards allow changing
the 1.5V to 1.6 or 1.7, and sometimes this helps I/O speed as well.
Your particular motherboard model doesn't have this option.)

No, it doesn't say AGP I/O voltage. Just CPU core voltages can be changed.
You could, for example, boot into Windows, set the SmartGART slider
to 4X, reboot, enter BIOS, change AGP drive strength by no more than
1 from the previous setting, save, exit, boot, and see whether 4X
"takes" or not.

That would be a good idea if there was any setting for AGP drive strength,
but there isn't.
The transition from 2X to 4X will offer a little more performance,
but I don't have any numbers for you. I was trying to make the point
that it is a diminishing return.

At least I got AGP texture acceleration to work in DX 9b, and from what I
hear and read, AGP texture acceleration is very important to have,
especially in games, even though I don't play demanding games, some
adventure games are more demanding than they used to be.
For example, there is currently an article on tomshardware.com about
what to expect from PCI Express video cards. They tested existing
cards at 4X and 8X AGP, and there was virtually no difference between
them. There will be a small difference between 2X and 4X, but is it
enough difference to tear out your motherboard and replace it ?

My feeling exactly! As long as 4X is not needed for some games now or in the
future.
Video card performance is limited more by processor speed than
anything else.

Is it? does a 9000 Pro video card with 128MB compensate for it being only
2X?
How important is the memory on a video card?
What you want are stability (pass the overnight AGP test), and
to have as many features enabled in the AGP driver as possible
(texture acceleration, without which some games won't load).
4X is gravy if you can get it.

Do you call it an AGP test? The tests ran fine all night were still running
the next morning.
And according to DX 9b, texture acceleration is enabled, and in the BIOS,
AGP Performance Control is enabled but AGP fast write is disabled, because
Smartgart disables it. I would have to delete Smartgart and enable it in the
BIOS, because as you say, Smartgart does its own thing for video controls
and speeds.
(This is the time you want a store with a tech and workbench on site.
That way, you can talk to the tech directly, so all the testing you've
done won't get lost in the translation.

Thats the best way, but not many stores do testing onsite. At least the one
I got my motherboard from, doesn't seem to do that. I wanted the guy to test
my card on a motherboard at the store to see what speed it showed as, but he
said they are too busy - not too busy to sell of course. He was willing to
exchange the video card and maybe would do that with the motherboard too,
but I am not sure I want to bother, because the guy says they don't test it
there and I would have to do without a computer. I didn't ask if they would
just give me another motherboard and even if they did, what if by chance it
had some other problem. It would be different if the motherboard and video
card didn't work at all and it was obvious one or the other were defective.
Its easier in that circumstance to get them to give you another board.

If the store regularly builds
systems with your exact hardware configuration, they may know a great
deal about the "kinks" in the system. If, on the other hand, they
don't know what they are selling, you might only end up with more
unnecessary grief.

Most of the time, the computer stores here (especially in the area I am in)
are all Chinese owned and operated and they sell items at very good prices.
but they sell on the hope that they can sell enough items to get some profit
and that they won't have much trouble after they sell.
I guess they all build complete systems but probably not in the store. I
think they might do some testing in the store, but only if they have to.

Ask at the store, if you can phone the people
at the warehouse directly - maybe based on the kind of response you
get from them, you'll be able to tell what kind of techs they are.

That would most likely be a problem even if they would allow it, as there
would likely be a language problem. This happened even in the store I bought
my stuff at. But actually, the guy with the poorest English was quite
willing to help, but the other guy who was there the next time, I went, was
not very helpful and with my luck, I would get him again. There are only two
of them there. They have 3 stores in the city, but thats the only one near
me.
If they insist you leave the whole system with them, make sure you
have a backup of any files on the hard drive - many computer stores
are known for reformatting systems without your permission.)

I doubt I will be going back unless my video card or motherboard completely
conk out.
All I can tell you about Catalyst, is one version used to crash on
me when the Windows desktop appeared. The next version of Catalyst
didn't do that. So, there are differences between versions. ATI has
had enough time, though, to work the bugs out of the SmartGART code,
which was being introduced around the time

I wonder if I could force 4X on my video card if I uninstalled Smart guard.
I would have to try it in Power strip if I did that, and I doubt Powerstrip
would allow it.

I was installing. Also,
for older cards, the code base should be stable - the latest drivers
normally would be concentrating on improving performance for the
latest cards, so if someone told you today that there was a new
version for download on the ATI site, I wouldn't expect any bug fixes
for your hardware combo.

I could try Catalyst 4.2 but I doubt I would be any better off video speed
wise. I wonder if the VIA drivers would do anything. Or the Asus motherboard
drivers.

Mary
 
Paul said:
(Paul) wrote:

To show you how little you are missing, I reran 3DMark2001SE on
my P4B 1.8GHz Radeon 8500 at 1024x768, and got:

AGP 2X - 7271 pts
AGP 4X - 7255 pts

Its about the same as what I got at AGP 2X. The first time I got 7241,but I
tried it the next day and it was higher - something like 7278. So very
similar to your 8500. How much memory do you have on your 8500?
Now, I take it at face value that the hardware was actually at
the stated values. When I look in my BIOS, for some reason I
am only offered 1X and 4X AGP rates. Yet in SmartGART, I can choose
1X,2X,4X and after a reboot, both 2X and 4X were successfully
set according to SmartGART. Also, the AGP setting was verified
within 3DMark's info window as being set.

So your 2X and 4X results above should be fairly accurate.?
(Looking at the P4B user manual, it almost looks like the 1X
choice is there to cause PCI only transfers, while 4X is for
AGP burst transfers, so I guess 4X implies 4X/2X/1X AGP if
required.)

I didn't realize that PCI have set transfer speeds. I thought settings in
the BIOS for AGP are only for AGP.
I don't know what to make of this, other than that there isn't
any texture transfer over the AGP bus, during the frame rate
tests. Note that in a lot of games, the "loading" interval
before a level starts in a game, is used to precompute and load
certain textures into the video card memory, so this can reduce
the bus mastering traffic required by the video card while the
game level is running.

Also, my motherboard uses SDRAM and not DDR, so if the video card
ever had to get a texture from main memory, the bandwidth would
be limited by the memory controller and not the AGP bus. On
a lot of the dual channel boards out now, the memory controller
is no longer the limit.

SDram is a little slower than DDR, so if you had a DDR mothrboard, wouldn't
your 8500 card be faster on 3DMark tests?.

Thanks.

Mary
 
Back
Top