• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

First shader replacements, now bridge chips?

E

^eMpTy^

2[H]4U
Joined
Jul 21, 2004
Messages
3,233
So suddenly after months of bitching shader replacements suddenly became perfectly OK when ATi implemented them in the latest driver...

Now after months of bitching about nVidia's "inferior bridged design" ATi will be making a bridge chip to turn PCI-E parts into AGP parts and vice versa.

What's next? My guess is PS3 which will suddenly become important right around the time the r520 is released.

Nothing like playing catchup.

http://www.theinquirer.org/?article=18853
 
They don't have a choice, ATi's getting screwed every which way this round ;)
 
I think the three of you are actually the same person...

I mean there really cant be that many people that care so much for so little...


Can there?

As to the topic...of course they'll do whatever they need to remain competitive in the marketplace..

If that means support of PS3.0 of course they'll do it..if it means bridge chips they'll do it...whatever.

In the end you can't ever get that worked up about what PR and Company spin is from any company.. because its just business.. one day it the worst feature to have the next day they will support it.
 
TheGameguru said:
I think the three of you are actually the same person...

I mean there really cant be that many people that care so much for so little...


Can there?

As to the topic...of course they'll do whatever they need to remain competitive in the marketplace..

If that means support of PS3.0 of course they'll do it..if it means bridge chips they'll do it...whatever.

In the end you can't ever get that worked up about what PR and Company spin is from any company.. because its just business.. one day it the worst feature to have the next day they will support it.
Click to expand...



There is no need for a bridge chip by next year, alot more people will be using pci-e motherboards, but that might be one of the reasons they had yeild issues, two different fabs means two different sets if issues, twice the amount of resources, twice the amount cost to mature the fabs.
 
TheGameguru said:
I think the three of you are actually the same person...

I mean there really cant be that many people that care so much for so little...
Click to expand...

I never thought about it that way, but that makes alot of sense.
 
s_s256 said:
I never thought about it that way, but that makes alot of sense.
Click to expand...

On what planet?

Terra - We don't need anymore conspiarcy theories...
 
I think ATi has always stated they would consider making PCI-E to AGP bridge for future designs depending on how quickly AGP dies off. Although honestly, ATi's product lines are diverse enough to not need a bridge IMO.

Since its meant to bridge PCI-E designs into AGP, it could/should be faster than native AGP parts. The videocache and videocore components can all be run at PCI-E speed, then simply bridged down when it has to talk to the motherboard. Unlike an AGP to PCI-E design which would always be limited to AGP speed of the core, and further could be even damaged by the AGP to PCI-E translation causing it to be slower than a native AGP design. Thankfully the Nvidia 6600 is a native PCI-E part, and the only designs crippled by the AGP to PCI-E are the totally unavailable 5700 and 5900 PCI-E variants.

If anyone wants to test the theory, you can try getting a PCI-E 5900 and try some HDTV video decoding/editing on it and see if it struggles more than an AGP 5900. Even though there is plenty of bandwidth, I'd be willing to bet it eats up more CPU or GPU depending on the task at hand. For 3D games it won't make a difference.
 
It doesn't really matter if ATI wants to use bridge chips. Just typical nonsense TheInq blows out of proportion, again. :rolleyes:

The real problem is when ATI lies about the performance of a bridge chip. NV's nv40/nv45 bridge is capable of 1GB/s from the chip (and at least 2GB/s to the chip), nearly 3x faster than needed for processing uncompressed 1920x1080 HDTV streams.
 
so the ati ceo told the inq that they will be using a bridge chip, and now that they heard some whispers from other sources about the bridge they wrote a story about it.

from now on i have to trust the inq a bit more since it looks like they really have some good sources, i mean if they need more sources then the ceo of the company they must really try to validate the accuracy of there storys before they go to print :p
 
^eMpTy^ said:
So suddenly after months of bitching shader replacements suddenly became perfectly OK when ATi implemented them in the latest driver...

Now after months of bitching about nVidia's "inferior bridged design" ATi will be making a bridge chip to turn PCI-E parts into AGP parts and vice versa.

What's next? My guess is PS3 which will suddenly become important right around the time the r520 is released.

Nothing like playing catchup.

http://www.theinquirer.org/?article=18853
Click to expand...

First of all its the Inq, need I say more hehe

As Zen Ops said I think ATI has always left the option open for going from PCI-E to AGP with a bridge which might be some odd % more effecient than a native AGP part due to the core running internally at PCI-E.

It is the going from AGP to PCI-E that becomes the question, is there any latency, is there enough bandwidth, etc... cause internally its an AGP core. And to that note the Inq says this:

We expect an ATI chip to be bidirectional to be able to bridge a native PCIe card to AGP and vice versa, but we don’t have confirmation yet for our speculation.
Click to expand...

The Inq is speculating here, they have no confirmation.
 
Bleh more PR drivel. Honestly, I can just imagine the job application; "lack of dignity and integrity preffered".

Where do ATI and Nvidia find these people?
 
The Batman said:
Bleh more PR drivel. Honestly, I can just imagine the job application; "lack of dignity and integrity preffered".

Where do ATI and Nvidia find these people?
Click to expand...

*looks at empty and ruined*

Who knows :).
 
Brent_Justice said:
First of all its the Inq, need I say more hehe

As Zen Ops said I think ATI has always left the option open for going from PCI-E to AGP with a bridge which might be some odd % more effecient than a native AGP part due to the core running internally at PCI-E.

It is the going from AGP to PCI-E that becomes the question, is there any latency, is there enough bandwidth, etc... cause internally its an AGP core. And to that note the Inq says this:

The Inq is speculating here, they have no confirmation.
Click to expand...

...that's not how I read it...

ATI's CEO told us that it will do it, but now we've also heard similar whisperings from several outside people.
Click to expand...

The way it reads it seems they have confirmation that there will be a bridge chip from ATi's CEO as well as other sources...the only speculation is whether or not it will be bi-directional.

And you and Zen-Ops can talk theory all you want...the fact is that every single benchmark shows the PCI-E version of the NV40 performing EXACTLY the same as its AGP counterpart...doesn't seem to be any latency related issues to me.

And the point here is that ATi has a tendency lately of bashing nvidia for things and then turning around and doing those same things a few months later...don't forget about shader replacements...6 months ago every ATi PR rep was spouting about how thier solution was superior because Nvidia was using shader replacement...and now, shader replacement is a NEW FEATURE with a snazzy name "AI"...

The Inq seems to be onto the story, and god knows ATI could use a damned bridge chip so they could get some AGP x800XT parts out the door at a reasonable price...
 
gordon151 said:
*looks at empty and ruined*

Who knows :).
Click to expand...

yeah...cuz I buy into all the PR these guys generate :rolleyes: ...overstatement of the year right there...
 
The whole industry has to do bridge chips to keep costs down, without them they can't dynamicly allocate chips. For example the highend might be split 50/50 for agp/pci-e, but the lowend might be 90% pci-e for the oems, but when that fluxuates they dont want to go to the fabs and tell them to start producing more of a certain design. That and taping out 1 design is much cheaper.
 
archevilangel said:
The whole industry has to do bridge chips to keep costs down, without them they can't dynamicly allocate chips. For example the highend might be split 50/50 for agp/pci-e, but the lowend might be 90% pci-e for the oems, but when that fluxuates they dont want to go to the fabs and tell them to start producing more of a certain design. That and taping out 1 design is much cheaper.
Click to expand...

exactly...the whole bridge chip thing made perfect sense to me from the get go...it's the most flexible way to handle the PCI-E transition...I just love that ATi bashed the strategy non-stop only to adopt it themselves 6 months later when they realized they made a mistake...
 
hell i'd love for them to start producing bridge chips or connectors, if that would allow me to hook up a 6800 ultra agp to a pci-e like I want. But then agian I have little to no info on what you all are talking about so yah...
 
^eMpTy^ said:
exactly...the whole bridge chip thing made perfect sense to me from the get go...it's the most flexible way to handle the PCI-E transition...I just love that ATi bashed the strategy non-stop only to adopt it themselves 6 months later when they realized they made a mistake...
Click to expand...

The whole fallacy with your statements is that you dont even remember that ATI had issue with bridging agp to pci-e since the latter is a much faster interface than the former. It's different bridging a faster interface into a slower one (pcx->agp). And you question why I made the earlier statement, sheesh.
 
gordon151 said:
The whole fallacy with your statements is that you dont even remember that ATI had issue with bridging agp to pci-e since the latter is a much faster interface than the former. It's different bridging a faster interface into a slower one (pcx->agp). And you question why I made the earlier statement, sheesh.
Click to expand...


And this came out to be false, since the AGP cards nV has that were bridged to PCI-e weren't really slower. Plus the fact since AGP x4 busses aren't even being pressured will PCI-e busses be?
 
gordon151 said:
The whole fallacy with your statements is that you dont even remember that ATI had issue with bridging agp to pci-e since the latter is a much faster interface than the former. It's different bridging a faster interface into a slower one (pcx->agp). And you question why I made the earlier statement, sheesh.
Click to expand...

Had trouble with it?

They said from the beginning that they would have "native PCI-E" support...you're right I don't remember them having trouble bridging...I don't even remember them trying...link me
 
rancor said:
And this came out to be false, since the AGP cards nV has that were bridged to PCI-e weren't really slower. Plus the fact since AGP x4 busses aren't even being pressured will PCI-e busses be?
Click to expand...

What are you talking about. ATI never said anything like that. The one of the main things they emphasized was the bidirectional bandwidth offered by a true pcx interface. Which is why they had a demo running multiple high def video streams to emphasize it.
 
^eMpTy^ said:
exactly...the whole bridge chip thing made perfect sense to me from the get go...it's the most flexible way to handle the PCI-E transition...I just love that ATi bashed the strategy non-stop only to adopt it themselves 6 months later when they realized they made a mistake...
Click to expand...

It's PR, made to sell cards at any cost, EVERYONE does PR... If you're smart enough to read past it then why be bothered by it? Not like their PR department is even remotely in charge of what their development team does...
 
gordon151 said:
What are you talking about. ATI never said anything like that. The one of the main things they emphasized was the bidirectional bandwidth offered by a true pcx interface. Which is why they had a demo running multiple high def video streams to emphasize it.
Click to expand...


Why do you assume that I'm talking about ATi saying something? It is faster but is it going to be used to the fullest extent or is it going to be like agp x8?
 
TheGameguru said:
I think the three of you are actually the same person...

I mean there really cant be that many people that care so much for so little...
Click to expand...

Hm...all three are [H]ardGawds, all three joined in the beginning/middle summer of 2004, all three back each other up when it comes to video card opinions, and they all seem to find their way to the same threads...very suspicious. ;) :cool:
 
gordon151 said:
What are you talking about. ATI never said anything like that. The one of the main things they emphasized was the bidirectional bandwidth offered by a true pcx interface. Which is why they had a demo running multiple high def video streams to emphasize it.
Click to expand...


You are referring to the capabilities of PCI E and they were refering to the bridge chips and their function
 
theoretical effects are different then real world effects. Yes fillrates are effects but it but are we using the fill rates to the max, no.

http://www.beyond3d.com/previews/nvidia/nv45/index.php?p=7

same article but with the games. There really isn't much of a different


Of course, under normal resolutions and 256MB frame buffers, current titles manage to nearly fully exist in the graphics memory, hence the only data being transferred is that of of the game data from the CPU down to the graphics, and this rarely exceeds current AGP bandwidths, hence PCI Express bandwidth's are not going to make any differences for these applications. The question is with future applications that are coded to make use of the full 4GB/s upstream and downstream bandwidth of PEG x16, and whether the HSI solution will be able to offer the full advantages - on the face of the specific testing done here, it would seem not. We'll have to wait to compare the results with NVIDIA's native PCI Express solutions, such as the recently announced GeForce 6600.
Click to expand...
 
rancor said:
theoretical effects are different then real world effects. Yes fillrates are effects but it but are we using the fill rates to the max, no.

http://www.beyond3d.com/previews/nvidia/nv45/index.php?p=7

same article but with the games. There really isn't much of a different
Click to expand...
OIf course there isn't much difference, pleanty of on board ram.
You think I was trying to pass it off as real world, when it cleary is theoretical?
 
Moloch said:
OIf course there isn't much difference, pleanty of on board ram.
You think I was trying to pass it off as real world, when it cleary is theoretical?
Click to expand...


No I don't think you said anything wrong, just clarifying :D
 
rancor said:
No I don't think you said anything wrong, just clarifying :D
Click to expand...
No need to, if the people here are [H] they won't need any clarifcation.
That test shows what would be theoretically possible, perhaps when games start using over 250MB it will matter, but the FPS drop should be massive anyway, that's why I think increasing the bandwidth is stupid, it's still a a far cry from the 28+GB/s the onboard ram does.
 
Moloch said:
No need to, if the people here are [H] they won't need any clarifcation.
That test shows what would be theoretically possible, perhaps when games start using over 250MB it will matter, but the FPS drop should be massive anyway, that's why I think increasing the bandwidth is stupid, it's still a a far cry from the 28+GB/s the onboard ram does.
Click to expand...


Very true, but also the pci e bridge still won't stop a massive frame per sec drop if page flipping occurs. So either way, let it be AGP or PCI-e, game makers will make sure they don't exceed the vram usage.
 
rancor said:
They don't have a choice, ATi's getting screwed every which way this round ;)
Click to expand...

Yup.... sad they must sink to the level of...... ;)

I now have nothing to hang my head high regarding ATI. Consider them the also-ran.
 
best

topic

ever

I'd like to nominate empty and tranzruined to be moderators.

Sticky this thread!!

no name calling.

ATI flip flops so its run by John Kerry right?

and last I'd like to sum up my feelings about this thread in one small mouse click ----> :rolleyes:
 
R1ckCa1n said:
Yup.... sad they must sink to the level of...... ;)

I now have nothing to hang my head high regarding ATI. Consider them the also-ran.
Click to expand...


its part of business sometimes ya win and sometimes ya don't, next round I'm sure ATi will get its butt into gear, nV knows that too, they have weaknesses in thier cards still, but at least now they can actaully stay close to ATi, which is all good, keep it comin more GPU power the better!
 
Mister E said:
best

topic

ever

I'd like to nominate empty and tranzruined to be moderators.

Sticky this thread!!

no name calling.

ATI flip flops so its run by John Kerry right?

and last I'd like to sum up my feelings about this thread in one small mouse click ----> :rolleyes:
Click to expand...
John kerry-"I've always had a clear stance on the war, when's im at a pro war place, I'm pro war, when I'm at a anti war place I'm anti-war"
But ati isn't flip flopping, they're not using a bridge chip for agp-pci, pci-agp!!!!!!!!!!!!
 
jacuzz1 said:
You are referring to the capabilities of PCI E and they were refering to the bridge chips and their function
Click to expand...

I'm referring to the fact that a bridge chip limits those same capabilities of pcx, which is what ati had been jumping on nvidia about when they first came out with the bridged solution for pcx.
 
You must log in or register to reply here.
Back
Top