Thats not really a good basis for operating though. IIRC this is exactly what Via did with AGP4X, only for the final specification to have changed and Via's initial AGP4X board causing grief for everyone.
More ATI Driver News from Derek Smart
- Thread starter antlers
- Start date
- Status
Thats not really a good basis for operating though. IIRC this is exactly what Via did with AGP4X, only for the final specification to have changed and Via's initial AGP4X board causing grief for everyone.
Joe DeFuria
Legend
Actually, it's not clear to me that ATI is recalling boards. There's a bit of contradiction going on at Rage3D on the issue. I've read that 1) Boards might be recalled, 2) A new BIOS may be issued, 3) They're still looking into the problem.
Regardless (assuming initial boards will in fact be recalled), I agree that it does indicate the issue may be on ATI's side. However, it may be a case (as I have often observed) in BOTH parties "meeting spec", but somehow that spec is "loose" in some ways and gets "interpreted" differently that results in some incompatibility.
I had read that ATI, for example, did extensive 8X testing on some Intel chipsets with no problems. So apparently, the Intel chipsets are different in some way. (Including perhaps they are "beyond spec" and are more tolerable of "out of spec" AGP boards.)
Well, it's been done in the past. nVidia had "ignored" problems of the original TNT (maybe TNT-2?) working on certain LX chipset motherboards. Neither the motherboard makers nor nVidia "corrected" the problem, and the solution was to either not use those motherboards, or choose a different video card. I certainly agree though, that it would be extremely difficult for ATI to not recall the boards.
Knowing Intel, most likely they are different in that they adhere to the specification exactly.
Doomtrooper
Veteran
The other 8X board the SIS Xabre is also causing lockups on some of these early 8X motherboard offerings..which to me points to two things:
1)Growing pains
2)The motherboards are not in spec
That would be two 8X AGP cards causing issues.
In fact looking at overclockers.com forums, some 8X AGP motherboards won't even post with a Xabre in it...some companies have released BIOS updates to at least get it to post, but lockups continue.
Personally I am waiting for AMD's SOLO chipset with 8X AGP and Hyperstransport and will keep my tried and true Kt266a board that is rock solid stable.
I remember 4X AGP causing the same kind of issues, and who to blame comes into play, the motherboard and the card company or both.
1)Growing pains
2)The motherboards are not in spec
That would be two 8X AGP cards causing issues.
In fact looking at overclockers.com forums, some 8X AGP motherboards won't even post with a Xabre in it...some companies have released BIOS updates to at least get it to post, but lockups continue.
Personally I am waiting for AMD's SOLO chipset with 8X AGP and Hyperstransport and will keep my tried and true Kt266a board that is rock solid stable.
I remember 4X AGP causing the same kind of issues, and who to blame comes into play, the motherboard and the card company or both.
Well, I usually just lurk and don't post much here, but since everyone else in the world seems to have gotten into the act... 
Nice to see this thread get all civil and some real good discussion taking place.
First, some comments re: w buffer.
I just installed the latest Catalyst drivers for my R9000, and I noticed that in the advanced options somewhere (can't remember exactly where, but it was easy to find), there is an option to "emulate w buffer". Don't know if this exists in the R9700 version of the drivers yet... maybe someone can comment on whether this emulation is present on the R9700 drivers and if it would be helpful to Derek Smart.
Next, some obervations re: other posts.
Incredibly, Chalnoth is starting to sound like a moderate in this thread...
No offense, Kassandra, but you really haven't been around these forums long enough to make any such sweeping judgements. OpenGL_Guy is one of the more intelligent people on this forum. Can you really blame him for being, at times, defensive over his team's work on the ATI drivers? Sure the drivers aren't perfect and things break, but they've generally gotten really a worse rap for driver quality than they deserve IMHO.
See above re: w buffer emulation in R9000 drivers. Don't know if this helps you or not.
Am I to infer that OpenGL_Guy has had some PM with you on the specifics of your MT problem? While have no doubt that the ATI devrel people are great folks, sometimes it really helps to be able to bypass the middlemen and go straight to the source. Good luck to both of you in sorting out the problem with your MT implementation.
Really ancient history... Kind of silly to make any judgements based on such ancient hardware. Otherwise I could bring up all kinds of horror stories spanning years of driver hassles re: my old Riva TNT. Is that really relevant to this particular discussion? I don't think so... How about we stick to hardware from the current decade at least.
And DoomTrooper really does have the best smilies!
Nice to see this thread get all civil and some real good discussion taking place.
First, some comments re: w buffer.
I just installed the latest Catalyst drivers for my R9000, and I noticed that in the advanced options somewhere (can't remember exactly where, but it was easy to find), there is an option to "emulate w buffer". Don't know if this exists in the R9700 version of the drivers yet... maybe someone can comment on whether this emulation is present on the R9700 drivers and if it would be helpful to Derek Smart.
Next, some obervations re: other posts.
Chalnoth said:
Chalnoth said:
Incredibly, Chalnoth is starting to sound like a moderate in this thread...
Kassandra said:
No offense, Kassandra, but you really haven't been around these forums long enough to make any such sweeping judgements. OpenGL_Guy is one of the more intelligent people on this forum. Can you really blame him for being, at times, defensive over his team's work on the ATI drivers? Sure the drivers aren't perfect and things break, but they've generally gotten really a worse rap for driver quality than they deserve IMHO.
Derek Smart [3000AD said:]btw, I'm with you on the DaveBaumann report. He's even tracked down some info for me (passed along via PM) related to implementing a W buffer via pixel shaders. Bless his heart.
And OpenGL_Guy is currently trying to find out what exactly is broken in the drivers, relating to my MT implementation.
See above re: w buffer emulation in R9000 drivers. Don't know if this helps you or not.
Am I to infer that OpenGL_Guy has had some PM with you on the specifics of your MT problem? While have no doubt that the ATI devrel people are great folks, sometimes it really helps to be able to bypass the middlemen and go straight to the source. Good luck to both of you in sorting out the problem with your MT implementation.
Steven Hartland said:
Really ancient history...
Kind of silly to make any judgements based on such ancient hardware. Otherwise I could bring up all kinds of horror stories spanning years of driver hassles re: my old Riva TNT. Is that really relevant to this particular discussion? I don't think so... How about we stick to hardware from the current decade at least.And DoomTrooper really does have the best smilies!
For the 8x issue...
1. 9700 IS working with P4S8X - confirmed, check forum@Anand. (Solution: mobo BIOS-flash. Chipset: SiS648.)
2. 9700 IS working with A7V8X - confirmed, check Rage3D, there's a named topic for this. (Solution: new mobo w/ new BIOS. Chipset: VIA KT400.)
Since there is one more available chipset w/ AGP v3.0 standard (VIA P4X400), probably we can assume: the story is about the mobo makers, isn't?
IF it would be 9700's fault then ALL the cards should be def in ANY AGP v3.0 mobo, right? But that's not the case, right?
I don't know...
...my 2 cents...
1. 9700 IS working with P4S8X - confirmed, check forum@Anand. (Solution: mobo BIOS-flash. Chipset: SiS648.)
2. 9700 IS working with A7V8X - confirmed, check Rage3D, there's a named topic for this. (Solution: new mobo w/ new BIOS. Chipset: VIA KT400.)
Since there is one more available chipset w/ AGP v3.0 standard (VIA P4X400), probably we can assume: the story is about the mobo makers, isn't?
IF it would be 9700's fault then ALL the cards should be def in ANY AGP v3.0 mobo, right? But that's not the case, right?
I don't know...
...my 2 cents...
Doomtrooper
Veteran
Interesting reading some of the fallout from this thread, the power of the internet at its best...
http://discuss.madonion.com/forum/s...amp;view=&sb=&o=&fpart=1&vc=1
http://discuss.madonion.com/forum/s...amp;view=&sb=&o=&fpart=1&vc=1
Derek Smart [3000AD]
Regular
Joe DeFuria said:
hmm, you're right. There might be some confusion going on because IIRC, they were already issuing RMAs to people who wanted their boards replaced.
And yes, the fault could be on either side, but until the dust settles, who knows? Its just going to be the traditional finger pointing going on. At least for a while.
Doomtrooper said:For example:
Rage3d has:
22,785 members
Nvnews has:
1,509
We had close to 10,000 members before we had to start the forums from scratch due to nV News being attacked. We couldn't recover since its Rackshack's policy not to allow access to a server that has been compromied without wiping it clean first. We are doing off-site backups now.
Some of your buddies over at Rage3D did have a good laugh at us though.
Doomtrooper
Veteran
There is bad apples on any forum Mike..there were also people that didn't have a laugh.
This wasn't a site comparison thing BTW
, it was meant to show alot of posts will be at Rage3D due to the userbase...people that post are usually people that are having issues...the others are enjoying their card.Derek Smart [3000AD]
Regular
Doomtrooper said:
Yeah, someone sent me email about that this morning. I registered to respond and didn't check my email for my password. Just got it. I've posted. The rest, is history.
Derek Smart [3000AD said:]Yeah, someone sent me email about that this morning. I registered to respond and didn't check my email for my password. Just got it. I've posted. The rest, is history.
Derek, what on earth is this:
?
Why can I activate T&L and see a difference?
http://www.beyond3d.com/reviews/ati/radeon9700pro/index.php?page=page17.inc
Derek Smart [3000AD]
Regular
DaveBaumann said:
Because its running in software emu? Dunno. Ask OpenGL_guy, he might have an answer on what exactly was broken in TnL and subsequently fixed. I'm at a loss, as I do not know all the specifics.
I think you're making blanet statements again - clearly those tests (becuase one is software and one is hardware) indicates that hardware gometry processing is not broken in all cases.
Bigus Dickus
Regular
It would seem that the limiting of broad sweeping statements being kept to a minimum would do wonders for preserving clear communication, quieting insulting backlash, and garnering respect.
For instance: "Multitexturing is broken on the 9xxx cards. ATI admitted it." This could be more accurately worded as "my technique for multitexturing is incompatible with current 9xxx drivers. ATI admitted it."
For instance: "The hardware TnL is broken..." could be more accurately worded as "there is a specific function of hardware TnL that is broken..."
Before you engage in a lecture about why they are saying the same thing, allow me to make a simple point: the first way of stating the problems indicates that it doesn't work, period, for any game, for any developer (that's what most consumers think of as broken... and it is the consumers you care so deeply for, correct?); the second way of stating the problems indicates that it doesn't work for you, on your game. Vast difference there.
You have a point... just make it, don't bury it in a pile of steaming, useless rhetoric. It only promotes these flame wars (which I'm sure you are well aware of). Perhaps that is the true point after all?
About that W-buffer... thanks Rev. So there is Mafia, and the BC series. Any others (still curious)?
For instance: "Multitexturing is broken on the 9xxx cards. ATI admitted it." This could be more accurately worded as "my technique for multitexturing is incompatible with current 9xxx drivers. ATI admitted it."
For instance: "The hardware TnL is broken..." could be more accurately worded as "there is a specific function of hardware TnL that is broken..."
Before you engage in a lecture about why they are saying the same thing, allow me to make a simple point: the first way of stating the problems indicates that it doesn't work, period, for any game, for any developer (that's what most consumers think of as broken... and it is the consumers you care so deeply for, correct?); the second way of stating the problems indicates that it doesn't work for you, on your game. Vast difference there.
You have a point... just make it, don't bury it in a pile of steaming, useless rhetoric. It only promotes these flame wars (which I'm sure you are well aware of). Perhaps that is the true point after all?
About that W-buffer... thanks Rev. So there is Mafia, and the BC series. Any others (still curious)?
Bigus Dickus
Regular
You beat me to it Dave...
Derek Smart [3000AD]
Regular
DaveBaumann said:
If it doesn't work as it should. Its broken. I think thats simple enough, isn't it?
For one thing, since when was software TnL the norm? And I specificcally said hardware TnL is most of the posts I made about this.
As for the MT, there is NO technique that I'm using that would cause it to not work. And until ATI come back and tell me exactly what is broken, why and how they fixed, all we can do is speculate.
That said.
You're not going to get anywhere by telling me that because one aspect doesn't work, so therefore its not broken.
This is not a Clinton issue where the issue of smoking pot is relegated to taking puffs or inhaling. Or the definition of what sex is (blowjobs not withstanding).
So....
if MT doesn't work in my game - or ANY game for that matter - its broken. Whether its one out of fifteen million aspects, is irrelevant. Broken means broken. No ifs ands or buts about it.
if TnL works in software (which defeats the purpose right there) and not in hardware (where its needed the most), its broken.
You folks can argue all you want about what the definition of broken is. And since I'm not having sex with anyone here, I don't have to debate the merits of what constitues sex.
But clearly, in those instances tested (although, lord knows whats going on with SS:SE) hardware T&L is not broken – you can obviously see that it is not operating in software because I am testing the difference between software and hardware geometry. (Also, although there is no specific test, its clear to see from the Dungeon Siege benchmarks that hardware geometry is in operation there).
- Status