nVidia Application mode slow because of "debugging"
- Thread starter martrox
- Start date
DaveBaumann said:
I never said nVidia had those fixes ready for years either
What I meant is that if they'd, for example, have it ready tommorow, it would be ridiculous to release it tommorow. Heck, many NV31 / NV34 reviews are already finished anyway...
Of course, if they had the Det 50 ready two months ago, then it'd make no sense to wait. But I doubt they finished it so early... So, assuming they didn't, it does make sense for them to wait for the NV35 launch.
Uttar
Does anyone else see the absurdity of this? nVidia, whose reputation has been built on great heardware and some of the best support, has produced a whole series of graphics cards with, at the least, questionable hardware and absolutely terrible drivers, to the extent they are claiming all (well, almost all?) will be corrected in the next update. Just how many times will this boy cry "wolf"!
Uttar said:I never said nVidia had those fixes ready for years either
Errr, eh?
OpenGL guy
Veteran
Er, yeah whatever.PSarge said:
"Catching these error" codes should not influence hardware limited cases.
Well, it seems like it is either a BS smoke screen to hide behind while they work on drivers to more successfully hide performance/image quality tradeoff issues from users, at the least as far benchmarks and colored mip level screenshots...or it is a delaying tactic for dealing with some real issue with a real solution.
Whether the real issue and solution is due to "debugging", some work around they are trying to see if they can avoid or will not have to use with subsequent hardware, or User Perception Engineering, I don't know. I can offer my guess as to which it is without further information, but if there really isn't going to be a delay in releasing the drivers in question, that doesn't seem productive or necessary.
I'll just get my "pester Wavey to test and find out for me" typing muscles limbered up.
"pester"
Wavey <-Warming up.
Whether the real issue and solution is due to "debugging", some work around they are trying to see if they can avoid or will not have to use with subsequent hardware, or User Perception Engineering, I don't know. I can offer my guess as to which it is without further information, but if there really isn't going to be a delay in releasing the drivers in question, that doesn't seem productive or necessary.
I'll just get my "pester Wavey to test and find out for me" typing muscles limbered up.
"pester"
Wavey <-Warming up.Uttar said:I never said nVidia had those fixes ready for years either
What I meant is that if they'd, for example, have it ready tommorow, it would be ridiculous to release it tommorow. Heck, many NV31 / NV34 reviews are already finished anyway...
Of course, if they had the Det 50 ready two months ago, then it'd make no sense to wait. But I doubt they finished it so early... So, assuming they didn't, it does make sense for them to wait for the NV35 launch.
Uttar
it doesnt make sense to give good driver to CUSTOMER rigth now because review are finish ???? I think the goal of a compagny is to make happy their customer ... no ?
Hellbinder
Banned
Exactly. I dont believe for one second that there is some mysterious *debugging* going on for their only True Tri based AF. I personally think its a complete Fabrication, likely by Nvidia themselves. Or perhaps it is slightly True in some non-related *we can stretch this one* type CYA action.
Or do you guys all really think that they would shoot themselves in the foot for 3 Nv3x driver releases since the first tests started.
Almost everything out of nV's marketing mouth since the FX's "launch" has been marketing spin, so forgive me if I doubt their truthfulness regarding their latest excuse. If nV has increased App speed, I'd like to see at what IQ cost it came, if any.
Moving to Q/B/P from A/Q/P is a step in the right direction, though.
Moving to Q/B/P from A/Q/P is a step in the right direction, though.
Usually, d3d debug runtime is doing all the checking. I think what's happening is that the slider influences mipmapping quality. Instead of you getting bilinear img filter for each mipmap + interpolation between two closest mip levels you get no bilinear fiter but interpolation or bilinear filter but no interpolation or some unique not avail. to dev. setting nvidia is tight lipped about.
OpenGL guy said:
Agreed, I wasn't really saying that's why they're getting bad performance, but more that I expect that's what this guy was meaning (I think it's BS anyway).
Put yourself in NV's shoes. You have to come up with some way to explain why your performance sucks in "Application" mode. Which sounds better "It's because we switch off all our dodgy shortcut code" or "It's because there's extra debug code run in that mode"?
martrox said:
Exactly. I don't buy it. More PR bs from the kings of spin in the videocard business.
Livecoma said:That strategy works great for Comcast. (Sorry had to vent.
Hmmmm... in our area, we went from MediaOne (terrible service) to AT&T (even worse service - was being sued by city!) to Comcast ( these guys really suck!)..... needless to say, I've got DirecTv.......
martrox said:Livecoma said:That strategy works great for Comcast. (Sorry had to vent.
Hmmmm... in our area, we went from MediaOne (terrible service) to AT&T (even worse service - was being sued by city!) to Comcast ( these guys really suck!)..... needless to say, I've got DirecTv.......
Well two of the greatest prime evils (AT&T Broadband & Comcast) have just merged. Next thing ya know Verizon will hook up with MCI and Enron will hook up with PSE&G... NVIDIA and MS to stay on topic...
