New ATI Beta/Hotfix Drivers

No, it's not.

It causes a big increase of support issues (for ISVs), compatibility testing have to be done for a lot more versions.
It is harder to make workarounds for all the combination of bugs you can get with different versions.

With frequent release new bugs introduced usually stay in 3-4 consecutive versions - there's little to no advantage there.

Monthly releases causing thight schedules at the IHV which causes less time to spend on compatibility and regression testing of each driver.

No, I don't think it's a good thing at all...

 

Sorry, but...



You're not really serious, are you? You're just trying to play devil's advocate, right?

 

He might have a point. It might be a pain for system builders to have to go through testing/validation each month. Similar to the frequent Microsoft patches in the enterprise sector. It is definitely a plus for the end user though - so as an end user I agree with ATI's approach. Not that it makes a lick of difference to me personally since I like trying out every beta driver I can get my hands on - officially endorsed or not

 

There's a search function on this forum it seems.
http://www.beyond3d.com/forum/viewtopic.php?p=282546&highlight=drivers+nv#282546
http://www.beyond3d.com/forum/viewtopic.php?p=280618&highlight=drivers+nv#280618
etc.

Yeah, but that's not what you said before. It's a good thing, but if the competition doesn't do it...

 

Older bugs are generally still resolved though, and these wouldn't be available for several months with less frequent updates.

Not really, not if the regression testing is separate from the qualification labs (as they are with ATI, AFAIK) - you still have to do the regression testing on the internal builds anyway.

I think ATI do have regression testing issues, and I told their VP Software that on monday. It seems that they are mainly looking at applications for regression testing, and not necessarily specific cases that developers bring up - if the functionality is defined withing an application or WHQL testing they try and catch it, however if they get ad-hoc developer issues with specific cases it seems that they don't catch them as well.

As for ISV testing, with monthly updates they at least know that these drivers adhear to WHQL testing; what drivers should you qualify and test at the moment for NVIDIA? Obviously 61.77 (which aren't actually WHQL) or pick and choose one of the leaks or OEM releases?Which one are most users actually using?

 

Isn't WHQL some form of weird compatibility test that OEMs usually like? Wouldn't monthly WHQL releases sort of negate that bit?

 

Time for the smiley war

Anyway - I was serious, maybe a bit too harsh.

My view is maybe more of a developer's, but it also a user's opinion.

Once you get past that "it impoved my 3dmark score by 30 point and got 330 fps in Q3 instead of 318" thing you might actually get to see there are more to drivers than benchmarks.

Benchmarks don't satisfy me when I got a > .1 sec delay of rendering in a game, hickups once every second when running something windowed, or 3fps instead of 150fps when I have my download manager running, or when everything is dark in fullscreen because gamma is not working.

 

But that's what WHQL testing is for, at least I thought it was...to insure compatibility.

If they release WHQL drivers every month there should be no compatibility issues so there should be no problems, like ATi does.

 

I think driver updates are good on a regular basis, but coming from a consumer point of view it looks bad. Case and point: I sell computers and such to get through college, and if I ever mentioned how ATI was good for updating drivers regularly the customer generally wouldn't purchase ATI even though I recommended them. To the average person, constant updates are a hassle and give the perception of faulty drivers. Worse off, they do not go and buy an Nvidia card, they keep their integrated cards even if they really need a discrete card. Customers do not like updating things, hobbyist do.

 

system builders can pick whatever ones they want to support. They can use drivers every other month , ever 4 months , whatever they want through thier websites . Thta is what most do anyway .

For power users they can go get monthly updates from ati though. I don't see a problem

Its even less of a problem if all the vendors are using the same refrence design.

 

And how do you sell Nvidia based systems? Using :"there will be updates but you have to catch them on the web because if you wait for official drivers on Nvidia website you may never see one ?

 

I'd say release when you have something to release.
Of course, releasing every 3 months exactly is much worse.

Well, if you think about the issue I brough up, the game since gone gold even in English (after quite a few months delay), and AFAIK they still haven't fixed the problem - so they now have an application to test it in.

I'm actually not that much up-to-date about nVidia (none of our developer machines nor my home computer are running with an nVidia card...), but I really wasn't trying to do a comparsion here.
(And I don't know what drivers would I use if nVidia had sent the card they had promised months ago...)

 

i could , but doom3 wont run because of nero on my pc

 

I wasn't saying Nvidia was better so what is your point? I was saying in general customers do not like the idea of updating, so they wouldn't look for updates in the first place. In the past year I have usually recommended ATI systems if they were a good price.

Microsoft uses the automatic updates because customers would not want to do it themselves.

Edited for clarity.

 

And why do you think ATI is going towards automated drivers updates ?

 

Sorry, I didn't mean to infer ATI is going to auto updates. I liked the updates being around regularly when I had my 9700 Pro.

 

How exactly do you do this registry hack? I changed the CatalystAI and CatalystAI_DEF keys to a value of 2. Is that how to do it?

Thanx,
Teflon

 

Just a question really. I saw this in a thread over at FM, it's from an FM employee.

So how did the scores increase so much if it wasn't because of the 256MB cards not being able to use all thier memory? i.m confused now.. :?

 

Not sure, but from what I can gather the reason 256MB cards had poor performance was not because they couldnt allocate memory above 128MB, rather that their Memory allocation system was bugged - resulting in poor performance for any allocation.

Feel free to correct me if Im wrong.

 

http://www.futuremark.com/download/?3dmark05hotfix110.shtml

fix for scores that are infated