Could Anand or someone please explain? I don't get this paragraph in the R300 pre/review

[RedShirt]

First, let me say if this point has been made, I'm sorry, I did a search for ATI and Drivers in the search and didn't find anything like this.

What is meant by this:

"ATI?s drivers have always been their weakness and, unfortunately, it does not look like much has changed with the R300. On the positive side, we didn?t encounter any performance, compatibility or image quality issues with the current build of the R300 drivers during our time with it. The card and its drivers ran through our entire benchmark suite just fine"

The drivers are their weekness, although it ran everything without a hitch??? Huh?

After reading this at the Rage3D forums, I thought I should repeat it here, because this statement makes 0% sense in my mind.

If he was referring to the lack of a unified drivers for this product, there was no unified drivers for the original Radeon... just when the 8500 was coming out is when they started talking about unified drivers, so ATI has had non-unified drivers in the past. Granted, the lack of a unified driver may be one weakness, but I don't think this is near as bad as it was in the past.

What I can't get past is how the drivers were flawless, but they are a weakness?

 

[Booster]

Anand wrote that ATI has a very bad driver reputation, and though he hasn't discovered any major flaws in the R300 drivers, it's too early to say they're all good b/c all the previous ones have had problems and it's hard to believe that absolutely no flaws would be discovered over time as users run different games and software. I think he made a pretty clear statement.

 

[wfbberzerker]

basically, ati has always had a problem with their drivers affecting performance. for example, with the radeon 8500, the drivers that came with the card made the card perform slower than even a geforce3, although on paper, the cards specs made it seem like it should outperform a geforce4. what is being referred to in this article is that although the drivers for the r300 do not have any major flaws (i.e. screwed up pictures, graphical errors, crashes, etc.), it still has the same problem of performance as in the past.

 

[SocrPlyr]

ATI has always had very poor driver support (updates coming about every 6 months...) they have been working on this problem recently but still aren't up to part w/ nvidia... nvidia has flaws in their drivers too but you see a fix very shortly most of the time while w/ ati it may be months... not to mention ati really had problems even with the 8500 drivers when it first came out, many games wouldn't play right etc...
but really Anand really said it all...
(it is great to see he didn't find any probs tho)

Josh

 

[acejj26]

i didn't understand that statement as well....what gave him the idea that the drivers would be a weakness when he said there were no glitches and it wiped the floor with the Ti4600, performance-wise?? if you can't find any glitches, no performance-hindering bugs, and the AA and AF work perfectly, i don't see how you can say that the drivers aren't up to snuff.

 

[mchammer187]

Originally posted by: RedShirt
First, let me say if this point has been made, I'm sorry, I did a search for ATI and Drivers in the search and didn't find anything like this.

What is meant by this:

"ATI?s drivers have always been their weakness and, unfortunately, it does not look like much has changed with the R300. On the positive side, we didn?t encounter any performance, compatibility or image quality issues with the current build of the R300 drivers during our time with it. The card and its drivers ran through our entire benchmark suite just fine"

The drivers are their weekness, although it ran everything without a hitch??? Huh?

After reading this at the Rage3D forums, I thought I should repeat it here, because this statement makes 0% sense in my mind.

If he was referring to the lack of a unified drivers for this product, there was no unified drivers for the original Radeon... just when the 8500 was coming out is when they started talking about unified drivers, so ATI has had non-unified drivers in the past. Granted, the lack of a unified driver may be one weakness, but I don't think this is near as bad as it was in the past.

What I can't get past is how the drivers were flawless, but they are a weakness?

I think what he meant was R300 would have been out the door a lot sooner had they not built the drivers from scratch

and he said it would be hard for them to write bug free drivers from scratch

I dont really understand why a UDA is so good i mean its like windows still uses the old source code it is so bloated

and they pile keep piling on stuff making it bigger and bigger and bigger and more bloated

I am sure a memory leak free OS could be written that is better than Windows but it would be difficult i guess

maybe i am understanding it wrong but this is what i get:

original driver has bugs

they patch

patch it again to work with new hardware (now bloated)

or new driver optimized for new hardware (less bloated)

i would think a UDA would have to be bloated for it to work for all cards IE det's work for GF1 -> GF4 Ti

 

[RedShirt]

Originally posted by: wfbberzerker
basically, ati has always had a problem with their drivers affecting performance. for example, with the radeon 8500, the drivers that came with the card made the card perform slower than even a geforce3, although on paper, the cards specs made it seem like it should outperform a geforce4. what is being referred to in this article is that although the drivers for the r300 do not have any major flaws (i.e. screwed up pictures, graphical errors, crashes, etc.), it still has the same problem of performance as in the past.

That's a pretty big assumption, sure I bet with future drivers the preformance will get up there, but when the card destroys a GF4 Ti4600, has no visual bugs, no preformance bugs, no bugs he can see at all, isn't it a pretty bold statement to say:

"it does not look like much has changed with the R300"

In the past ATI had numorous bug issues with early drivers, in almost anything they sold. As far as Anand can presently see, there are none with the R300.

Now, I am not saying that there are no bugs, cause I bet there are (probably a lot) And I'll also be the first to admit that ATI has had very bad driver support in the past (and still to this day), but this statement still makes no sense to me. He makes it with no proof to back up his claim, only proof to the contrary.

 

[Mavrick007]

Originally posted by: acejj26
i didn't understand that statement as well....what gave him the idea that the drivers would be a weakness when he said there were no glitches and it wiped the floor with the Ti4600, performance-wise?? if you can't find any glitches, no performance-hindering bugs, and the AA and AF work perfectly, i don't see how you can say that the drivers aren't up to snuff.

ATI has had bad driver support for the Radeons especially in the past so it has been their weakness since the hardware is technically better than nVidia's but they have not been able to get the performance out of their cards due to drivers.

This next gen card wipes the floor with the Ti4600 with everything they had time to try it with "so far". When new games and software comes out they are hoping that it will stand up to the test and not crap out with certain hardware configurations or software.

I'm sure that ATI has put alot of time into their drivers this time since they have received alot of bad impressions due to poor drivers. They had to re-engineer them to work correctly and from what Anand's seen, they seem to have done the job right.. but like he said, it's too early to tell cause it's not even a fully released product with drivers operating on consumers machines yet.

 

[dexvx]

I think that poor drivers can have 2 definitions:

1) they are not stable
2) they are not up to performance

You can have a driver that is rock solid and does everything as its supposed to be, but it can also be a poor performer. For instance, there could've been many optimizations to do or getting rid of a lot of redundancy, doubling up, etc. Personally, I'd rather have drivers that have lower performance than sacrificing stability.

Example, Intel's first IP-SEC (IP Security for Network cards) drivers offered very poor performance (beta drivers... wasnt much better than software IP-SEC) but offered the same stability you'd expect from Intel. It took time to optimize their IP-SEC drivers so that it performed to the point where it takes almost no CPU strain.

 

[Mavrick007]

Originally posted by: dexvx
I think that poor drivers can have 2 definitions:

1) they are not stable
2) they are not up to performance

You can have a driver that is rock solid and does everything as its supposed to be, but it can also be a poor performer. For instance, there could've been many optimizations to do or getting rid of a lot of redundancy, doubling up, etc. Personally, I'd rather have drivers that have lower performance than sacrificing stability.

Example, Intel's first IP-SEC (IP Security for Network cards) drivers offered very poor performance (beta drivers... wasnt much better than software IP-SEC) but offered the same stability you'd expect from Intel. It took time to optimize their IP-SEC drivers so that it performed to the point where it takes almost no CPU strain.

True, you can have stable drivers that offer poor performance, such as the drivers released a few months ago from ATI. They had been getting larger frame rates per second with previous drivers but they were highly unstable and giving bsods or kicking out. The new drivers offered better stability but the performance or speed here was alot slower and dropped them from being close. They then caught back up with later driver releases (much more frequent releases) to increase speed and stability.

 

[RedShirt]

Originally posted by: dexvx
I think that poor drivers can have 2 definitions:

1) they are not stable
2) they are not up to performance

You can have a driver that is rock solid and does everything as its supposed to be, but it can also be a poor performer. For instance, there could've been many optimizations to do or getting rid of a lot of redundancy, doubling up, etc. Personally, I'd rather have drivers that have lower performance than sacrificing stability.

Example, Intel's first IP-SEC (IP Security for Network cards) drivers offered very poor performance (beta drivers... wasnt much better than software IP-SEC) but offered the same stability you'd expect from Intel. It took time to optimize their IP-SEC drivers so that it performed to the point where it takes almost no CPU strain.

Here is what I see:

The R300 has the same MSRP as a Geforce4 TI 4600 and has much higher preformance in anything not CPU limited.
The R300 had no performance, compatibility, or image quality issues with what they tested it with.

So the preformance is higher than a a card priced exactly the same, and it was stable.

Yet, this was said:

"ATI?s drivers have always been their weakness and, unfortunately, it does not look like much has changed with the R300."

Where is the weakness that he can see? He doesn't say, he found none.

This is what I don't understand.

Is he saying the preformance will get substantially higher with better drivers? And if so, how does he know this?

And it's not like we haven't seen Nvidia cards get substantial preformance increases with new drivers either...

That remark just seems like a shot at ATI, one which they don't deserve based on the factual evidence Anand presently has.

Later though, after some investigation, it would be fine to make that remark, if it could be backed up.

 

[subhuman]

If you actually finish reading the page there, you'll see that Anand is refering to the fact that they STILL aren't using unified drivers - these are from scratch. In the long run, this is definitely an issue, you have to start over, and that IS a problem.

He's saying they should have a unified driver architecture - if you finish the article, you'll see him say it straight up something along the lines of 'for ati's sake we hope they move to unified drivers.'

 

[WyteWatt]

Does the ATI R300 seem to good to be true for anyone here? I mean as of right now it has like 20 to 50% + more performence in games than the GF 4 ti 4600, the drivers did not cause any problems or stabilty issues, the image quality amazing, still has very good performence even with FSAA on, etc ? I am not saying it won't do well i am just saying don't they say when something seems to good to be true most of the time its not? I mean do you all think the drivers will hold out really well when people start buying the R300 and playing games,etc?

 

[RedShirt]

Originally posted by: subhuman
If you actually finish reading the page there, you'll see that Anand is refering to the fact that they STILL aren't using unified drivers - these are from scratch. In the long run, this is definitely an issue, you have to start over, and that IS a problem.

He's saying they should have a unified driver architecture - if you finish the article, you'll see him say it straight up something along the lines of 'for ati's sake we hope they move to unified drivers.'

Yes, I read the whole article, and that is what he was talking about there. But still, his statement in that paragraph is about ATI's drivers in general, not about the UDA, and ATI has not had a UDA in the past, only around the time of the 8500. So really, the drivers are better than in the past (as far as he can see), yet he says they are just as bad as in the past.

This could be exactly what he meant by this statement, If so, I think he worded it completly wrong, I would have liked to see something like:

While we experienced no problems with our limited time testing the card, problems may arise, as ATI has always been very bad in the past with their drivers. It is too early to say how well these drivers will compare to their drivers of the past, but we will remain skeptical until we can do a fully featured review of this product.

Not having a unified driver architecture hurts ATI, as if they do find flaws in their driver, they will have to update every driver package of each card they sell. Having a unified driver package would allow ATI more time for increasing performance instead of using their time to update every individual driver package with bug fixes.

Instead we got a simple short paragraph that can be taken a million ways. I take it that Anand doesn't think the driver quality is up there, because that is what he has said in the past and he said these drivers are like past drivers from ATI.

 

[dexvx]

What I'm saying is that the R300 could have better performance than it currently has due to its drivers. It performs well now, but it could have better performance.

 

[zzzz]

He said the drivers perform flawlessly in the benchmark suites..
May be they have trouble in games which were not included in the benchmarks?