From ATI to AMD back to ATI? A Journey in Futility @ [H]

Lets be real though, have AMD drivers ever been awesome on launch for new cards? Rarely. As much as they have made significant improvements in the driver side of things, I'm not expecting them to absolutely nail it for the first few weeks to month.. hopefully with this 'extended alpha launch' they can tighten them down a little more.
 
N4CR said:
Lets be real though, have AMD drivers ever been awesome on launch for new cards? Rarely. As much as they have made significant improvements in the driver side of things, I'm not expecting them to absolutely nail it for the first few weeks to month.. hopefully with this 'extended alpha launch' they can tighten them down a little more.
Click to expand...
we are talking about a 20% difference though that is pretty large for just drivers on more than one game, and we know they had this GPU ready since last year lol.
 
razor1 said:
we are talking about a 20% difference though that is pretty large for just drivers on more than one game, and we know they had this GPU ready since last year lol.
Click to expand...

I've seen drivers pushing 10-15% or more improvements regularly over the last decade or two.. sometimes even more at 25%+ lol. This is AMD we're talking about, cut them some slack to fuck some drivers up will ya! ;p
 
hmm not over all, I have seen it happen in limited scenarios over specific titles, but over all never.

Drivers are made well before silicon is ready, they have banks of computers emulating how a new GPU works and this is where you need drivers done even before it goes to the fab to manufactured.

Once they have the silicon in hand they will tune drivers. We know it was in Dec they had silicon in hand. Now its June, they already had six months for final tuning of the driver.

Now if AMD is purposefully holding back performance for launch day, I can see that happening, but not because a driver isn't already ready.
 
Last edited:
I don't know. The benchmarks that were out there for the 480 at stock running 1080p showed it significantly behind my 390@1100/1650.

Then again, it also showed my 390 numbers were almost matching the 980 stock numbers they posted for 1080p, so either their benchmark numbers for other cards were all on the low side or there really isn't as much throughput difference between $300 and $450 cards from the start of this year.
 
New architecture, new drivers. To think that they won't improve performance is disingenuous at best. Both vendors improve quite a bit as the tech matures. Please consider the source people.
 
Bowman15 said:
New architecture, new drivers. To think that they won't improve performance is disingenuous at best. Both vendors improve quite a bit as the tech matures. Please consider the source people.
Click to expand...

Really 20%+ improvement in a life time on all applications? When have we seen that? Show me some driver comparison benchmarks that valid that statement.
 
razor1 said:
Now if AMD is purposefully holding back performance for launch day, I can see that happening, but not because a driver isn't already ready.
Click to expand...

Likely a mix of both which isn't unheard of. They often just go after most popular titles initially. I'd wager they don't have enough budget to work at the same level as Nvidia with simulations of GPUs, not to the same extent at least.
 
You can't have a successful hardware company without a solid driver team, nV can throw more resources at it in theory, but only up to a certain point, you end up with a certain point where more resources just doesn't help, it will actually be detrimental. Old adage too many cooks spoil the broth.

AMD has been been doing a good job with Day 0 drivers for Dx12 for the past few months.

the time they have to do create drivers is stretched over a long period of time. Added to this the pipelines of Polaris are staying very similar to older GCN products. There are some front end changes but the shader arrays are very similar. So drivers wise you aren't going to be going back to the drawing board and starting from stretch. Many of the optimizations that were done in the past will pan over to Polaris of course they will probably have to tweak parts or rewrite segments of the driver.
 
razor1 said:


video comparison overclocked 970 vs rx 480 probably stock. GTA V
Click to expand...


Lets wait for launch and card drivers. Benchmarks where the video doesnt even know what speeds it used, are kind of BS... Most of these benchmarks show no improvements over what a 380x would do with 1266 clocks, so I dont think the right drivers are released yet.

And no, we are not talking about the difference between "initial drivers vs improved drivers"... We are talking about the difference between "incorrect drivers vs correct drivers"... No one is claiming the drivers arent optimized yes, they are claiming they are outright using the card wrong as they arent the right ones.
 
So you don't want anyone to post anything till NDA is lifted?

Come on, info is info what people want to believe in is up to them, lets discuss. I have already stated something doesn't look right, but no need to keep silent and post nothing at all.
 
I've posted my thoughts on the bechmarks... And yes, you have to... But you continue to post the benchmarks without disclaimer, so other reads that havent seen your statement will simply think that these are accurate. They are misleading posts in that context... Thats why I add the disclaimer for you, in my post :)
 
N4CR said:
Likely a mix of both which isn't unheard of. They often just go after most popular titles initially. I'd wager they don't have enough budget to work at the same level as Nvidia with simulations of GPUs, not to the same extent at least.
Click to expand...

Look at what happened with the 290X and how long it has lasted compared to the 780 Ti when they are able to code the latest drivers for individual games. There you can defiantly get anywhere from 10-30% bump in performance with patches and drivers.
 
  • Like
Reactions: N4CR
like this
Bowman15 said:
Look at what happened with the 290X and how long it has lasted compared to the 780 Ti when they are able to code the latest drivers for individual games. There you can defiantly get anywhere from 10-30% bump in performance with patches and drivers.
Click to expand...

10% change there are reviews of that out, The entire nV's been holding back the 780 ti performance, thing........ Which was false.

GTX 780 Ti vs R9 290X; The Rematch - Page 8
 
Last edited:
It will be interesting to see where it pans out after the dust settles. X800 had some decent bumps and even after market drivers..
 
N4CR said:
It will be interesting to see where it pans out after the dust settles. X800 had some decent bumps and even after market drivers..
Click to expand...

There is supposed to be a release day driver on the 29th . I"m not sure if that will be the same driver reviewers get or not?
 
The driver team will probably be tweaking the driver until launch day. I'm sure as more reviewers get cards RTG will find places where it needs to be tweaked. AKA, completely normal launch behavior.
 
They also need to give time for the reviewers to use them though, so hopefully not too close to launch.....
 
Driver development never really stops, though. When to release the drivers is an art in and of itself. But, yes, release drivers are probably done and waiting for silicone to be sold and in the hands of users so they can use their new shiny cards; but there will undoubtedly be performance tweaks that will be available after release.
 
Algrim said:
Driver development never really stops, though. When to release the drivers is an art in and of itself. But, yes, release drivers are probably done and waiting for silicone to be sold and in the hands of users so they can use their new shiny cards; but there will undoubtedly be performance tweaks that will be available after release.
Click to expand...

Yeah but if history tells us anything it will probably be the review driver released on the 29th for us peasants. :p
 
lolfail9001 said:
You are just exposing how little you know, man. And your signature still contains contradictions.


Also, while we're on that, if you ask me, Tonga was larger failure than Polaris 10 from technical PoV. Of course now i have suspicion that failure was set up by AMD with the whole GCN's SME thing, but nonetheless.
Click to expand...

Prove to me how little I know. Also, well you are at it, prove to me the "contradictions". LOL Accusations are fine as long as they are backed up with fact.
 
Moving on to the leaked gaming benchmarks. In the five games that have been benchmarked, the RX 480 delivered performance that’s very close to the R9 Fury and GTX 980. Often trading blows with both cards whilst running at the same settings.
Click to expand...

If this holds, I feel really sorry for anyone who bought into Fiji - especially at full price :p
 
ManofGod said:
Prove to me how little I know. Also, well you are at it, prove to me the "contradictions". LOL Accusations are fine as long as they are backed up with fact.
Click to expand...
Let's start with simple fact, that EMA stands for Explicit Multi-Adapter. Yes, the hint is in the name, decipher it yourself.

As for contradiction, you had it pointed out, yet you stand blind, so i'll keep poking at that for now.
 
razor1 said:


3Dmark comparisons different cards to the rx 480



CS preview with the rx480.
Click to expand...

Something's up with Firestrike results. I mean, is the disparity between 390x and 980 that big there O_O? Though i suppose it may have something to do with tesselation.
 
lolfail9001 said:
Something's up with Firestrike results. I mean, is the disparity between 390x and 980 that big there O_O? Though i suppose it may have something to do with tesselation.
Click to expand...

The disparity between the 390x and 980 depends on the test. The latest performance summary from TPU shows the 980 slightly ahead at 1080p, the 390x slightly ahead at 1440p, and further ahead at 4k when 20+ games are averaged.
 
Daniel_Chang said:
The disparity between the 390x and 980 depends on the test. The latest performance summary from TPU shows the 980 slightly ahead at 1080p, the 390x slightly ahead at 1440p, and further ahead at 4k when 20+ games are averaged.
Click to expand...
I talk about FireStrike disparity, i know that in games they are neck and neck.
 
Algrim said:
Driver development never really stops, though. When to release the drivers is an art in and of itself. But, yes, release drivers are probably done and waiting for silicone to be sold and in the hands of users so they can use their new shiny cards; but there will undoubtedly be performance tweaks that will be available after release.
Click to expand...

Yep.
Its not just down to getting the best performance from the card, its also about getting the best performance from new games.
Many have been shipped broken, especially for multi card.
And sometimes the driver team will spot a way to bypass tedious code.

This post about a guys experience in the Vista era highlights the issues.
Things are probably a bit better now and the new DX12 theoretically takes the load off the driver team, but we cant be sure yet.
Some or much of this is bound to still ring true.
AMD R9 390X, Nvidia GTX 980 Ti and Titan X Benchmarks Leaked
 
You must log in or register to reply here.
Back
Top