What will be AMD'S next Move?

[amenx]

I can not support NVidia if AMD has a reasonable alternative. This is due to how NV does things compared to AMD. NV seems to go out of it's way to lock out AMD, or hurt it's performance. This hurts the industry and consumers, even those who have NVidia since it drastically cuts the adoption of these technologies.

You wouldnt have an Intel CPU btw?

 

[amenx]

But even if 390X beat 980 by 50% for $399, NV fans would keep waiting for GM210 even if it took 6-9 months to come out after the 390X.

No way. Stuff like this what market sensations are made of and buyers would all jump over it in a minute... and then maybe upgrade to GM210.

 

[KaRLiToS]

Nvidia strenght is their marketing and their lies.

MFAA is a cheat.

AMD marketing sucks.

 

[Jumpem]

Nvidia strenght is their marketing and their lies.

MFAA is a cheat.

AMD marketing sucks.

The dozens of independent 970 and 980 reviews are all lies?

 

[f1sherman]

Nvidia strenght is their marketing and their lies.

MFAA is a cheat.

AMD marketing sucks.

They just need to copy it.

Same like they did with previous Nvidia cheat.
What's the name of it... Nvidia GPU Boost!

 

[MTDEW]

I don't understand how MFAA could be a cheat if it works and isn't used in review benchmarks to skew the results vs the competition using traditional 4x MSAA.

 

[Abwx]

MFAA is a cheat.

What is MFAA.??.

 

[Keysplayr]

What is MFAA.??.

You haven't read any GTX980/970 reviews?

 

[notty22]

I am not talking about market share. I am talking about how NV fans refuse to acknowledge price/performance, overclocking and dual GPU value when it comes to AMD. I've been following this forum for so long and with seeing 290s vs. 780Ti and now the hype of 970 SLI, it's blatantly obvious there is a double standard. I will stick to my original assertion that NV fans just don't care for AMD products at all. They only care in a sense that when AMD forces NV to drop prices, they like buying cheaper NV cards.

Right now the market share erosion of AMD is coming from brand agnostic users such as myself switching sides. But even if 390X beat 980 by 50% for $399, NV fans would keep waiting for GM210 even if it took 6-9 months to come out after the 390X. And this is the scariest situation for AMD since they cannot rely on brand agnostic users to only buy AMD products like NV can rely on its Apple-like userbase.

You should apply for Dr. Phils job. You rant this personal Psychological evaluation of fellow forum posters to no end, I personally find it repetitive, off topic and nonsense.


edit: Just to add, I do enjoy the majority of your posts, and can appreciate your observations/experience with computers.

 

[MTDEW]

MFAA = Multi-Frame Sampled AA
Its supposed to give 4x MSAA image quality with only the performance hit of 2x MSAA.

 

[Keysplayr]

I don't understand how MFAA could be a cheat if it works and isn't used in review benchmarks to skew the results vs the competition using traditional 4x MSAA.

Agreed. I'd like to hear this one.

 

[Grooveriding]

I thought MFAA was interesting at first, but you quickly realize it will mostly be useless because you need to apply it at the driver level. Hardly any games that release today allow that as they are almost all using deferred rendering. Crysis 3, BF4, Metro LL, Watchdogs etc. etc. none allow driver level AA. Almost every game released is now deferred so there is not much use for it as a feature in anything but older games where MSAA is not going to be a problem performance wise.

 

[Abwx]

You haven't read any GTX980/970 reviews?

I mainly did read around the power management scheme after THG published their findings, i still didnt had the time to look at other caracteristics, i dont like to just have an horizontal knowledge of things, i prefer more vertical, in depth analysis, even if this require some time to have a clue about all the aspects.

Agreed. I'd like to hear this one.

So no need to read reviews, a simple post is enough after all...

 

[Alatar]

no one talked about Nvidia terrible power consumption with Fermi

You can't possibly be serious. The flak AMD has gotten from power consumption and temps is nothing compared to Fermi.

Allow me to illustrate my points with these:

http://www.youtube.com/watch?v=2QkyfGJgcwQ
http://i.imgur.com/J6wCN47.gif
http://i.imgur.com/lznqQ1l.jpg
http://i.imgur.com/GhxsOfa.jpg
http://i.imgur.com/nRtAe56.jpg
http://i.imgur.com/J6sq7sJ.jpg
http://i.imgur.com/IUdYo5t.png
http://i.imgur.com/lINic1P.jpg
http://i.imgur.com/1aVrQKU.jpg

^And there are countless pictures gifs etc. like this (and they're all hilarious).

For many enthusiasts Fermi launch is right up there with Bulldozer.

 

[Abwx]

MFAA = Multi-Frame Sampled AA
Its supposed to give 4x MSAA image quality with only the performance hit of 2x MSAA.

Ok, thanks, i did read something from Hardware.fr but so far the reviewer states that it should be verified, i post the full text and then a gross transcription , other could use google translate to check the accuracy.

Faire varier la position des samples, c'est déjà ce que faisais ATI et AMD avec le Temporal AA. Et c'est ce que va faire Nvidia avec le MFAA qui sera disponible d'ici quelques temps via de nouveaux pilotes. Nvidia ne s'est cependant pas contenté de recopier ce principe mais y a ajouté un filtre plus avancé qui combine downsampling et composante temporale basée sur l'image précédente. De quoi se rapprocher de la qualité du MSAA 4x avec seulement du MSAA 2x avec alternance de la position des samples (MFAA 2x), selon Nvidia. Il faudra vérifier en pratique ce qu'il en est, notamment en mouvement.

"Varying the samples positions, thats what ATI and AMD already did with Temporal SA.

And this is what NVvidia will do with MFAA wich will be available in some times via new drivers.

Nvidia didnt just copy this principle but they also did add a more advanced filter wich combine downsampling and temporal component based on the previous frame.

That should get close of MSAA 4X quality with only MSA 2X s samples alternance position (MFAA 2X) according to Nvidia.

It should be checked in practice how it fares , particularly when in motion."

http://www.hardware.fr/articles/928-23/gm204-smm-nouveautes.html

 

[BrightCandle]

MFAA is potentially a cheat and there is a good reason to be wary of the claims.

When everything is still you do in essence get MSAA 4x by doing half of the work on every other frame. Indeed you could in theory do some sort of MSAA 1x and sample 4 frames if you really wanted when all is still and you aren't moving as it would have the same effect. It will produce screenshots that look the same as MSAA 4x applied, but at a higher performance.

However when you are moving the previous frames sampling isn't likely to help you since the object has moved. So you either throw the data away (what I suspect Nvidia does and why they say worst case it looks like MSAA 2x) or you some how move it along with the object and then blend it (unlikely, probably not possible).

The end result might very well be the sort of thing that looks good in screenshots but that in actual gameplay and motion is no better than MSAA 2x. All depends whether you notice the jaggies while you are moving or whether you notice them when still. Personally its the shimmering todays MSAA technique causes that annoys me more than the jaggies themselves and depending on what Nvidia does while in motion depends on if this helps. I hope its more like TXAA in that regard and the temporal sampling helps the shimmer effect but it doesn't work yet so no one has reviewed what its like in practice.

 

[AtenRa]

For many enthusiasts Fermi launch is right up there with Bulldozer.

Thats why i have both :biggrin:

 

[rgallant]

I am not talking about market share. I am talking about how NV fans refuse to acknowledge price/performance, overclocking and dual GPU value when it comes to AMD. I've been following this forum for so long and with seeing 290s vs. 780Ti and now the hype of 970 SLI, it's blatantly obvious there is a double standard. I will stick to my original assertion that NV fans just don't care for AMD products at all. They only care in a sense that when AMD forces NV to drop prices, they like buying cheaper NV cards.

Right now the market share erosion of AMD is coming from brand agnostic users such as myself switching sides. But even if 390X beat 980 by 50% for $399, NV fans would keep waiting for GM210 even if it took 6-9 months to come out after the 390X. And this is the scariest situation for AMD since they cannot rely on brand agnostic users to only buy AMD products like NV can rely on its Apple-like userbase.

I agree with all you said
but high lighted the what if nv knows what amd has coming down the pipe.

 

[Zodiark1593]

Thats why i have both :biggrin:

Does a 5.5 GHz bulldozer and quad GTX 480s fit the bill?

 

[SirPauly]

MFAA is potentially a cheat and there is a good reason to be wary of the claims.
.

Why? ATI offered temporal AA and their x2 offered similar quality as x4; as long as you had 60 FPS -- could get away with 48-50fps threshold at times. Hopefully, this is more nVidia advanced -- and would like to see if MFAA is possible with TXAA, SSAA, TRAA moving forward.

 

[SirPauly]

I thought MFAA was interesting at first, but you quickly realize it will mostly be useless because you need to apply it at the driver level. Hardly any games that release today allow that as they are almost all using deferred rendering. Crysis 3, BF4, Metro LL, Watchdogs etc. etc. none allow driver level AA. Almost every game released is now deferred so there is not much use for it as a feature in anything but older games where MSAA is not going to be a problem performance wise.

There is a setting called enhance the application!

 

[ams23]

Crysis 3, BF4, Metro LL, Watchdogs etc. etc. none allow driver level AA.

Well clearly your statement is inaccurate because NVIDIA actually measured MFAA performance using Watch Dogs (in addition to Far Cry 3, Hitman Absolution, Assassin's Creed IV Black Flag, and Call of Duty: Ghosts):

 

[Gloomy]

Why? ATI offered temporal AA and their x2 offered similar quality as x4; as long as you had 60 FPS -- could get away with 48-50fps threshold at times. Hopefully, this is more nVidia advanced -- and would like to see if MFAA is possible with TXAA, SSAA, TRAA moving forward.

TXAA already does something like MFAA right? It's 2xMSAA + some temporal mumbo jumbo. Not sure how MFAA fits in, since it's also 2xMSAA + some temporal mumbo jumbo.

 

[ams23]

TXAA already does something like MFAA right? It's 2xMSAA + some temporal mumbo jumbo. Not sure how MFAA fits in, since it's also 2xMSAA + some temporal mumbo jumbo.

According to Tech Report, MFAA doesn't require integration into a game engine such as TXAA. MFAA requires Maxwell Gen 2 hardware with it's new more programmable and more flexible pixel and subpixel sampling pattern ability in the ROP's, so we won't be seeing this feature on Kepler or older gen hardware.

 

[SirPauly]

TXAA already does something like MFAA right? It's 2xMSAA + some temporal mumbo jumbo. Not sure how MFAA fits in, since it's also 2xMSAA + some temporal mumbo jumbo.

TXAA's foundation is built around multi-sampling and may be possible for x2TXAA with MFAA to offer the quality of x4 TXAA!

Maxwell's new RAM-based sample position technology can still be programmed with standard MSAA and TXAA patterns,