AMD Pure Hair comes at a very low performance hit - Great stuff

Rroff · 2 Feb 2016 at 11:57

Don't take my word for it - ask AMD - they are quite vocal on twitter I'm sure they'd be happy to confirm or deny it.

I'm not at all trying to diss PureHair which is a great product - its just tedious when people use a poor understanding of it to try and use as ammo to attack HairWorks/GameWorks.

I'm a big fan of advanced physics in games I don't care who brings it to the table just that we see advances in it.

Nexus18 · 2 Feb 2016 at 21:48

Well I got this game and it is stunning but I found this rather weird.... the start up screen, you got nvidia gameworks logo, no AMD logo and this as a "tip" in the main screen:

No acknowledgement towards AMD at all......

N19h7m4r3 · 2 Feb 2016 at 21:51

Nexus18 said:
Well I got this game and it is stunning but I found this rather weird.... the start up screen, you got nvidia gameworks logo, no AMD logo and this as a "tip" in the main screen:

No acknowledgement towards AMD at all......

Issue is there no acknowledgement from AMD themselves anywhere either. They're completely silent on the entire matter.

AlamoX · 2 Feb 2016 at 21:59

Nexus18 said:
Well I got this game and it is stunning but I found this rather weird.... the start up screen, you got nvidia gameworks logo, no AMD logo and this as a "tip" in the main screen:

No acknowledgement towards AMD at all......

that's not an acknowledgement of Nvidia, it's sponsoring by Nvidia, they paid for that sentence, AMD didn't, and it's just like an add banner space, you want it you pay.

Marine-RX179 · 2 Feb 2016 at 22:10

AlamoX said:
that's not an acknowledgement of Nvidia, it's sponsoring by Nvidia, they paid for that sentence, AMD didn't, and it's just like an add banner space, you want it you pay.

The irony of that sentence is that the direction that Nvidia heading is not "pushing graphic performance", but "pushing graphic performance away" in favor of graphics "quality and features" that offer little benefits in return at the cost of heavy loss of graphic performance

N19h7m4r3 · 2 Feb 2016 at 22:19

Marine-RX179 said:
The irony of that sentence is that the direction that Nvidia heading is not "pushing graphic performance", but "pushing graphic performance away" in favor of graphics "quality and features" that offer little benefits in return at the cost of heavy loss of graphic performance

Yup, because HBAO+ is such a performance killer in Rise of the Tomb Raider. Has less of a impact on the FURY X than the 980Ti

http://www.overclock3d.net/reviews/...c_performance_retested_with_new_amd_drivers/6

AlamoX · 2 Feb 2016 at 23:49

Marine-RX179 said:
The irony of that sentence is that the direction that Nvidia heading is not "pushing graphic performance", but "pushing graphic performance away" in favor of graphics "quality and features" that offer little benefits in return at the cost of heavy loss of graphic performance

well pushing performance away yes, but pushing further for one vendor more than the other, one can argue
and graphic quality is still a selling point, even if there is no real effective increase in quality, marketing the notion of image quality could be beneficial for the vendor.

Nexus18 · 3 Feb 2016 at 00:00

AlamoX said:
that's not an acknowledgement of Nvidia, it's sponsoring by Nvidia, they paid for that sentence, AMD didn't, and it's just like an add banner space, you want it you pay.

Yup spot on, I have been saying this for years, anything that has a logo at the start up of something or somewhere is always a "paid for advertisement", do people think the ads in football stadiums, on tv, before films start in the cinema etc. etc. etc. are done for free....

This is even more true if a game is being given away with a product for free, do people really think that a publisher is going to give thousands of game codes away with nothing to show for in return???

And before people think this is just an anti-nvidia post, the same also applies for AMD and their "sponsored" games.

N19h7m4r3 said:
Issue is there no acknowledgement from AMD themselves anywhere either. They're completely silent on the entire matter.

Yup, this part is confusing, if AMD had no involvement at all, I honestly don't think Hallock would have posted what he did and if it was a lie, I'm sure nvidia and the TR developers would have quickly responded to that.... The best I can think of is AMD had paid/offered a decent amount of money for the game to be sponsored by them (and maybe even provided on site support) but then nvidia came along in the last few months before release and offered an even larger sum and this has caused the issues/silence, rename of tressfx etc. Assuming CD and EM have used AMD's source code then that is very poor show from them regardless of nvidia's involvement.

It just seems odd that a game, which was going to have some form of dx 12 and be AMD sponsored all of a sudden changes a month or 2 before release and no mention about AMD's work/product at all.....

Edit: who knows, maybe amd ****** up something and/or annoyed CD and EM......

This whole async and dx 12 stuff being cancelled/delayed or whatever atm is down right dodgy too.

AlamoX · 3 Feb 2016 at 00:40

Nvidia have enough market share to have their words heard by Devs, if they dont want games not using async for now they probably can influence the cadence somehow at some extent, beside Win7 desktop is still at 52% down from 55%, so there is still no real incentive for DX12 games, Vulcan lagging behind, Nvidia is really getting lucky

because if really as ppl were expecting and dx12 games started rolling in 2015/2016, it would have been a bloodbath, and probably a significant change in market share.

Rroff · 3 Feb 2016 at 07:46

Nexus18 said:
but then nvidia came along in the last few months before release and offered an even larger sum and this has caused the issues/silence, rename of tressfx etc. Assuming CD and EM have used AMD's source code then that is very poor show from them regardless of nvidia's involvement.

It just seems odd that a game, which was going to have some form of dx 12 and be AMD sponsored all of a sudden changes a month or 2 before release and no mention about AMD's work/product at all.....

Eidos showed demonstrations of what it called PureHair March last year and were talking about it months before that it isn't a last minute rename of TressFX.

Stanners · 3 Feb 2016 at 08:36

Nexus18 said:
Yup spot on, I have been saying this for years, anything that has a logo at the start up of something or somewhere is always a "paid for advertisement", do people think the ads in football stadiums, on tv, before films start in the cinema etc. etc. etc. are done for free....

This is even more true if a game is being given away with a product for free, do people really think that a publisher is going to give thousands of game codes away with nothing to show for in return???

And before people think this is just an anti-nvidia post, the same also applies for AMD and their "sponsored" games.

Yup, this part is confusing, if AMD had no involvement at all, I honestly don't think Hallock would have posted what he did and if it was a lie, I'm sure nvidia and the TR developers would have quickly responded to that.... The best I can think of is AMD had paid/offered a decent amount of money for the game to be sponsored by them (and maybe even provided on site support) but then nvidia came along in the last few months before release and offered an even larger sum and this has caused the issues/silence, rename of tressfx etc. Assuming CD and EM have used AMD's source code then that is very poor show from them regardless of nvidia's involvement.

It just seems odd that a game, which was going to have some form of dx 12 and be AMD sponsored all of a sudden changes a month or 2 before release and no mention about AMD's work/product at all.....

Edit: who knows, maybe amd ****** up something and/or annoyed CD and EM......

This whole async and dx 12 stuff being cancelled/delayed or whatever atm is down right dodgy too.

Do you actually think that companies pay sums of money without a contract to protect them and allows another to stroll in at the last minute. Game gazumping!

Insanties_birth · 3 Feb 2016 at 08:52

N19h7m4r3 said:
Yup, because HBAO+ is such a performance killer in Rise of the Tomb Raider. Has less of a impact on the FURY X than the 980Ti

http://www.overclock3d.net/reviews/...c_performance_retested_with_new_amd_drivers/6

That does a good job of highlighting HBAO+ is quite efficient and doesn't hamper performance but to try and treat the argument as if HBAO+ is all that gameworks do and none of the other features could cause performance hampering is a bit one sided. Still that is good news for HBAO+ use at least.

AlamoX · 3 Feb 2016 at 09:07

Insanties_birth said:
That does a good job of highlighting HBAO+ is quite efficient and doesn't hamper performance but to try and treat the argument as if HBAO+ is all that gameworks do and none of the other features could cause performance hampering is a bit one sided. Still that is good news for HBAO+ use at least.

actualy hbao+ hamper perf, but his point is that it impacts 980Ti more than is does with Fury
my question thou is the AO effect is the AMD Ambient Occlusion FX in gpuopen ? or is ambient occlusion from the devs

KillBoY_UK · 3 Feb 2016 at 10:03

AlamoX said:
actualy hbao+ hamper perf, but his point is that it impacts 980Ti more than is does with Fury
my question thou is the AO effect is the AMD Ambient Occlusion FX in gpuopen ? or is ambient occlusion from the devs

Doesn't GTA5 have a AMD branded HBAO effect as well as Nvidia's HBAO+ ?

Nexus18 · 3 Feb 2016 at 10:06

Hbao+ also ran very well in the division beta.

Rroff said:
Eidos showed demonstrations of what it called PureHair March last year and were talking about it months before that it isn't a last minute rename of TressFX.

Ah fair enough, I thought I saw a few people in here posting about that.

Stanners said:
Do you actually think that companies pay sums of money without a contract to protect them and allows another to stroll in at the last minute. Game gazumping!

Of course there will be a contract of some kind although companies can always get out of contracts if one or more of the T&C's weren't adhered to by said companies or if both parties agreed to revoke the contract.

Do you not find it odd how it was going to be an amd sponsored game up until the last month or 2 and now no mention of amd at all, just nvidia... I would find it very hard to believe that amd didn't have some form of a contract so something must have happened but we will never know exactly what.

fs123 · 3 Feb 2016 at 10:07

The game seems to get impacted by Anisotropic Filtering quite a lot. There is a difference of almost 10 fps going from 4x to 16x. Anyone else experience this?

N19h7m4r3 · 3 Feb 2016 at 10:23

Insanties_birth said:
That does a good job of highlighting HBAO+ is quite efficient and doesn't hamper performance but to try and treat the argument as if HBAO+ is all that gameworks do and none of the other features could cause performance hampering is a bit one sided. Still that is good news for HBAO+ use at least.

As it stands HBAO+ is the only Gamework's tech in RoTTR, so it's rather apt here. It's certainly come a long way since I first tried to use HBAO+ in older games.

KillBoY_UK said:
Doesn't GTA5 have a AMD branded HBAO effect as well as Nvidia's HBAO+ ?

Yes, the game has tech from both sides, and it also recently got NVIDIA PCSS+ shadows in there as well.

fs123 said:
The game seems to get impacted by Anisotropic Filtering quite a lot. There is a difference of almost 10 fps going from 4x to 16x. Anyone else experience this?

Oh yes, I'm not sure why it has such a dramatic impact at all. I haven't experienced something like this since I used a GeForce 4 and 6.

maonayze · 3 Feb 2016 at 10:31

Forgetting all the other ins and outs for a minute....is anyone having favourable results from the new 16.1.1 driver?

I have tried it and the stuttering is still present. Has anyone eliminated it completely and if you have, what graphic and display settings are you using.

thanks

Rroff · 3 Feb 2016 at 10:43

N19h7m4r3 said:
Oh yes, I'm not sure why it has such a dramatic impact at all. I haven't experienced something like this since I used a GeForce 4 and 6.

Bit out of my depth here as most of my GPU programming experience predates DX9 but I think it has some impact on fragment shaders - higher levels of anisotropic filtering increases the amount of texture accesses for your sampler states or something - anisotropic filtering is only applied at certain angles so I assume at lower levels it reduces the amount of samples you have to deal with on distant objects - at higher levels I assume this forces increased samples to apply the effect over the now higher detail area. Someone with more upto date experience can probably set me right.

Razor Time · 3 Feb 2016 at 11:09

AF hit between 2x and 16x is marginal as shown here:

http://www.geforce.com/whats-new/gu...rise-of-the-tomb-raider-anisotropic-filtering

Performance: Anisotropic Filtering has a negligible performance cost on PC, with the delta being just 1.2 frames per second between 2x and 16x. Trilinear Filtering does give you a significant speed-up, but no sane person should select the option unless they wish for the complete loss of texture fidelity game-wide.