NVIDIA Controls AIB Launch and Driver Distribution @ [H]

So we have cards coming out in the next month that has ray tracing in hardware that people are actively poo-pooing but silicon from RTG that doesn't even exist is going to out-do it? There's speculation and then there's speculation with drugs. Maybe choose your poison more wisely in the future?
 
Just whitelisted the site, and happy to do it since seeing Kyle act on poll results.

I don't have much money to give but the site's gonna need every bit of help it can get moving forward. Stick it to nvidia.
 
RT cores are their own thing AFAIK. They’ve been labeling them separate.

nVidia is using tensor cores for DLSS which is exactly how deep learning works for practically all applications but using it for games is pretty genius imo.

View attachment 100094

Bingo......hence why they said that RTX wasn't just Ray Tracing......the details are all there, in what they say and what should be obvious in what they don't say. :)
 
Just whitelisted the site, and happy to do it since seeing Kyle act on poll results.

I don't have much money to give but the site's gonna need every bit of help it can get moving forward. Stick it to nvidia.

"Stick it to Nvidia"?
In what way are you going to "stick it" to them that you feel will impact them in any notable way?
Not being sarcastic, maybe you have a unique strategy to take down a billion dollar corporation that employs some of the best people in the world, that no one has ever thought of before?
 
So we have cards coming out in the next month that has ray tracing in hardware that people are actively poo-pooing but silicon from RTG that doesn't even exist is going to out-do it? There's speculation and then there's speculation with drugs. Maybe choose your poison more wisely in the future?
If that's your take-away from what I said, you may want to read it again and not skim it? Or at least read it slower?
Otherwise, if that's seriously how it reads, I do apologize, as it wasn't my intent. I said what I did as a COMPLETE hypothetical. I was in no way implying that AMD will manage to do so. All I was doing was explaining why I couldn't see nVidia making it an open standard, as they wouldn't be able to live with themselves IF a Radeon card did end up doing it faster. They typically do everything in their power to close doors like that, which as my post went on to say, in the off chance someone finds an "open door" then they make sure to quickly slam it on everyone.
EDIT: Misinterpreted them, my bad :)

RT cores are their own thing AFAIK.

nVidia is using tensor cores for DLSS which is exactly how deep learning works for practically all applications but using it for games is pretty genius imo.
Yea I was just taking a peek at what was on Wiki about this all and it answers a number of the questions asked, provided it's accurate.

For example, it uses/can use M$'s DXR, Vulkan (WIP), or nV's own OptiX to accomplish Ray Tracing.
OptiX, which has been around since at least 2011 for Ray Tracing, supports CUDA, and even CPU when there is "no supported hardware". In other words, as the spelling would imply, it's "PhysX" for Ray Tracing. OptiX is also in the GameWorks library.
I suppose by extension, one could say that their Ray Tracing is... quasi-open? lol It's about as open as PhysX I guess? Which is like being allowed to be part of a blockbuster movie, except your part is that you sit in a trailer in a back lot and make sure a light on a switchboard doesn't go out. haha
Tensor Cores on the Desktop graphics seem to... be the Antenna for Ray Tracing?? As it's worded, it seems like the actual Ray-Tracing is done on a supercomputer and then your graphics card learns from that on how to apply it to what's being rendered. It honestly almost sounds like it's streaming... I hope it isn't the case though, since it'd be rather absurd to have your new graphics card require an internet connection to utilize Ray-Tracing :confused:
Beyond that, the Tensor Cores do seem underutilized or somewhat repurposed.
Sources: https://en.wikipedia.org/wiki/Turing_(microarchitecture)#Tensor_core; https://en.wikipedia.org/wiki/OptiX; https://en.wikipedia.org/wiki/Nvidia_RTX


Not being sarcastic, maybe you have a unique strategy to take down a billion dollar corporation that employs some of the best people in the world, that no one has ever thought of before?
Sinkholes. Right under their HQ and any offsite data center where they store source code and research backups. :pompous:
 
Last edited:
Quote "Given this, and the fact that our readers have made their voices heard (67% Against Signing / 33% For Signing), HardOCP will not be signing the NDA."

You made the right choice Kyle, being effectively gagged for 5 years is just not worth it.
It is also in a frivolous way to create an opening for Nvidia for Lawsuits to make it into the courts system where Nvidia for practical matters have unlimited funds while the defendant can go broke. Frankly what Nvidia appears to be doing makes no sense particularly any reason why they would even have to is beyond my understanding.
 
If that's your take-away from what I said, you may want to read it again and not skim it? Or at least read it slower?
Otherwise, if that's seriously how it reads, I do apologize, as it wasn't my intent. I said what I did as a COMPLETE hypothetical. I was in no way implying that AMD will manage to do so. All I was doing was explaining why I couldn't see nVidia making it an open standard, as they wouldn't be able to live with themselves IF a Radeon card did end up doing it faster. They typically do everything in their power to close doors like that, which as my post went on to say, in the off chance someone finds an "open door" then they make sure to quickly slam it on everyone.

I'm sorry you took my reply to be a personal one to you; your post was rather good at prognostication (as are most of yours that I've seen in general). Others were rather farcical and my comment was more directed to that particular crowd.
 
This is easy to navigate.. Create a second entity or subsidiary that only does video card reviews for NVidia cards and have that entity sign the agreement. If you ever enter into a spot where you are caught up in something stupid, dissolve the entity and release whatever info you want.

On a side note, I dont consider any core work product at HardOCP as being conducted by "the press" or as an act of "journalism". Yes, I have been at the site long enough to hear all the reasoning as to why you feel you *are* journalists, etc., but sticking to reviews would be fine in my book. If you feel otherwise, the above scenario could also be used to cover your bases. Granted, I am not an attorney, but I have worked in a technical capacity with numerous law firms and attorneys (even testified in court and grand jury as an expert witness) enough to know that a lawyer is never going to lose their job by telling you not to do something. I have also signed contracts that were extremely vague, told I was in violation of the agreement by doing whatever I did, then had court agree the agreement was 100% non-enforceable. If the agreement is crap, it wont stand anyone challenging it. You might think you would have a lot to lose by going to court against NVidia, but look at it from their perspective.. You think NVidia wants to actually take you to court for telling the truth about something they did?
Except they don't have to intend to win, just make it miserable and costly for you is an option for Nvidia, make an example of you. You would have to sue them to recoup losses if you could prove it. Anyways that is how it looks to me, the NDA looks utterly ridiculous how it reads.
 
Last edited:
Keep Calm 2.png
 
Except they don't have to intend to win, just make it miserable and costly for you is an option for Nvidia, make an example of you. You would have to sue them to recoup losses if you could prove it. Anyways that is how it looks to me, the NDA looks utterly ridiculous how it reads to me.

Precisely... I agree with you 100%. But you arent following my point to the end in that there is a real tangible cost to NVidia too (both monetarily and with regard to the public) and that while HardOCP does have something to lose, NVidia has MORE to lose which means HardOCP does indeed have leverage of their own. I even suggested a way to limit the liability of HardOCP even more by creating another corporate entity (how about HardNVDA?) that only does video card reviews of NVidia hardware and THATS IT.. Have that corporate entity sign the agreement with NVidia, etc. I also mentioned that a lawyer has never been fired by telling their client not to do something. So while I am sure HardOCP's lawyer is good, he is never going to be wrong by saying "no", so lawyers will basically always say "no" if you ask them anything about what you should do in a legal scenario involving risk.
 
GN is a reputable review and hardware site. My point is that lawyers opinions differ on the NDA, as GN took the time back in June to look into it. I imagine that an NDA from AMD is pretty standard and vague as well. Not that I have seen one, I just can't imagine NDAs to be all that different.
One lawyer is an IP lawyer and shares legal opinion of some here including myself and far more experienced people than me. The other lawyer isn't an IP lawyer and is advising a customer who has signed the nda on how to not get sued.
Of course their opinion will differ. Jesus is great but no longer independent nor impartial on such matters ( by contract) and now cannot be totally trusted on nvidia issues.
Yes their simple review format is different but they still can't take a big shit on ngreedia.
 
Precisely... I agree with you 100%. But you arent following my point to the end in that there is a real tangible cost to NVidia too (both monetarily and with regard to the public) and that while HardOCP does have something to lose, NVidia has MORE to lose which means HardOCP does indeed have leverage of their own. I even suggested a way to limit the liability of HardOCP even more by creating another corporate entity (how about HardNVDA?) that only does video card reviews of NVidia hardware and THATS IT.. Have that corporate entity sign the agreement with NVidia, etc. I also mentioned that a lawyer has never been fired by telling their client not to do something. So while I am sure HardOCP's lawyer is good, he is never going to be wrong by saying "no", so lawyers will basically always say "no" if you ask them anything about what you should do in a legal scenario involving risk.
The easier answer appears to be just don't sign a questionable NDA that may impede what you do. Plus if a separate entity comes into existence, plus all the accounting, taxes and time required to be a legal entity - Nvidia may decide just not to sample anything to ya.
 
How do you explain all the others signing the NDA but only this site having a problem with it? I didn't quite understand that part.
 
If that's your take-away from what I said, you may want to read it again and not skim it? Or at least read it slower?
Otherwise, if that's seriously how it reads, I do apologize, as it wasn't my intent. I said what I did as a COMPLETE hypothetical. I was in no way implying that AMD will manage to do so. All I was doing was explaining why I couldn't see nVidia making it an open standard, as they wouldn't be able to live with themselves IF a Radeon card did end up doing it faster. They typically do everything in their power to close doors like that, which as my post went on to say, in the off chance someone finds an "open door" then they make sure to quickly slam it on everyone.
EDIT: Misinterpreted them, my bad :)


Yea I was just taking a peek at what was on Wiki about this all and it answers a number of the questions asked, provided it's accurate.

For example, it uses/can use M$'s DXR, Vulkan (WIP), or nV's own OptiX to accomplish Ray Tracing.
OptiX, which has been around since at least 2011 for Ray Tracing, supports CUDA, and even CPU when there is "no supported hardware". In other words, as the spelling would imply, it's "PhysX" for Ray Tracing. OptiX is also in the GameWorks library.
I suppose by extension, one could say that their Ray Tracing is... quasi-open? lol It's about as open as PhysX I guess? Which is like being allowed to be part of a blockbuster movie, except your part is that you sit in a trailer in a back lot and make sure a light on a switchboard doesn't go out. haha
Tensor Cores on the Desktop graphics seem to... be the Antenna for Ray Tracing?? As it's worded, it seems like the actual Ray-Tracing is done on a supercomputer and then your graphics card learns from that on how to apply it to what's being rendered. It honestly almost sounds like it's streaming... I hope it isn't the case though, since it'd be rather absurd to have your new graphics card require an internet connection to utilize Ray-Tracing :confused:
Beyond that, the Tensor Cores do seem underutilized or somewhat repurposed.
Sources: https://en.wikipedia.org/wiki/Turing_(microarchitecture)#Tensor_core; https://en.wikipedia.org/wiki/OptiX; https://en.wikipedia.org/wiki/Nvidia_RTX



Sinkholes. Right under their HQ and any offsite data center where they store source code and research backups. :pompous:

I’ll have to check out those wikis later. It didn’t even occur to me that might be a good place for info.

DLSS which uses the tensor cores for AI AA requires nVidia to use a super computer. It basically compares crappy images to perfect “ground truth” images from a game. It keeps iterating it’s code until it can predict the perfect image from the lesser. They then take that code “millions of parameters” and shoves it into the driver and your RTX card’s “neural net” uses it to upscale your games. Jensen used the example of rendering at 1440p and outputting 4k. I think it’ll work well everyone is just curious of the IQ. If you’re bored here’s the part of the keynote.
https://www.twitch.tv/videos/299680425?t=9816s

Ray tracing is more game side implemented. So we have one tech that requires nVidia and the other the game producers. Not exactly ideal but I have highb hopes since nVidia committed a decent portion of their die space.

I wonder which ray tracing tech the game producers are actually using.
 
Good on [H] for sticking to their guns. Honestly, I would think a majority of the readers here aren't in a rush to buy the 20-series at inflated launch prices. Wait for launch drivers, push out the review a couple weeks after launch and I don't think anyone will be annoyed at the "lateness". I personally don't care how long it takes, i just want to read one well written and unbiased review of these dumb things, and this site is the only place anyone going to get that (and maybe gamers nexus).

Can't wait for 7nm lol
 
The easier answer appears to be just don't sign a questionable NDA that may impede what you do. Plus if a separate entity comes into existence, plus all the accounting, taxes and time required to be a legal entity - Nvidia may decide just not to sample anything to ya.
The other part is that if an average read can see right through it, then so can nVidia, and while it might be legally acceptable, it won't hurt nVidia to get slapped down in court, but it would break [H] just to defend being in the right.
 
I am a long time lurker here and joined the forum just yesterday mainly to say kudos to Kyle and all the readers who voted NO to this shitty NDA. Well done all and no one will lose if a website reviews a GPU a week after it goes on sale with public drivers and not obligations to the vendor as they bought it. More trustworthy tech journalism is what we all need. Except for fanboys and their beloved companies. Tbh, I like AMD more than nVidia and Intel just because they are the small ones and it makes them more customer-friendly atm. If tables turn around I will go to the most customer-friendly company then. If that thing doesn't exist, used products will be enough to me not awarding the bully companies for their attitute.
 
  • Like
Reactions: Meeho
like this
Well, a major German website is refuting all claims that Nvidia is acting bad here. Apparently AMD does exactly the same and controls who get's what. The main reason for that is apparently that no AIB has enough samples for everyone, so Nvidia and AMD are working like a controller to make sure there's no chaos and most media will get a card from someone. That sounds VERY different from the claims made here.

http://www.pcgameshardware.de/Gefor...rteilung-Boardpartner-Custom-Designs-1264075/
 
Well, a major German website is refuting all claims that Nvidia is acting bad here. Apparently AMD does exactly the same and controls who get's what. The main reason for that is apparently that no AIB has enough samples for everyone, so Nvidia and AMD are working like a controller to make sure there's no chaos and most media will get a card from someone. That sounds VERY different from the claims made here.

http://www.pcgameshardware.de/Gefor...rteilung-Boardpartner-Custom-Designs-1264075/
I'm working from a translated version, but nothing there refutes Kyle's claims. There are some generalized statements of similar practices, but nothing substantial regarding AIB cucking on the AMD side and nothing about the new driver distribution control.
 
The easier answer appears to be just don't sign a questionable NDA that may impede what you do. Plus if a separate entity comes into existence, plus all the accounting, taxes and time required to be a legal entity - Nvidia may decide just not to sample anything to ya.

Yes.. Correct again.. So why complain about a document someone on their own accord has refused to sign? If you dont want to sign because of liability, there are always ways to limit exposure. Deciding not to lower the liability then complaining you cant sign due to liability makes no sense.
 
I'm working from a translated version, but nothing there refutes Kyle's claims. There are some generalized statements of similar practices, but nothing substantial regarding AIB cucking on the AMD side and nothing about the new driver distribution control.

They say Nvidia and AMD have been and are controlling the distribution of review samples because no single AIB can give samples to all testers, so with control by Nvidia and AMD, more testers can get at least one - and all this is only valid during the NDA. Later, all AIB can send samples as they wish. Also drivers always come from AMD and Nvidia for review samples, according to PCGH. So there is no real difference there, either. They also refer to the fact that HardOCP did not sign the NDA and that AMD and Nvidia boycott media that are known to have broken a NDA - though they are not claiming that HardOCP ever did this.

Also, just to make my personal stance clear: F*** Nvidia and their control-freak and anti-competetive behavior. We need great AMD graphics cards, because Nvidia needs a hard kick in the butt.
 
  • Like
Reactions: GHRTW
like this
I believe the Tensor cores are just glorified FP64 units that can support 4 compute queues per Tensor depending on the power of two equation 16+16+16+16 or 32+32 calculation grid for example. I read the Tesla whitepaper which talks about those Tensor cores and in reality, the resemblance is striking. I could be wrong though.
 
...is an extension of DirectX. Meaning that AMD, Intel, and whoever else that implements needed DirectX features can run Gameworks middleware.

AFAIK Microsoft does not allow extensions on DX11 or DXCap like they did on DX9. So Gameworks is just a middleware developed by nVidia using DX11 features and pack them into libraries for easy access and implementation. Microsoft does use feature extensions like DX11.1, DX11.2 etc but its done by them and implemented by the hardware vendor, on the DX9 was the other way around where the hardware vendor would create new features and implementations that then Microsoft would have to adapt and implement on the API (Reference 2.0, nVidia SM 2.0a, AMD SM2.0b, reference 3.0) Good times lol
 
Kyle, could you comment on the comparison of AMD and NVidia NDAs?
Talking about NDAs as a whole in the industry are usually what I consider "embargo" NDAs. These NDAs are product specific and basically say, "We will give you access to the hardware, the driver, and technical documentation about Product X, and you are not to go public with any information till this date(s)." Very basic so that sites with access coordinate around a product launch date. Really that simple. Much of AMD's NDAs lately have been a simple email asking if we agree to the embargo date. It is fairly mundane and everyone understands that they just don't want benchmarks or documentation released ahead of the launch date.
 
First, this is exactly what I feared would be the direction they took, and it's happened...so, yeah. Awesome.
Could be coincidental, but probably some pseudo psychology parallel to today's political climate but ANYWAY...., and any entity that tries to influence or control a citizen's opinion or access to information, is in the wrong.

They give the facts, we should come to our own conclusions.

Second, the 'feeling' is that the 2080 performance hype is going to fail, neither here nor there IMO & it is what it is until the cards come out, HOWEVER....

If the performance is at a point where we see it being 'justified' and acceptable......how many are going to forget about this whole NDA thing?

Could be part of their evil master plan...aha.
It wouldn't be the other way around, for actual performance is not going to go away...

This is all a very weird situation, does not align with decades of expectations.
 
Thanks for that explanation.. I'm seeing a few sites claiming that AMD has been using the same sort of controls on hardware and AIB drivers and such.. here's the one from DVH..

https://www.dvhardware.net/article69457.html

From the article..
Well first of all, "AMD does it too," is not much of an argument as they are not laying out facts in how all this happens.

When AMD runs GPU launches, it many time invites AIBs to have their cards be part of the launch when it has not produced its own cards. It is AMD's launch, not the AIBs. I have never seen AMD use the driver controls put in place either. That said, recently AMD has built its own cards for launches, so AIB cards were not included.

What is different here is that AIBs are handling its launches for its cards. NVIDIA is dictating to the AIBs who and more importantly who cannot be sampled directly by the AIBs. And then NVIDIA is requiring the AIBs to identify who is getting the AIB cards, asking for names, phone numbers, and emails, so that it can control who gets the driver, most likely through a secured GeForce Experience site. These are two very different things. If those people are not able to comprehend that, and the only thing they have to say is "AMD does it too," then they do not understand what exactly is going on.
 
This better be a wake up call to those AIB partners that Nvidia gives zero fucks about them. The writing has been on the wall ever since Nvidia launched their own founders edition of their cards. Nvidia is going to cut out the middle man if they can, they have the money and resources to do so if they wanted.
 
I think you misunderstood my point. Not trying to get you to be a Social Justice Warrior! :p

Just do it for yourself, and maybe that German site. Flaunt it...
Thanks, but no thanks.
 
Interesting video on the effect of FE cards and MSRP leading to price hikes since maxwell.

 
Getting PREVIEWS of anything is a privilege. No one is stopping anyone from reviewing the cards with official drivers once they are on the actual market. Think about what this article is really saying: "I deserve the right to get a sneak peak". A launch is a company's own party and they should be welcome to invite anyone they want. If they want an exclusive soiree that makes us all resentful, that's their call.

Now, I do think that being exclusive and pricey and cagey are not the greatest of behaviors and can breed suspicion and resentment. I totally understand that. But people gotta put their outrage in perspective. Having preview access to the launch of a luxury item is a privilege -- simple as that.

And I want to emphasize that these are a LUXURY item. No one needs one. There is no rush to buy one. NVIDIA has the right to put an asking price of a million dollars if they want. The only thing making people all anxious is vanity and greed -- people are literally whining about the fact that something that didn't exist a year ago is not that affordable today! First world problem...
 
How do you explain all the others signing the NDA but only this site having a problem with it? I didn't quite understand that part.

The real issue is about the nature of "news" and "reporting". Except for unpaid social warriors, most reporting outlets are meant to make money which automatically starts creating conflict of interest. For example, maybe your business model is to simply make click-bait and then will emphasize bias in your information to attract interest/reaction. Or maybe you want to be honest reviewer but get privileges like special access to previews and behind the scenes -- it becomes suspicious once you're all chummy and pampered by the vendor you're reviewing.

In the case of people who really want to be honest they fall into the category of investigative reporter. However, what vendor, celebrity, politician, etc. is ever going to give an investigative reporter early, unfettered access to information?

If you're cynical you could say that people who don't like investigative reporters are those that have something to hide. But that is like saying you don't need a lawyer if you're innocent -- the reality is you can lose the control of the narrative which can have dire consequences.

So in this case of NVIDIA's you could say "what are they hiding?" but I look at it as I understand why someone wants control over the narrative -- that is just prudent.

You asked why some people sign an NDA that others won't -- well some news outlets only want to make money by repackaging pre-canned marketing material, some news outlets want to be trusted but are willing to generally to stick to the high level stuff without a lot of hard questions, some people want to be investigative journalists who can't have anything restricting what they want to say. Many categories would just be fine with an NDA, but not the investigative journalist.

So basically, a company should not let an investigative journalist into their inner circle and also an investigative journalist should not sign almost any NDA.
 
Getting PREVIEWS of anything is a privilege. No one is stopping anyone from reviewing the cards with official drivers once they are on the actual market. Think about what this article is really saying: "I deserve the right to get a sneak peak". A launch is a company's own party and they should be welcome to invite anyone they want. If they want an exclusive soiree that makes us all resentful, that's their call.

Now, I do think that being exclusive and pricey and cagey are not the greatest of behaviors and can breed suspicion and resentment. I totally understand that. But people gotta put their outrage in perspective. Having preview access to the launch of a luxury item is a privilege -- simple as that.

And I want to emphasize that these are a LUXURY item. No one needs one. There is no rush to buy one. NVIDIA has the right to put an asking price of a million dollars if they want. The only thing making people all anxious is vanity and greed -- people are literally whining about the fact that something that didn't exist a year ago is not that affordable today! First world problem...
I am not outraged about anything. I knew when I ran the GPP stories I would be cut off. I was hugely surprised when NVIDIA offered access, pending me signing its 5 year NDA. I made my bed and I slept in it. No biggy. We have already spent $4300 on 2080 cards for review.

That all said, I do think it is important for people to understand just how the system under NVIDIA is changing. Go back and read everything I have written about this, I have not shown any outrage or butthurt. It is what it is, and our readers should know the truth about what is going on behind the scenes when it comes to reviewing video cards on day one.
 
I wonder what it would take for the SEC to launch a probe into NVIDIA. Or maybe this is something more up the FTCs alley. After all, this isn't exactly Apple, so they might not get gentle treatmebt if caught in either of these departments crosshairs.

Anti-consumer behaviour, stock price manipulation by paper-launching a new product, forcimg their AIBs to eat 10-series inventory to bolster revenue numbers.

Maybe one of the best and brightest here can answer the following questions:

Is NVIDIAs behaviour outright illegal or just unethical? Where do we draw the line?
 
Already did, bought a Vega 64 last month, couldn't be happier with it, and it's performance!
I have bought 3. But quite abit longer ago. One is in my Mac Pro. I changed the thermal compound on all 3 cards and performance improved immediately. They had this ultrathick crumbly grey paste that was way to thick. My thermals improved quite abit and it stays clocked higher. I also upped my fan speeds by default abit and now they really run well. I considered water but I don't think the cost is worth it given how well they are running. Just a thought :)
 
The real solution to this is for AMD to up their game and get in nvidias face at the top end. They need to do the nvidia what they're doing to Intel.

At 4K (which is what I run at) the best AMD offering is on par with the 1080 while costing a little bit more (Asus Strix on amazon, R64 is $50 more than 1080). They tend to trade places by a few FPS depending on the title.

However, for me, the big thing is timely game support. AMD really lost me a long time ago in the days waiting months and months for driver updates for various titles and eternal beta drivers.

I do remember when AMD thought this longer driver delay would work with delivering higher quality drivers but waiting inbetwen. That didn't work out and they have long abandoned that plan. We get drivers quite regularly and game release drivers. Just a heads up that did change.
 
As someone who once had enough disposable income to just get the toys I wanted, now I have to be a lot more careful.

It is still fun, mind you. Makes buying decision more interesting and challenging.

My personal opinion/guess, is that NVidia and AIBs have so much 10 series stock left over, they have nothing to loose.

To me, all it means is that I might just need to get a 1080Ti, instead of a 2080Ti at whatever price. Or worse, keep my Vega 64 for a bit longer. Neither of those options are going to make me cry too much over my pillow.
 
Back
Top