The thread which sometimes talks about RDNA2

[EastCoastHandle]

They are reacting different because Jensen knows how to sell his stuff. They hear from influencers that DLSS is better than native or that upscaling from 1080p to 4k is good enough and they become convinced that it is true. Same for the RT they really believe they will play full real time path traced games, they don't understand it is a feature Jensen will exploit for several generations of cards, then he will move to another feature.

 

[sprite-]

If you think I'm trolling you it's only because I don't care about how you feel. I provide my opinion, give facts and provide insight as it relates to the discussion. For example the contravsory of Nvidia's launch.
https://www.overclockers.co.uk/foru...-event-thread.18904106/page-242#post-34227443

Your conversation about stock levels are irrelevant. As I stated that the release difference between Ampere and RDNA2 was September 17. The release date for 6000 series was November 18. At that time Nvidia had a head start and people were still waiting in line for cards. I've also posted:

https://www.overclockers.co.uk/forums/posts/34269835
So that make what you posted a lie. As I suspect that prices will should normalize next year once stock levels normalize.

Are you ok down there? You've dug yourself a pretty deep ditch.

Yup, you use the "i don't care" line allot aswell. I highlighted that one as you also used that in plenty of your posts. 2 pages of results, all from arguing with people about Nvidia / AMD. You don't care so much that I'm beginning to think you do care, as I said before. You're just some cuddley carebear. You're nothing if not predictable lol .

You slammed Nvidia on launch day for their 3080 stock availability, yet AMD launch comes along and...

1. At first you claim blue in the face that the launch wasn't as bad, until the retailers started publishing statements and you really couldn't keep pushing that line without looking silly.

2. You go "ill judge it by end of Q1 2021" which is odd, since you ran to judgement so quickly on launch day with Nvidia. Which is a pretty transparent way of confirming your personal bias?

 

[nvidiamd]

Again that stupid legend with Nvidia sharing their tech with MS. Dude you people believe everything you hear or read if it's something good about Nvidia.
I like to play a game as the artist made it, i don't want to rely on AI upscaling, especially if i pay a ton of money for a last gen videocard. I understand AI upscaling can be useful ( for example on consoles after some years ) but i don't understand how people who paid $1000-1500 for their cards can be happy not playing a game at native resolutions.

The main purpose of RT and DLSS is to control how fast a videocard will age and to be a marketing tool, not to give you great games. Nvidia is one of the reasons for why Cyberpunk was launched now in the current state, because it was sponsored to be an asset for the 3000 series. The other reason was CDPR greed but that is another story.

Think about it. Look here in the Control footage someone posted earlier. Is that real RT? Is that realism/immersion? Or it is some bling that Nvidia paid for to sell more cards?

 

[EastCoastHandle]

Yup, you use the "i don't care" line allot aswell. I highlighted that one as you also used that in plenty of your posts. 2 pages of results, all from arguing with people about Nvidia / AMD. You don't care so much that I'm beginning to think you do care, as I said before. You're just some cuddley carebear. You're nothing if not predictable lol .

You slammed Nvidia on launch day for their 3080 stock availability, yet AMD launch comes along and...

1. At first you claim blue in the face that the launch wasn't as bad, until the retailers started publishing statements and you really couldn't keep pushing that line without looking silly.

2. You go "ill judge it by end of Q1 2021" which is odd, since you ran to judgement so quickly on launch day with Nvidia. Which is a pretty transparent way of confirming your personal bias?

I just want to make sure we are clear on our communication . But you are replying to my posts. Not the other way around. That is the important part to this that you missed, going back to being triggered.

It's OK he's just a bit insecure and troll like.

He ran away from me in a previous debate when I highlighted that if you look through his forum history he'd used the same jibe multiple times (incidentally, the same thing he's doing now, the "you mad bro?" approach), made him look a little silly.

Now that's I've proven you wrong, corrected your lies and overall see that you don't know what you are talking about you use Ad Hominen. So I do ask, "You mad bro?"
Thanks for the laugh.

--------------------
-------------------
I still wonder if we will see a toxic 6000 series variant though. On water 6800xt can hit and sustain a 2700MHz+ OC. That's quite impressive feat if you ask me. But how do you get that from air cooled variant as you have to keep both the vregs/mosfets just as cooled as you do the die and vram ICs.

I haven't seen many benchmarks with OC's that high yet though. But I do believe, if the drivers are optimized for that suites of games benchmark it should be on par, if not better then a 3090.
Which makes the choice of which brand of card to buy? Go with a Nitro+ or go OEM and watercool it?

There are a few tricks one can use if they get a HSF 6000 series variant. Which requires re-pasting and applying 1 Fujiploy thermal pads 17mw/k (you have to check on thickness .5/1.0/1.5mm). As the thermal pads used on those cards aren't rated that high thermally. For some reason some cards have a series of dots formulating a square dab of paste. Which isn't very effective method IMO. That helps reduce thermals a bit more which can stabilize higher clock rates.

 

[sprite-]

Again that stupid legend with Nvidia sharing their tech with MS. Dude you people believe everything you hear or read if it's something good about Nvidia.

Stupid legend? They've jointly published several articles with Microsoft, also presented at events with them talking about both DirectML and DLSS.... Not to mention the approach mirrors Nvidias entirely... Mainly because its in their own interest, they know full well DirectML will become the defacto standard used across consoles and PCs. Nvidia are ensuring that they can still offer a competitive product by DirectXL services calling the DLSS architecture when an Nvidia card is used and sell their training models as better quality.

As for the other stuff you said. You won't have a choice, DLSS / DirectML will just be considered part of the rendering pipeline for all games eventually. As for "as artist made it", that's entirely the point. The training models will become so accurate, they will be.

Think you need to strap in bud, because AI training models are being used for everything from games, identifying cancer or even mapping genomes to create medicine. It's accuracy that matters, once it hits a point where predictability is equal to or higher than what is observable or below human rate of error... Its the standard.

 

[sprite-]

I just want to make sure we are clear on our communication . But you are replying to my posts. Not the other way around. That is the important part to this that you missed, going back to being triggered.

Now that's I've proven you wrong, corrected your lies and overall see that you don't know what you are talking about you use Ad Hominen. So I do ask, "You mad bro?"
Thanks for the laugh.

It's always fun when you reach a point with someone by presenting the facts that they're unable to dispute them so choose not to acknowledge them at all. Bit like a flat earther

Which colour are you?

 

[EastCoastHandle]

It's always fun when you reach a point with someone by presenting the facts that they're unable to dispute them so choose not to acknowledge them at all. Bit like a flat earther

Which colour are you?

Haaa!
So you've tired yourself out. Always a pleasure.

 

[sprite-]

Haaa!
So you've tired yourself out. Always a pleasure.

Tired? It's hard to continue a debate when someone refuses to respond to the facts because it makes them look bad. It's essentially like running away but sticking their fingers in their ears going "lalalalala". Poor show, if you can't compete don't get into the ring little one.

 

[EastCoastHandle]

Tired? It's hard to continue a debate when someone refuses to respond to the facts because it makes them look bad. It's essentially like running away but sticking their fingers in their ears going "lalalalala". Poor show, if you can't compete don't get into the ring little one.

Hahahah!!!!!
Looks like someone needs a nap. You cant even mustard up a descent response to nvidiamd counter points.

 

[sprite-]

Hahahah!!!!!
Looks like someone needs a nap.

There you go, running away again bud. You need to hang in there, respond to the information presented to you and fight your case. Chin up laddie.

 

[EastCoastHandle]

There you go, running away again bud. You need to hang in there, respond to the information presented to you and fight your case. Chin up laddie.

It's ok, craddle yourself to whatever you believe to help you cope. I've already addressed you.

 

[sprite-]

It's ok, craddle yourself to whatever you believe to help you cope. I've already addressed you.

Looks.... Nope... Can't see anywhere where you've addressed your inconsistency in judging the Nvidia and AMD launches, or why you use the same handful of debate mechanisms (as proven by several pages of search results).

Feel free to take a stab at it though.

 

[Richeh]

@sprite- just don't bother. It is the RNDA2 thread so more of an echo chamber is to be expected. I took annoyance to him trolling the CP2077 thread as he was just posting stuff nothing to do with the games graphics, but here, it's a lose lose.

 

[nvidiamd]

Dude it is one thing to say they were or are working with MS because as you said, Direct ML will become the default AI upscaling. And another thing to claim they shared their modeling approach with them. Both Nvidia and AMD keep close to Microsoft and i've seen AMD fanboys in the past claiming that Direct ML is the result of AMD work with Microsoft.
I know it is the future because it is much cheaper than building more powerful cards ( just as i know heavy RT is not the future for the same reason ) but i also hope there will be enough fools like me who will still appreciate native rendering. Even when we will hear 24/7 that AI upscaling is better than native.

 

[sprite-]

Dude it is one thing to say they were or are working with MS because as you said, Direct ML will become the default AI upscaling. And another thing to claim they shared their modeling approach with them. Both Nvidia and AMD keep close to Microsoft and i've seen AMD fanboys in the past claiming that Direct ML is the result of AMD work with Microsoft.
I know it is the future because it is much cheaper than building more powerful cards ( just as i know heavy RT is not the future for the same reason ) but i also hope there will be enough fools like me who will still appreciate native rendering. Even when we will hear 24/7 that AI upscaling is better than native.

Literally a video here, where Microsoft at end of the presentation talk (which they did so partnered with Nvidia) about models shared with them by Nvidia and used it with a Titan V with DirectML. Also a white paper published on it after the event. Microsoft are tied in with Nvidia significantly on AI due to their enterprise partnerships, so it's not surprising.

http://on-demand.gputechconf.com/si...-gpu-inferencing-directml-and-directx-12.html

Nvidia aren't stupid. They want the entire industry to move in the direction they're going, where they sacrifice raw gaming architecture on their GPUs for implementation of more complex AI cores and workloads. Especially as this is a major strength of theirs and isn't for AMD. They can only ensure that happens through partnerships. Sure, they don't give away the farm. They'll still wanna slap a "Use Nvidia and get better quality AI upscaling" but they'll move it along enough to make it a standard.

 

[CAT-THE-FIFTH]

So anything new about RDNA2?? Or is become the battle of people justifying whatever GPUs and games they like?? Maybe the thread title needs a change. I can do that because I am certainly not reading much about RDNA2 based GPUs,or anything remotely related here.

Edit!!

If people want to drone on about Cyberpunk 2077 the there is already a large thread about here on this subforum:
https://www.overclockers.co.uk/forums/threads/cyberpunk-2077-ultra-performance.18910855/page-53

 

[EastCoastHandle]

So anything new about RDNA2?? Or is become the battle of people justifying whatever GPUs and games they like?? Maybe the thread title needs a change. I can do that because I am certainly not reading much about RDNA2 based GPUs,or anything remotely related here.

Edit!!

If people want to drone on about Cyberpunk 2077 the there is already a large thread about here on this subforum:
https://www.overclockers.co.uk/forums/threads/cyberpunk-2077-ultra-performance.18910855/page-53

 

[Curlyriff]

I would say a couple things in CDPR's defence, since we've moved on from just RT.

1. I notice GTA 4 was used in this comparison. In many ways, GTA 5 was a regression from GTA 4. The collision physics was much dumbed down in 5, and your vehicle became almost indestructible (like CP). GTA 4 had much better physics, handling, etc. Both in terms of realism and fun (imho).
2. GTA 4 was Rockstar's 6th (or so) attempt at an open world driving game. This is CDPR's first attempt.

OTOH, hasn't CP2077 been in development for 10 years... so far longer than GTA 4.

OTOOH, Rockstar had over 1,000 people working on GTA 4 according to Wiki. That's probably more than CDPR.

I <3 GTA4. I absolutely hated GTA5 tho. Worse in almost every way.

Just to give figures for CP2077 it has had 650 developers working on it not including any third parties/contract workers. Solid working was closer to 4 years with a small team working on it since around 2010 in terms of getting the IP brought and setup and getting some basic stuff into engine to show premise. With that be it GTA4 or GTA5, both have better physics than CP2077 in my view and the interaction generally in city and with NPC's are better. The vehicle destruction I don't mind so much as it is 2077, the cars could be built from different materials and such but they lack animations, sounds, ragdoll and basic AI stuff to make the game feel alive tbh. It just kinda feels dead with almost zombies walking about in it.

 

[xcx xcxvgyt]

On water 6800xt can hit and sustain a 2700MHz+ OC. That's quite impressive feat if you ask me. But how do you get that from air cooled variant as you have to keep both the vregs/mosfets just as cooled as you do the die and vram ICs.

No need water or extra cooling with good binned 6800xt.

Also this is how perform against 3090 at valhalla with Sam

 

[nvidiamd]

Literally a video here, where Microsoft at end of the presentation talk (which they did so partnered with Nvidia) about models shared with them by Nvidia and used it with a Titan V with DirectML. Also a white paper published on it after the event. Microsoft are tied in with Nvidia significantly on AI due to their enterprise partnerships, so it's not surprising.

http://on-demand.gputechconf.com/si...-gpu-inferencing-directml-and-directx-12.html

Nvidia aren't stupid. They want the entire industry to move in the direction they're going, where they sacrifice raw gaming architecture on their GPUs for implementation of more complex AI cores and workloads. Especially as this is a major strength of theirs and isn't for AMD. They can only ensure that happens through partnerships. Sure, they don't give away the farm. They'll still wanna slap a "Use Nvidia and get better quality AI upscaling" but they'll move it along enough to make it a standard.

I think you are confusing Tensorflow with Nvidia tensors. Just because they run some of their models on an Nvidia Titan or a CPU it doesn't mean Nvidia shared something with them. They can run the same models on a Radeon on an Intel GPU.
The tech goes from Microsoft to AMD, Nvidia and Intel and not the other way around. You can find enough papers where Microsoft thanks AMD or Intel for their colaboration too, that doesn't mean anything.

CohhCarnage however has a decent Youtube playthrough recorded from Twitch, but again it's hard to appreciate the RT effects unless you are in control. He's using 1440p, everything maxed and DLSS Quality also. I like to enjoy the game first and watch some Youtube afterwards to see what I missed or even a different play style.

Personaly I play steathy using non lethal take downs, before gathering my victims together, shooting them in the head and dumping the bodies in the refuse bins or even the available fridge. Cohh takes a more direct approach at times. There is exciting action from 20:30 onwards.

Thanks for that i noticed he doesn't recommend Cyberpunk unless people understand what they are getting into ( he is a big fan though ). But he recommends this game:

At full price.