Will AMD release any high performance GPU's in Q2 or Q3?

[Fascinate]

No, its not possible lol.

If you have a FE next to a vega rx with the same drivers and they are boosting to the same clocks, the ONLY possibilities are the rx vega ties it, or is slower.

[Shinzai]

That's not the question lol.

Again, the question is how do people think the rx vega can be a faster card than the FE. Of course when having this dicussion you would have to assume the same drivers.

Many people try to make this argument that its "not a gaming card", well the titan is marketed as a deep learning card and smokes the gtx 1080.

The 1080 is a different card entirely from the Titan X(P). The 1080ti is the comparable card here. Your argument is rendered irrelevant by this fact. The 1080 is a GP104 chip, the 1080ti and the Titan X(P) are both GP102 variants.

The GP102 is in a whole different ballpark to the GP104. What AMD are releasing is their GP102 equivalent - the FE and RX variants. The FE compares directly to the Titan X and the RX will compare directly to the 1080ti if you want a 1:1 comparison. However, I've no doubt that the pricing of the RX will put it in the same price bracket as the 1080 vanilla.

So the RX Vega being above the 1080 shouldn't really be any surprise, given that the FE variant barely falls behind a 1080 in the benchmarks referenced. It would be easy for AMD to tweak the RX heavily for gaming, the same way the 1080ti is tweaked.

[Fascinate]

The 1080 is a different card entirely from the Titan X(P). The 1080ti is the comparable card here. Your argument is rendered irrelevant by this fact. The 1080 is a GP104 chip, the 1080ti and the Titan X(P) are both GP102 variants.

The GP102 is in a whole different ballpark to the GP104. What AMD are releasing is their GP102 equivalent - the FE and RX variants. The FE compares directly to the Titan X and the RX will compare directly to the 1080ti if you want a 1:1 comparison. However, I've no doubt that the pricing of the RX will put it in the same price bracket as the 1080 vanilla.

So the RX Vega being above the 1080 shouldn't really be any surprise, given that the FE variant barely falls behind a 1080 in the benchmarks referenced. It would be easy for AMD to tweak the RX heavily for gaming, the same way the 1080ti is tweaked.

That isn't what is being discussed.

If the rx vega gets tweaked to be faster than the 1080 (which it wont) so will the FE.

People need to stop suggesting the rx vega can be faster card than the FE, because that is a literal impossibility given same clocks/drivers.

And of course i know the titan uses a different chip than the 1080, i only used that because people like to use the "not a gaming card argument" for vega FE. Titan has never been marketed as a gaming card, even tho it has gtx badging.

[Shinzai]

That isn't what is being discussed.

If the rx vega gets tweaked to be faster than the 1080 (which it wont) so will the FE.

People need to stop suggesting the rx vega can be faster card than the FE, because that is a literal impossibility given same clocks/drivers.

And of course i know the titan uses a different chip than the 1080, i only used that because people like to use the "not a gaming card argument" for vega FE. Titan has never been marketed as a gaming card, even tho it has gtx badging.

OK. Simple question: Is the 1080ti faster than the Titan X(P)? Yes, or no?

If yes, then the RX can be faster too. It's that simple.

Addendum:

Don't even attempt to pull the "Titan X wasn't marketed as a gaming card" bullshit.

http://www.geforce.co.uk/hardware/10...-x/#redirected

3 main features of the card as per that site:

"UP TO
3X
FASTER PERFORMANCE

LATEST
GAMING
TECHNOLOGIES

NEXT-GEN
VR
EXPERIENCES"

[Fascinate]

OK. Simple question: Is the 1080ti faster than the Titan X(P)? Yes, or no?

If yes, then the RX can be faster too. It's that simple.

No actually, it isnt.

There are two versions of the titan xp, bet you didnt know that.

Psst @Shinzai:
https://www.youtube.com/watch?v=EYlMrxIgoME#t=4m39s

[Shinzai]

No actually, it isnt.

There are two versions of the titan xp, bet you didnt know that.

Psst @Shinzai:
https://www.youtube.com/watch?v=EYlMrxIgoME#t=4m39s

To paraphrase what he just said: "The 1080ti was faster than the Titan X, until they made a new Titan X.". So what's your point?

[Fascinate]

To paraphrase what he just said: "The 1080ti was faster than the Titan X, until they made a new Titan X.". So what's your point?

Two things:

1. Again my 1080 vs titan comparison is not an apples to apples to vega FE and rx vega, i made this point clear. It was to show the marketing of the cards and to combat when people say the vega FE "isnt a gaming card". You tried calling me out on this and that the titan was marketed as a gaming card, i linked the exact point in the video where nvidia told jay he wasnt going to be getting these card because they are "not gtx gaming graphics cards".

2. You didn't know nvidia came out with a refreshed titan xp and thought the fact that the older titan xp losing to the 1080ti was some sort of proof that the the rx vega had a chance to be faster than the vega FE. There is another titan xp....

[Shinzai]

Two things:

1. Again my 1080 vs titan comparison is not an apples to apples to vega FE and rx vega, i made this point clear. It was to show the marketing of the cards and to combat when people say the vega FE "isnt a gaming card". You tried calling me out on this and that the titan was marketed as a gaming card, i linked the exact point in the video where nvidia told jay he wasnt going to be getting these card because they are "not gtx gaming graphics cards".

2. You didn't know nvidia came out with a refreshed titan xp and thought the fact that the older titan xp losing to the 1080ti was some sort of proof that the the rx vega had a chance to be faster than the vega FE. There is another titan xp....

... I specifically used the X(P) to specify the early version of the Titan X Pascal. Which is called Titan X, which is the same name as the Maxwell Titan X. If I was going to talk about the new one, I'd say Titan XP.

The 1080ti is faster than the Titan X(P). Fact.
The Titan XP, which came out after the 1080ti is faster again. Fact.
If they made a 1080 tiP, it would be faster than the Titan XP. Conjecture, but accurate given the past decade or so of 'ti's vs Titans.

Your whole point, or attempt at a point is that you can't make an architecture faster just by tweaking it, correct? And yet, you've pointed out that Nvidia has done this twice with a single architecture, so again, what is your point?

[Fascinate]

Wow now you are really trying to twist things arent you?

Just admit you had no idea nvidia made two versions of the titan xp, nvidia never marketed either of them as gaming cards, and cant even grasp the notion that the only possibility rx vega has to be faster than vega FE is with higher clocks (which simply comes down to cooling solution).

People really cannot admit defeat on a forum lol, does it bruise your ego too much?

[Shinzai]

Wow now you are really trying to twist things arent you?

Just admit you had no idea nvidia made two versions of the titan xp, nvidia never marketed either of them as gaming cards, and cant even grasp the notion that the only possibility rx vega has to be faster than vega FE is with higher clocks (which simply comes down to cooling solution).

People really cannot admit defeat on a forum lol, does it bruise your ego too much?

Please explain what exactly I'm twisting?

You're trying to suggest that it's only the clocks that can make a difference. So you're telling me that the 1080ti outstrips the Titan X(P) because its base clock is 50mhz faster? And the Titan XP is faster than the 1080ti only because its Vram is clocked higher? That's really what you're saying?

[Fascinate]

Please explain what exactly I'm twisting?

You're trying to suggest that it's only the clocks that can make a difference. So you're telling me that the 1080ti outstrips the Titan X(P) because its base clock is 50mhz faster? And the Titan XP is faster than the 1080ti only because its Vram is clocked higher? That's really what you're saying?

Wow you really dont get it lol.

I have said MULTIPLE times my titan/1080 comparison was in regards to people saying you cant judge the gaming performance of vega based on the FE card because it "isnt a gaming card".

The fact you didnt realize there were two titan xp's is simply besides the point, and has no relevance to what is being discussed.

[Shinzai]

Wow you really dont get it lol.

No actually, it's the opposite.

The fact you didnt realize there were two titan xp's is simply besides the point, and has no relevance to what is being discussed.

So... you just stated that you trying to say about the X(P) and the XP is irrelevant?

I've known about the different Titan X's as long as they've been out. We've discussed them on this forum months ago, whether you were a part of the discussion or not. Go to videocardz.com and tell me whether the 2016 Titan X(P) is called the "Titan X (Pascal)" or not and whether the new one is called "Titan XP" or not.

I have said MULTIPLE times my titan/1080 comparison was in regards to people saying you cant judge the gaming performance of vega based on the FE card because it "isnt a gaming card".

I told you the reason for using the X(P) and the 1080ti as a more accurate "apples to apples" comparison in my first post. That hasn't changed, it won't change and has no reason to.

Just look at your own posts:

No, its not possible lol.

If you have a FE next to a vega rx with the same drivers and they are boosting to the same clocks, the ONLY possibilities are the rx vega ties it, or is slower.

No there are people that actually think that the rx vega is going to be faster than the FE...

I dont even know why you would speculate card a is faster than card b if the drivers/clocks arent normalized, whats the point?

Again the only possibilities here are the rx vega is the same as FE or its slower, there is NO CHANCE that its a faster graphics card.

THIS is what you are saying. It's blatantly false when the usage of the cores and how the data is moved around on an architecture can completely be modified and sped up for specific purposes.

If a Titan X(P) a 1080ti and a Titan XP all got benched together, using the exact same clock speeds, from what you're saying, they would all give exactly the same results. And that's bullshit.

We have a basis of comparison, a direct source to reference the differences that companies can make on a single architecture and how they can produce different and faster results. The clock speeds are NOT the only deciding factor.

[Fascinate]

It has no relevance because i was not comparing 1080 to the titan in the sense they had the same GPU's, but rather the titan was a 1000 dollar graphics card meant for the prosumer market just as the vega FE is.

You came into this conversation late it seems, many have used the "vega fe isnt a gaming card" to predict the performance of the rx models. You simply cant use that as an excuse, and i am trying to show people if we assume the GPU is the exact same on the rx vega as vega FE (best case scenario) the only positive outcome can be the aftermarket cards can hit a higher clockrate than the FE. This does not make it a faster GPU than FE, simply means it has a better cooler strapped onto it.

There is no scenario where clock for clock the rx vega is a faster card than the vega FE, that is what people need to get.

[Shinzai]

It has no relevance because i was not comparing 1080 to the titan in the sense they had the same GPU's, but rather the titan was a 1000 dollar graphics card meant for the prosumer market just as the vega FE is.

You came into this conversation late it seems, many have used the "vega fe isnt a gaming card" to predict the performance of the rx models. You simply cant use that as an excuse, and i am trying to show people if we assume the GPU is the exact same on the rx vega as vega FE (best case scenario) the only positive outcome can be the aftermarket cards can hit a higher clockrate than the FE. This does not make it a faster GPU than FE, simply means it has a better cooler strapped onto it.

There is no scenario where clock for clock the rx vega is a faster card than the vega FE, that is what people need to get.

It's not an excuse. And the FE and RX are part of the same architecture, which is why it makes 100% sense to use a 1080ti/Titan X(P) comparison.

There is every possibility that at the same clocks, the RX performs better in games, exactly the same as the ti vs X(P). Or should we then assume that things like Teslas and Quadros with higher core speeds than GTX range cards can outperform them in games? That kind of thinking is beyond basic.

[Tiberria]

I think its highly unlikely that FE vs RX gaming performance differs by any real margin. They are going to basically be the same card in the same way the Titan X and 1080 Ti are basically the same card. With AMD selling the FE as a professional card with strong gaming performance, how would it make sense for them to gimp the gaming performance of something they are charging $1000 for? It makes no sense, just like Nvidia claims the Titan cards are not targeted towards gamers but they still perform great in games. Where the FX might start outperforming the FE is when AIB vendors start making custom cards with better cooling solutions, in the same way that top AIB Nvidia cards are up to 10% better than the base reference blowers.

And, it's absolutely possible driver improvements will cause huge performance increases down the road. The question is whether AMD can get those driver optimizations in place in enough time for it to matter, before the press/review narrative has already written these cards off. They had that issue with the RX 480, with the R9 Fury, and even had it with Ryzen at launch. They need to get their shit together and realize that hardware is not just the pure hardware pieces - the driver and software side is just as important and needs to be ready for launch to have any chance of looking like they are on the Intel/Nvidia level instead of amateur hour.

[Fascinate]

It's not an excuse. And the FE and RX are part of the same architecture, which is why it makes 100% sense to use a 1080ti/Titan X(P) comparison.

There is every possibility that at the same clocks, the RX performs better in games, exactly the same as the ti vs X(P). Or should we then assume that things like Teslas and Quadros with higher core speeds than GTX range cards can outperform them in games? That kind of thinking is beyond basic.

My god dude you really just dont get it lol.

I dont disagree the 1080ti is a more valid comparison to the titanx, but even in that scenario my predictions become more true.

The 1080ti is clocked HIGHER than the original titan xp. When you clock them the same they offer identical performance, only difference being the titan had 1gb more vram.

The reason i originally said 1080 is because that is the card the rx vega is hoping to compete against, maybe i should have said 1080ti so i could have avoided your onslaught of nitpicks.

[Shinzai]

I think its highly unlikely that FE vs RX gaming performance differs by any real margin. They are going to basically be the same card in the same way the Titan X and 1080 Ti are basically the same card.

But... this statement is completely contrary to itself. You say there will be no real margin of difference, yet the X(P) vs the 1080ti shows exactly the opposite. The 1080ti was way better at gaming, with a bunch of X(P) circuitry/chips shut off or missing.

With AMD selling the FE as a professional card with strong gaming performance, how would it make sense for them to gimp the gaming performance of something they are charging $1000 for? It makes no sense, just like Nvidia claims the Titan cards are not targeted towards gamers but they still perform great in games.

Of course it makes sense. They specifically said they want a card like the Titan. They'll sell it at high prices for a few months, make some ridiculous money from early adopters, then put out the RX for everyone else. And people will pay ridiculously for new cards from their favourite brand, regardless of whether it's a good idea or not. Nvidia's proved that time and time again with the Titans always being followed by 'ti's.

My god dude you really just dont get it lol.

The 1080ti is clocked HIGHER than the original titan xp. When you clock them the same they offer identical performance, only difference being the titan had 1gb more vram.

It's 50 mhz. It doesn't add up to the difference in benchmarks. Stop being asinine.

[Fascinate]

But... this statement is completely contrary to itself. You say there will be no real margin of difference, yet the X(P) vs the 1080ti shows exactly the opposite. The 1080ti was way better at gaming, with a bunch of X(P) circuitry/chips shut off or missing.



Of course it makes sense. They specifically said they want a card like the Titan. They'll sell it at high prices for a few months, make some ridiculous money from early adopters, then put out the RX for everyone else. And people will pay ridiculously for new cards from their favourite brand, regardless of whether it's a good idea or not. Nvidia's proved that time and time again with the Titans always being followed by 'ti's.



It's 50 mhz. It doesn't add up to the difference in benchmarks. Stop being asinine.

Are you being funny with me now?

http://www.tomshardware.com/reviews/...ti,4972-8.html

They are margin of error differences with the titan taking wins in some games. Once you clock them the same its the same card minus 1gb of vram.

You are killing yourself with your own argument now, which wasnt even relevant to my 1080/titan comment.

The only way you can be this misinformed is you saw a 1080ti being benchmarked against a maxwell titan.

Surely we wont be seeing shinzai post in here again.

[Deleted]

Shinzai no matter how you slice it you are just plain wrong. Titan x(p) and 1080ti are exactly the same chip - perhaps the ti is from newer series where the manufacturing process matured hence higher clocks are possible, or just from better part of the wafer doesn't matter. They perform exactly the same - save for error margin and the TI having higher clocks on both chip and ram. Truth is unless AMD got another better chip in stores for RX there is no way it will perform better, not with the same drivers.

Think of it this way the Vega FE and RX are the Titan X(p) and 1080ti - cut version of gp102. Unless AMD got something that would be equivalent to gp102 uncut(Titan Xp) or gp100 there is no way they can release card that will beat Vega FE.

[Artorius]

1080Ti - GP102-350-K1-A1 - 3584:224:88 - 1480MHz - 1582MHz
Titan X Pascal - GP102-400-A1 - 3584:224:96 - 1417MHz - 1531MHz
Titan Xp - GP102-450-A1 - 3840:240:96 - 1405MHz - 1582MHz

Pay attention to the naming scheme.

The Titan Xp uses a GP102 in the same configuration as the Quadro P6000, the GP102 in the Titan X Pascal is a cut-down version of it, and the GP102 in the 1080Ti is an even more cut-down version of it. The differences between them are not only clocks and firmware/software-side tweaks. At first Nvidia just couldn't sell Titans with the entire die, so we got a cut-down GP102. Later they were probably having to cut-down perfectly working GP102s due to low demand of the Quadros, which made them release another Titan, with the full-die this time.

But anyway, for pro tasks what you're after is consistency, stability, predictability and reliability. You can do software side tweaks on top of the hardware configuration differences to make a product better suited to task X or Y, but I don't think that's the only deal with the RX Vega. I'm not sure they can close the gap they need to close with only this.

Yes the Vega FE is a "Titan" from AMD, but at this point we have no idea if the RX Vega will even ship with the same die. We have rumours saying that the delays are the result of a hardware respin, and we also have some benchmarks where the FE loses to Fiji-XT at the same clocks in games. And others where it stomps it in compute tasks. Being honest the theory that there's something wrong with this card makes sense, but who knows? We certainly don't. It could be that the product was never really designed with gaming in mind first, which isn't a bad thing really, and they're only working in improving it now.

In any case, we'll have all the answers we seek in a few weeks if it doesn't get delayed again.