AMD Vega Architeture Speculation

It's not a rape if everyone lines up for months to get one.

Unfortunately Nvidia has demonstrated that the market for $800+ video cards is there and it has crazy demand. PC gaming has always been a pretty cheap hobby compared to a lot of other hobbies, so maybe a trend upwards in pricing was inevitable.

You can expect AMD to try and stay close to Nvidia's pricing if they are close on performance. They won't leave money on the table just because, they have shareholders who demand results like any other company.
 
Fury X didn't have the performance. Wrong pricing must have caused AMD some money rather than positioning it even as a 600 bucks part.

Even now there are "deals" on Fury X for 399 but there are also "deals" on 980 Ti OC editions for 400-450 and 1070s are already showing up at 420 onwards. Both practically slaughter the Fury X unless you are looking at one or two benches and crappy games.

So everytime AMD has a card since 7970, it is beat in performance (at the point of time of release) by another NVidia card unless we talk about sub-250 parts where AMD has ruled (IMO) for quite a while now even since the 4850 days. I still think a 290X at sub-200 is quite a steal.
 
AMD's next-gen Vega 10, Vega 11 GPUs spotted in recent OpenCL driver

Read more: http://www.tweaktown.com/news/53032...-gpus-spotted-recent-opencl-driver/index.html

KAC: the Doom and AoTS situations will be more and more when more and more devs will use API's...This is a gamble that AMD made when releasing Mantle...Some buyers may gamble on this or not...Is just a more adventurous path and for some is more fullfiling...

If AMD will come up with a vega version this year the fury line prices will deep much more...Remember also that the new consoles with upgraded (AMD) hardware are around the corner and you port those games that are made for the new consoles .... you know where things will go ...
Nvidia - conservative side - guaranteed constant FPS
AMD - adventurous path -- more big leaps in possible future

AMD may consider the Fury X more relevant as we go in the future...This is why they kept the price around 980ti level..Also the out of box liquid cooling is also keeping a pretty premium price...The 980ti reference design didn't have the out of box liquid cooling...Factoring the price for liquid cooling if you have to buy it from third party on 980ti will lead a much more expensive experience for 980ti...The air cooled 980ti are more value oriented but if you want liquid cooling enthuziast card i do feel the Fury x is a better value proposition than 980ti even with all it's weaknesses...
 
It's not a rape if everyone lines up for months to get one.

Unfortunately Nvidia has demonstrated that the market for $800+ video cards is there and it has crazy demand. PC gaming has always been a pretty cheap hobby compared to a lot of other hobbies, so maybe a trend upwards in pricing was inevitable.

You can expect AMD to try and stay close to Nvidia's pricing if they are close on performance. They won't leave money on the table just because, they have shareholders who demand results like any other company.


Unless you first keep it lower to build up your market share......Once that is back up to near 50 / 50 between AMD and Nvidia, then feel free to gouge on the prices.


It's no where near that right now as we all know.
 
I think AMD will price their cards cheaper, unless they have a massive performance advantage. A lot of people just reflexively buy Nvidia cards. People who aren't really into hardware don't even know about AMD.
 
The problem with pricing your cards cheaper is that you become known for being 'cheap'; so not only do you make less money but people will still buy the more expensive option, even if the performance difference doesn't make sense for the cost difference, due to perception that a more expensive product must be better.

Then, later, when you try to rise prices back to the 'correct' level people think you are over pricing and continue to buy your competitors products.

Pricing and perception is a right pain and is a real uphill battle for AMD frankly.
 
The problem with pricing your cards cheaper is that you become known for being 'cheap'; so not only do you make less money but people will still buy the more expensive option, even if the performance difference doesn't make sense for the cost difference, due to perception that a more expensive product must be better.

Then, later, when you try to rise prices back to the 'correct' level people think you are over pricing and continue to buy your competitors products.

Pricing and perception is a right pain and is a real uphill battle for AMD frankly.


For those that have the money to spend, that's likely the case but only a relatively small amount of people are in this position.....if we were to apply this very problem in the car world, it's like comparing a Corvette Z-06 to a Porsche 911 turbo or worse yet, a Ferrari which are insanely more expensive than the corvette, and would have a hard time out running it.



I'll take the corvette and keep the rest of the money in the bank, thanks.....:D


I had this very discussion just a few days ago, and unless a GPU vendor makes a huge mistake in the design of their product, performance comparisons between 2 cards of the same generation, and that are aimed for the same market segment, are usually within 10% of each other in overall performance, with some games being won by one card and others being won by the other.



Even if the performance lead was always consistent towards that one card by that ~10% in all games, but they want an extra 150~200$ because it's faster?.....They can shove it you know where.....:lol:
 
I'm seeing speculation in the form of math I don't actually understand from info some have gathered so far that Zen has about the same IPC performance as Haswell. I'll take that.
 
I'm seeing speculation in the form of math I don't actually understand from info some have gathered so far that Zen has about the same IPC performance as Haswell. I'll take that.


I'd do the same if the prices are a fair amount lower since as I said above, that extra 5~10% extra performance wins benchmarks, but if it comes at at large price premium and when the day comes that the slower option just doesn't cut it anymore, the marginally faster one ( that extra 10% ), won't run it much better anyhow.



These small performance leads are being seriously over hyped and marketing departments try to milk everything they can out of them, retail price wise.
 
Last edited:
The Titan Variant will probably use HBM2 whereas the Ti variant will probably be GDDR5X on a 384bit bus. I don't think they will give the Ti HBM2 because it will justify the cost of the Titan and performance difference.

What it will come down to is cost. I can see Nvidia releasing the Ti at the same price as the 1080 is now ( not FE ) and then dropping it's price on the 1080 down to 299/399 area which would decimate AMD'S lineup even more. The Titan will be around 1499/1599 area....

Close but not close enough. The Titan is launching with 12gb GDDRX ram at an estimated price of $1200. No HBM2 but it is slated for release on 2nd August. You were right with the 384bit bus but not for the TI.

This rumor is from WCCFtech so could be crap but if accurate where does it leave the TI which can't, logically, launch later with HBM2. Strange times.
 
Close but not close enough. The Titan is launching with 12gb GDDRX ram at an estimated price of $1200. No HBM2 but it is slated for release on 2nd August. You were right with the 384bit bus but not for the TI.

This rumor is from WCCFtech so could be crap but if accurate where does it leave the TI which can't, logically, launch later with HBM2. Strange times.



It's not a rumor anymore.....

http://techreport.com/news/30422/nvidia-unveils-a-pascal-powered-titan-x-with-11-tflops-on-tap


The GP102 GPU within the Titan X has 12 billion transistors, a 384 bit memory bus and 12GB of GDDR5x with a total of 480GB/sec of memory bandwidth.


So it seems Nvidia said "screw it" with regards to HBM until they get production going for real, and went with GDDR5x instead.....


More than ever AMD needs to show what Vega is all about.....
 
The new Titan X (2016) has 480 GB/s of memory bandwidth with GDDR5X. I really doubt that card needs more than that and it's being bandwidth-starved for consumer GPU applications.

AMD is really going to Fury X themselves (again) by waiting for HBM2 which is unlikely to actually do anything useful because Vega simply won't need that much bandwidth. They should have gone with GDDR5X like Nvidia did and pushed Vega out sooner. Instead they are giving Nvidia at least 6 months of completely uncontested sales at the high end between the 1070, 1080, and Titan X (2016).

Nvidia is going to suck all the money and sales out of the high end and by the time Vega comes out there won't be anybody who waited beyond the few most diehard AMD fans and no one will buy it because they have already had their 1070's, 1080's, and Titan X (2016) for more than 6 months.
 
The new Titan X (2016) has 480 GB/s of memory bandwidth with GDDR5X. I really doubt that card needs more than that and it's being bandwidth-starved for consumer GPU applications.

AMD is really going to Fury X themselves (again) by waiting for HBM2 which is unlikely to actually do anything useful because Vega simply won't need that much bandwidth. They should have gone with GDDR5X like Nvidia did and pushed Vega out sooner. Instead they are giving Nvidia at least 6 months of completely uncontested sales at the high end between the 1070, 1080, and Titan X (2016).

Nvidia is going to suck all the money and sales out of the high end and by the time Vega comes out there won't be anybody who waited beyond the few most diehard AMD fans and no one will buy it because they have already had their 1070's, 1080's, and Titan X (2016) for more than 6 months.

Nvidia helped develop GDDR5X, that's why AMD isn't using it currently. Same situtation as to why we never saw a Maxwell with HBM1.

The announcement by Micron indicates that the company will be the only supplier of GDDR5X memory for NVIDIA’s GeForce GTX 1080 graphics adapters, at least initially

The option currently would be to release a 4GB HBM1 variant which the industry would laugh at. Or try and rush a small amount of Vega's out the door, but that really looks unlikely before Q12017 and even then it will probably not be a wide availability.
 
Based on people here, I don't think you can actually suck the money out of high end sales. If Vega is faster, they'll all spend another $1000 buying it.

However, I agree Nvidia is smart to release this card ASAP since they have the whole market to themselves. Prices will probably have to come down when there's actual competition.
 
Back
Top