• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Polaris performance numbers perf/$

defaultluser said:
Right, Tonga/Fiji has Delta Color Compression, but it's not as aggressive as Nvidia's version (that's why the 380 has to have twice the memory bus as the 960 to compete). That leaves some holes for improvement.

And Hawaii had zero frame buffer compression, so there's a lot of room for it's replacement to improve.

Reality: Nvidia was able to make a card that competed fairly well with the 390X with only slightly faster clocked 256-bit ram. If AMD can bump up the speed to 8Gbps, and introduce the same level of compression as Maxwell, they could do the same exact thing, or slightly faster! Add in a little more L2 cache over Maxwell, and you can make up the bandwidth difference enough to get within 10% of the 980 Ti!

Of course, they would lose their advantage of scaling well at higher resolutions, but that's what you lose when you give up that extra memory bandwidth. Would be an amazing card at $330 though :D
Click to expand...


That makes sense. Still getting it up to the 980ti even as close as 10% of it all the time, I can see that being a problem. Its not just bandwidth its shader array and throughput is going to make a difference too. At higher resolutions I think it shouldn't be too much of an issue if they have more ROPs as well. Also the color compression will help with the ROPs too.

Also GDDR5x is more expensive than GDDR5 too, so to put that on a card that is less that $350 which looks to be where this card is coming out, some people are saying less that $300, would be cutting into margins quite a bit.
 
TaintedSquirrel said:
The GTX 1070 is confirmed to use regular GDDR5, the 1080 is confirmed to use GDDR5X so it is very unlikely any Polaris cards will use it.
Which makes AMD's choice to go with the 256-bit bus even more odd, and also contradicts the "near 980 Ti" rumors since you don't put a weak bus on a card like that. They're also sporting 8 GB.

As to why GDDR5X is missing, maybe too expensive? Or low availability. The 1080 might not even be launching in June.

Frankly I don't care about GDDR5X or HBM. Most people don't. These aren't 4K cards, hell they're potentially not even 1440p cards.
Click to expand...

It would be disappointing if they're not. Clearly resolution increase have been a big trend in the past couple of years. A good 1440p (or similar) IPS monitor is cheap and performs well - I'm using one right now with my AMD 290X and it handles just about everything at max in-game settings. 4K resolutions are becoming prominent even on laptops today (ie Razer Blade Stealth, with a 4K IGZO Touchscreen, no GPU because you're supposed to use a Thunderbolt 3 external with a desktop card) and more affordable by the day through this year and next. This is to say nothing for the power needed to run VR if it truly takes off. The Playstation "4.5 / NEO" is said to be upgraded for dealing with 4K gaming and upscaling older content. I don't see how AMD or Nvidia can come out with brand new cards and not make these target features with reasonable performance (Except perhaps the 1080 Ti? ) If AMD basically says "these are cards for 1080p, or even 1440p " they'll be shooting themselves in the foot.

If one of these companies really wants to cement themselves in the market or change the direction of things, maybe they should be willing to take a lower profit on more expensive hardware and make it up in volume and future standing in the market (ie AMD becomes dominant with a lower priced high end that is almost as good as the 1080 Ti, their focus on Vulkan and other open tech is favored as opposed to Nvidia's proprietary crap. Its kind of like how in the R9 200 era the 290 and 290X were well ahead and locked up the entire cryptocurrency mining market too)

AMD and Nvidia alike are both pushing the VR and 4K gaming features with their materials previewing their new cards so I hope it isn't all bluster. They shouldn't skimp on the tech, RAM (8 gb min), or overall card power I really think AMD has the potential to move ahead, just like they did with Fury, but if they fall short it would be very disappointing.
 
"If you look at the total install base of a Radeon 290, or a GTX 970 or above [the minimum specs required for VR], it's around 7.5 million units," explained Taylor. "But the issue is that if a publisher wants to sell a £40/$50 VR game, there's not a big enough market to justify that yet. We've got to prime the pumps, which means somebody has got to start writing cheques to big games publishers. Or we've got to increase the install TAM [total addressable market].

"The reason Polaris is a big deal," continued Taylor, "is because I believe we will be able to grow that TAM significantly. I don't think Nvidia is going to do anything to increase the TAM, because according to everything we've seen around Pascal, it's a high-end part. I don't know what the price is gonna be, but let's say it's as low as £500/$600 and as high as £800/$1000. That price range is not going to expand the TAM for VR. We're going on the record right now to say Polaris will expand the TAM. Full stop."
Click to expand...

Why a reformed AMD is going all in on VR, rather than badmouthing Nvidia

Basically saying, we don't have any high end, we will focus on mainstream, hoping nvidia forgot about this market.
What world are they in? lol
 
xLegendary said:
Why a reformed AMD is going all in on VR, rather than badmouthing Nvidia

Basically saying, we don't have any high end, we will focus on mainstream, hoping nvidia forgot about this market.
What world are they in? lol
Click to expand...
Bizarre response but I expect nothing less from Roy.
In their defense, if Polaris/Pascal is a repeat of the HD4000/GTX 200 series then what they're saying is fine. A "mid-range" chip that performs 10% slower and costs half the price of its competition.
 
LOL yeah it is Roy,

yeah That is what I'm expecting, at least against the gtx 1070, nV does have room for the higher end product, but I don't think they will be willing to drop prices, we all know how they love their margins.
 
razor1 said:
LOL yeah it is Roy,

yeah That is what I'm expecting, at least against the gtx 1070, nV does have room for the higher end product, but I don't think they will be willing to drop prices, we all know how they love their margins.
Click to expand...

Precisely, which is why I think AMD has a shot at cutting NVidia down by the knees rather than chopping off its head as some want.

If AMD can get market share from budget-conscious builders, and build their reputation that way, they have a long term shot at taking it to NVidia, and NVidia might end up being buried by their own margins.
 
  • Like
Reactions: N4CR
like this
TaintedSquirrel said:
Really depends on GP106.
Click to expand...

Exactly. AMD is already struggling with the "budget" image and it hasn't helped the bottom line or increased market share yet. NV isn't going to give up any part of the market they dominate and it's wishful thinking that they would even consider that. AMD is going to have to make their way with good product and good marketing. Unfortunately for them they are only good at one of the two.
 
Like mentioned above, the difference is during AMD/ATI's victories like the 4000 series (maybe 5000 and 6000, can't remember), and the R9 200 - the high end like 290X ended up being maximum 10-15% slower (in some games equal, in some games better) than Nvidia's current high end part, being often several hundred dollars cheaper, while also giving a full/parity play experience. If Polaris 10's metric is the last generation high end, that doesn't bode well (unless of course the 1080 and 1080Ti are really only a 10% improvement or so over the 980 Ti ). Likewise, if there's a big difference in how VR or 4K games play (more than just 8x AF vs 16x AF) using high end Polaris versus high end Pascal all else being equal, that won't be good either. I guess we'll have to see how it progresses, but I really hope this turns out well..
 
DooKey said:
Exactly. AMD is already struggling with the "budget" image and it hasn't helped the bottom line or increased market share yet. NV isn't going to give up any part of the market they dominate and it's wishful thinking that they would even consider that. AMD is going to have to make their way with good product and good marketing. Unfortunately for them they are only good at one of the two.
Click to expand...


AMD isn't struggling with a "budget" image. AMD is struggling with a "Overhype and underperform" image. Plain and simple.

AMD was celebrated back in the day for being able to offer intel-like performances on their chips for almost half the cost at times. Back then, playing that role never hurt them. What hurt them is that intel finally found a chip that could put distance between the two, and AMD never counter-punched.


This time it's different. AMD isn't saying they are going to beat NVidia for the top, fastest cards. No, instead, they made a business decision, found a segment in which they felt they could compete, and if we are to believe the rumors, will be offering a huge amount of performance for the cost of mid-to-low end enthusiast cards. They are counting on NVidia not being able to match up their skus and prices with what they are offering, meaning that those people building on a budget will turn to AMD and be happy with it - at least that's AMDs plan.

The cards that NVidia are putting out matter less than the price point they put them out at. Can NVidia offer gamers the same value as AMD? That's the crux of it. If they pull this off, it won't earn them bragging rights, but it will earn them market share and a following, which they will need to compete going forward.

It's a sound business plan that all depends on how well they can execute, and how bound Nvidia is by their own margins.
 
defaultluser said:
So you think they'd eat the cost of mounting HBM1 (not cheap, even with a single stack) PLUS GDDR5 in the same system, and building a memory controller than can handle both at the same time, rather than just going 384-bit wide GDDR5? Cause I think you should share the crack you're smoking :D

Plus, I already came up with a REAL way they could get that level of performance with low cost. It's a lot easier to reverse-engineer your competitor's more advanced compression algorithm, rather than developing it on your own.
Click to expand...
Well if you think reverse engineering a color compression technique will help cut bandwidth usage for increasingly compute heavy applications that can't use it helps, I guess that would work. Seems backwards from the forward looking approach AMD has used in the past. They already have a memory controller that can handle it, so not sure what you're getting at there. You're arguing to increase the size of the chip and lower yields to fit a 384bit bus as opposed to adding a single stack of HBM1 with an interposer. Even without HBM there's a chance these chips were designed to use an interposer because of Zen. That being the case they could use almost any sized bus they wanted as it would effectively make the chip larger.
 
Zion Halcyon said:
AMD isn't struggling with a "budget" image. AMD is struggling with a "Overhype and underperform" image. Plain and simple.

AMD was celebrated back in the day for being able to offer intel-like performances on their chips for almost half the cost at times. Back then, playing that role never hurt them. What hurt them is that intel finally found a chip that could put distance between the two, and AMD never counter-punched.


This time it's different. AMD isn't saying they are going to beat NVidia for the top, fastest cards. No, instead, they made a business decision, found a segment in which they felt they could compete, and if we are to believe the rumors, will be offering a huge amount of performance for the cost of mid-to-low end enthusiast cards. They are counting on NVidia not being able to match up their skus and prices with what they are offering, meaning that those people building on a budget will turn to AMD and be happy with it - at least that's AMDs plan.

The cards that NVidia are putting out matter less than the price point they put them out at. Can NVidia offer gamers the same value as AMD? That's the crux of it. If they pull this off, it won't earn them bragging rights, but it will earn them market share and a following, which they will need to compete going forward.

It's a sound business plan that all depends on how well they can execute, and how bound Nvidia is by their own margins.
Click to expand...

Yeah that is one fine post Zion! Couldn't have said it better myself! It all comes down to execution and if nV wants to compete with the their initial Pascal lineup. We know they have a GP106 coming out but its one quarter out from the launch of Polaris, so any one that is looking for a quick upgrade will do it fairly quickly and by the time GP106 is out, its just who ever is left over...... It should grab marketshare from nV.
 
Anarchist4000 said:
Well if you think reverse engineering a color compression technique will help cut bandwidth usage for increasingly compute heavy applications that can't use it helps, I guess that would work. Seems backwards from the forward looking approach AMD has used in the past. They already have a memory controller that can handle it, so not sure what you're getting at there. You're arguing to increase the size of the chip and lower yields to fit a 384bit bus as opposed to adding a single stack of HBM1 with an interposer. Even without HBM there's a chance these chips were designed to use an interposer because of Zen. That being the case they could use almost any sized bus they wanted as it would effectively make the chip larger.
Click to expand...

ROP traffic still have to happen. That will get compressed.

Yeah, the compute load is slowly growing, but it will be awhile before it overtakes ROP bandwidth,.
 
It all sounds like the planets are going to align for AMD. Honestly, I think a lot of us have been looking at the entire graphics segment the wrong way. AMD has quietly grabbed the entire gaming industry with respect to PC's and consoles. We, as enthusiasts, look toward the flagship, but most don't. Most just want to play games. They don't care about justifying purchases and playing fanboy. Image a world where software and hardware actually play nice and are damn near standardized. If this all plays out we will be living it.
 
Zion Halcyon said:
AMD isn't struggling with a "budget" image. AMD is struggling with a "Overhype and underperform" image. Plain and simple.

AMD was celebrated back in the day for being able to offer intel-like performances on their chips for almost half the cost at times. Back then, playing that role never hurt them. What hurt them is that intel finally found a chip that could put distance between the two, and AMD never counter-punched.


This time it's different. AMD isn't saying they are going to beat NVidia for the top, fastest cards. No, instead, they made a business decision, found a segment in which they felt they could compete, and if we are to believe the rumors, will be offering a huge amount of performance for the cost of mid-to-low end enthusiast cards. They are counting on NVidia not being able to match up their skus and prices with what they are offering, meaning that those people building on a budget will turn to AMD and be happy with it - at least that's AMDs plan.

The cards that NVidia are putting out matter less than the price point they put them out at. Can NVidia offer gamers the same value as AMD? That's the crux of it. If they pull this off, it won't earn them bragging rights, but it will earn them market share and a following, which they will need to compete going forward.

It's a sound business plan that all depends on how well they can execute, and how bound Nvidia is by their own margins.
Click to expand...

I would argue AMD is struggling with both a "budget and overhype/underperform" image. I far too often see AMD proponents arguing perf/$ and at the same time not upgrading till AMD lowers their prices to fire sale rates. That's bad if you ask me. However, they do it because recent history says that's what AMD does.

Of course AMD was celebrated back in the day. I loved the Athlon chips and didn't leave them till Core2 released. When that happened their chips went on fire sales and this added to the budget image. Go to any forum and AMD proponents are arguing that you should buy a AMD cpu or gpu if you are on a slim budget. This is doing AMD no favors. That's why I say "budget" image. Unfortunately for the gpu division they are hobbled by the crappy cpu's of the last several years. Believe it or not customers link their gpu's and cpu's.....kinda makes sense for the average buyer.

I think NV was onto something when they dropped the 970 nuke and I expect them to go for blood again this round. NV likes their market position and I don't believe they will allow it to be taken from them. Mind share is closely related to perception and if NV has the top dog again then they will hold their customers since AMD doesn't seem to be attacking that market till next year. I think AMD made a mistake with their strategy this time, but they really didn't have much choice considering their market position, mind share, and lack of hard cash to push the limits. In the end we all will eventually pad the margins of NV and we know whose fault it is.......AMD.
 
razor1 said:
Yeah that is one fine post Zion! Couldn't have said it better myself! It all comes down to execution and if nV wants to compete with the their initial Pascal lineup. We know they have a GP106 coming out but its one quarter out from the launch of Polaris, so any one that is looking for a quick upgrade will do it fairly quickly and by the time GP106 is out, its just who ever is left over...... It should grab marketshare from nV.
Click to expand...

I disagree with this. Time after time the market has proven that most customers are waiting for the next shoe to fall from NV. The mind share they have gained doesn't just disappear overnight. AMD needs another ATI 9700 Pro. I'd love to see that. One of the best cards ever made.
 
I see what your saying Dookey, but if its priced right.... It could just that. There are many market factors I agree, but price (with decent performance) is definitely the biggest motivators in that segment.
 
DooKey said:
I disagree with this. Time after time the market has proven that most customers are waiting for the next shoe to fall from NV. The mind share they have gained doesn't just disappear overnight. AMD needs another ATI 9700 Pro. I'd love to see that. One of the best cards ever made.
Click to expand...

I agree. If AMD gets another 5% of the market, and a much larger percent of the gamers that actually buy VR headsets, it will be a HUGE win for them. When I worked retail people bought everything based off name brand. Everyone and their mother knows the name Nvidia. Those consumers, the mainstream consumer AMD is targeting, only buy Nvidia. Now if they order a Dell and it happens to have an AMD card in it, then it's fine. They will never go to a store and intentionally ask for an AMD card. The name Nvidia carries too much weight. Just like Intel carries weight. I never saw a person ask for an AMD CPU when they were the fastest. Not once. If the PC on sale happened to have one installed in it, then they went home with an AMD desktop.
 
Yeah, but this has the potential to be a disruptive product, much like the 4870 was. It doesn't have to own performance-wise, it just has to be a good value for a surprising amount of performance. They'll have the opportunity next year to see if Vega can roll P100.

Nvidia won't be caught dumb ever again after the 9700 Pro. But there's only so much they can pput into GP106, so in the meantime AMD can enjoy owning the $200-$300 market again.
 
AMD has owned the $300 market for a long time, aside from the 970's brief stint from Sept 2014 through June 2015.
 
I think people are overstating how amazing (or how unbelievable/unlikely) these targets are.

In reality you'd only need about a 50% performance improvement at the mid/mainstream segments (whatever you want to call it) which really should be what people expect given a 18 month+ newer generational product.

Cards like the 980ti are already well into (anything after GTX 970 and R9 390 start to scale poorly) the poorer perf/$ range as typically exhibited by high end products.

As such I'd actually argue that falling short of these targets would be disappointing to consumers and hitting them should be what consumers expect.
 
It's a matter of how you look at it. On the surface with how the information presented it, a 2x perf/$, it gives the impression that next gen products will be giving you 2x the performance at the same cost, which does go into the unlikely side of the bracket. However that isn't really, nowhere near, what needs to happen.

Basically what if we rephrase the question as "Do you think next gen GPUs can hit roughly 50% performance improvements in at least some product segments?"

As an interesting aside, GTX 970 at black friday sale prices (so 5 months ago) already hit about 2x perf/$ over launch GTX 980ti.
 
Last edited:
limitedaccess said:
It's a matter of how you look at it. On the surface with how the information presented it, a 2x perf/$, it gives the impression that next gen products will be giving you 2x the performance at the same cost, which does go into the unlikely side of the bracket. However that isn't really, nowhere near, what needs to happen.
Basically what if we rephrase the question as "Do you think next gen GPUs can hit roughly 50% performance improvements in at least some product segments?"
As an interesting aside, GTX 970 at black friday sale prices (so 5 months ago) already hit about 2x perf/$ over launch GTX 980ti.
Click to expand...

28 nm vs 14 nm and there is where AMD cuts cost, it allows more chips to be made with less material it took before
 
No the cost is about the same transistor for transistor. Even at half die size the cost of chip will be the same cost of double the die size at 28 nm......
 
razor1 said:
No the cost is about the same transistor for transistor. Even at half die size the cost of chip will be the same cost of double the die size at 28 nm......
Click to expand...

Slightly more actually, can't remember exact numbers for transistors at 14/16nm ff are more expensive, not just the same
 
It will take quite a bit of time for the prices to drop. New node transitions are more expensive now as Alkaladur on a per transistor level.
 
OK, so it's more that it's a new process, nothing intrinsic to the finfet tech. That's good.
 
schizo said:
OK, so it's more that it's a new process, nothing intrinsic to the finfet tech. That's good.
Click to expand...

No, it is specific to FF at "14/16nm" aka 20nm

At no other point in history has a node transition resulted in EQUAL transistor cost, let alone it costing more.
 
Yeah this is the first time this has happened, and this will continue to happen as tooling and other things (R&D, etc) are just much more expensive as the node size drops.
 
razor1 said:
Yeah this is the first time this has happened, and this will continue to happen as tooling and other things (R&D, etc) are just much more expensive as the node size drops.
Click to expand...

GP100 costs more than double what GM200 did. Food for thought.
 
That's not what I meant-- I was asking if there's anything inherent to the new 16nm finfet that would maintain high costs, or if it's just transition costs. Understand that the transition costs are much higher than previous nodes.

So while GP100 is >200% the cost of GM200 today, that cost will come down over time, while 28nm is very mature so it probably won't.
 
Well the inherent costs of initial development is what will keep the price high, eventually as more and more companies use it the process and as the process matures, price will drop, just like anything else in volume production, but since starting at a higher price the price drop will be relative to that.


Waiting on 20nm graphics cards from Nvidia and AMD? Don't. | TechSoda


20nm-cost-crossover.jpg
 
"I just got my hands on some Polaris cards, better run a bunch of useless OpenGL benchmarks nobody has ever heard of."
 
razor1 said:
Well the inherent costs of initial development is what will keep the price high, eventually as more and more companies use it the process and as the process matures, price will drop, just like anything else in volume production, but since starting at a higher price the price drop will be relative to that.


Waiting on 20nm graphics cards from Nvidia and AMD? Don't. | TechSoda


20nm-cost-crossover.jpg
Click to expand...

What is a X'over

I get X'tor

but whats a transover :p

That graph is so satisfying though. Seeing transistor cost correctly described as a function of scaling factor, wafer cost, and yields - with the latter being a function of time- is just so accurate it tickles my math pickle

Brackle said:
Huge shocker I know!

Watch the next set of leaks will be a useless game. Like Call of Duty!
Click to expand...

Don't be stupid Brackle, everyone knows the REAL performance metric is CS:GO at 8K
 
Last edited:
You must log in or register to reply here.
Back
Top