NVIDIA to make even more FE cards - could overall prices rise?

[Komshija]

NVIDIA cards cost more because they perform better (power, heat and noise, not just framerate) so that's not an argument. Also, both sides have sponsorship deals to make their cards work better on certain games. It's a dirty political trick to boost sales that I don't like, but it's a fact of life unfortunately.

That's only partially true. Some NVidia cards are terribly hot and very loud. Noise much more depends on fan/radiator design than solely on graphic chip.
The point is that NVidia knows that it can sell 10-15% more expensive GPU than AMD's version with equal performance. Most buyers will buy NVidia instead of cheaper and sometimes faster AMD, because NVidia's marketing department and some popular gamers who never even built a PC brainwashed them.

 

[FYFI13]

That's only partially true. Some NVidia cards are terribly hot and very loud. Noise much more depends on fan/radiator design than solely on graphic chip.
The point is that NVidia knows that it can sell 10-15% more expensive GPU than AMD's version with equal performance. Most buyers will buy NVidia instead of cheaper and sometimes faster AMD, because NVidia's marketing department and some popular gamers who never even built a PC brainwashed them.

For me it isn't because of marketing - it's because of really poor drivers. Granted, in some games/setups they work just fine, but there are scenarios where it makes you want to throw that GPU into the wall. I tried to "adopt" R9 380, then RX 480 lately. No more. Not touching AMD cards in foreseeable future. I will keep using them in my client builds when requested, but not for myself. Even if Vega turns out twice faster than GTX 1080Ti and for half price - i wouldn't use it in my rig because of damn drivers.

 

[Komshija]

For me it isn't because of marketing - it's because of really poor drivers. Granted, in some games/setups they work just fine, but there are scenarios where it makes you want to throw that GPU into the wall. I tried to "adopt" R9 380, then RX 480 lately. No more. Not touching AMD cards in foreseeable future. I will keep using them in my client builds when requested, but not for myself. Even if Vega turns out twice faster than GTX 1080Ti and for half price - i wouldn't use it in my rig because of damn drivers.

AMD improved a lot during the past few years. I would say that they made one huge step forward considering drivers.

 

[qubit]

That's only partially true. Some NVidia cards are terribly hot and very loud. Noise much more depends on fan/radiator design than solely on graphic chip.
The point is that NVidia knows that it can sell 10-15% more expensive GPU than AMD's version with equal performance. Most buyers will buy NVidia instead of cheaper and sometimes faster AMD, because NVidia's marketing department and some popular gamers who never even built a PC brainwashed them.

Yeah, I don't quite agree with that. I'm ok to agree to differ.

In the end buy whatever brand makes you happy, regardless of the whatever the "best" brand may be. From my point of view, since I went NVIDIA in 2009, I've had an excellent experience with their products and they have the best performing top end cards, which is why I buy them, so this works for me. I sincerely hope that Vega gives NVIDIA a kicking as us consumers will then win, on several fronts.

For me it isn't because of marketing - it's because of really poor drivers. Granted, in some games/setups they work just fine, but there are scenarios where it makes you want to throw that GPU into the wall. I tried to "adopt" R9 380, then RX 480 lately. No more. Not touching AMD cards in foreseeable future. I will keep using them in my client builds when requested, but not for myself. Even if Vega turns out twice faster than GTX 1080Ti and for half price - i wouldn't use it in my rig because of damn drivers.

I'm surprised you've had so many problems with the drivers as they seem to have improved a lot in recent years according to what I read on here. Note that I'm not giving a personal experience opinion on them here since I've been using NVIDIA cards exclusively since 2009 so wouldn't know.

@erocker is one of the primary culprits for saying that AMD drivers work ok nowadays, so blame him.

 

[rtwjunkie]

For me it isn't because of marketing - it's because of really poor drivers. Granted, in some games/setups they work just fine, but there are scenarios where it makes you want to throw that GPU into the wall. I tried to "adopt" R9 380, then RX 480 lately. No more. Not touching AMD cards in foreseeable future. I will keep using them in my client builds when requested, but not for myself. Even if Vega turns out twice faster than GTX 1080Ti and for half price - i wouldn't use it in my rig because of damn drivers.

FWIW, I know it's only anecdotal, but I also run AMD, and have yet to encounter any driver issues. I handle them the way i handle my NVIDIA drivers: find a good working one with acceptable features, and stick with it until that next really good candidate to replace it comes along, which might be 6 months.

 

[qubit]

FWIW, I know it's only anecdotal, but I also run AMD, and have yet to encounter any driver issues. I handle them the way i handle my NVIDIA drivers: find a good working one with acceptable features, and stick with it until that next really good candidate to replace it comes along, which might be 6 months.

Wot, you don't immediately install the latest one with wild abandon like I do? tsk rtw!

I did actually come unstuck a couple of months ago doing just this lol. But it's ok, I can fix it hehe.

 

[londiste]

i don't get the entire premise of this thread.

fe cards are a non-issue and nvidia has been very open from day 1 about what they are doing and why.

fe cards are meants to ensure market presence for specific cards for the duration that nvidia wants to have it. if aib-s decide to not make cards, fe should still be available. higher prices are not price gouging but rather a measure to get aib-s to be ok with what nvidia is doing. as many have pointed out already nvidia is effectively pricing fe cards out of the market. this is intentional as these were never meant to compete with aib cards. even 1070/1080 availability issues in summer of '16 were not split across fe/aib.

titan x is the exception here and aib-s have been pretty vocal about their disappointment. the reason they have not been more vocal is probably because titan x is way up in the ultra-high-end and profits from its sales would not be very significant anyway.

 

[rtwjunkie]

i don't get the entire premise of this thread.

fe cards are a non-issue and nvidia has been very open from day 1 about what they are doing and why.

fe cards are meants to ensure market presence for specific cards for the duration that nvidia wants to have it. if aib-s decide to not make cards, fe should still be available. higher prices are not price gouging but rather a measure to get aib-s to be ok with what nvidia is doing. as many have pointed out already nvidia is effectively pricing fe cards out of the market. this is intentional as these were never meant to compete with aib cards. even 1070/1080 availability issues in summer of '16 were not split across fe/aib.

titan x is the exception here and aib-s have been pretty vocal about their disappointment. the reason they have not been more vocal is probably because titan x is way up in the ultra-high-end and profits from its sales would not be very significant anyway.

The problem with your premise is that the FE is nothing exceptional it is a bog-standard reference card with fancy marketing to allow a price mark-up. Because AIB cards have always been better performers in both speed and cooling than reference cards, AIB manufacturers charge more for what is a better product. The problem now though is the FE has skewed that starting price point all to hell, because those AIB manufacturers KNOW the FE is just a reference card.

 

[FYFI13]

YI'm surprised you've had so many problems with the drivers as they seem to have improved a lot in recent years according to what I read on here. Note that I'm not giving a personal experience opinion on them here since I've been using NVIDIA cards exclusively since 2009 so wouldn't know.

Actually it's just a few issues, but what made me really sad was that same issues that i had reported to AMD were still present after more than a year. For example, if i change my monitors refresh rate from 60Hz to 75Hz (overclock) from Crimson Settings i get black screen after next reboot. The only way to fix is is to reboot into safe mode and uninstall display drivers completely. However, it works just perfectly fine if i do so with Custom Resolution Utility. Another thing is poor performance in Arma 3 multiplayer because cards downclocking itself due to low GPU utilization which causes even lower FPS and very stuttery gameplay. In single player it works fine since then you're not being limited by servers performance. Again, using third party tool (written by a German kid, btw) you can force GPU clocks to stay at the maximum. Well, almost maximum. GPU clocks were advertised as 1328MHz, and i want card to operate that way instead of "saving power" for me when i don't need it. In most games this is not an issue, yet in Arma 3 it clocks down to 450MHz because "there is no need for more". Wrong. With Nvidia it's easy, just go to CP, select Arma 3 profile and set power level to "prefer maximum performance". Sorted.

 

[yogurt_21]

I hadn't had a driver issue with my R9 290 until recently and even then it's game specific. So I've ignored it. Will likely replace the gpu in june with whatever is out...and of course new stuff will come out in august just to spite me. I've learned to roll with it.

At any rate I won't be going FE because I'm not an early adopter by any means so having FE cards doesn't matter to me. I will take a look at both AMD and Nvidia, find the best card in my pricepoint, and buy it. I've done that ever since my Kentsfield/9600gt rig and I've never regretted it.

Before that I was a fanboy for AMD/ATI rigs and had several times when I could have swapped to Intel or Nvidia for the same money and had a better rig but didn't because fanboy. Not a fanboy anymore, just a consumer who wants the best deal.

 

[kn00tcn]

Why should we care, I'm dumfounded, but I'll stick with my wallet cares mate and he gets the biggest vote.
It's exactly the mid range kit for top end prices that stops dead ,people like me buying Nvidias wares.
I vote with my wallet and bare much mocking for my peasantry, but I'll live.

intel just announced a 24 core 48 thread cpu for almost $9,000, it's the same situation

if the power draw goes down yet there's still a performance increase, that leaves room to make a bigger chip than what was previously possible & to make a smaller chip that's as fast or faster than what used to be the biggest chip, so would you suddenly price a new gen that beats everything old at a constantly lower price, making the tons of old stock obsolete e-waste?

meanwhile nv's laptops started using the same gpus as desktops, with the performance to show for it, unlike the past mess of model numbers

then we can forget about amd competing at all if 1060 was GP104 at $250... i dont even like nvidia but i'm not going to ignore the fact that price to performance is an important concept for everyone involved

not sure you understood the 'why should i care' part, i'm saying if you always pick the $300 tier, if it's always 30% faster than last gen, if eventually one gen moves from Gx00 to Gx04, but every other variable is identical, then there is no difference, you wouldnt even know about a bigger gpu if it wasnt public information or if they only kept the big one for professional compute, your upgrade schedule & wallet cost has not changed

with the exchange rate here, gtx1070 is like $700, of course that's too much, this tier is supposed to be $300-350 (400-460 converted), so i'm not into high prices, but not due to arbitrary gpu sizes

 

[londiste]

The problem with your premise is that the FE is nothing exceptional it is a bog-standard reference card with fancy marketing to allow a price mark-up. Because AIB cards have always been better performers in both speed and cooling than reference cards, AIB manufacturers charge more for what is a better product. The problem now though is the FE has skewed that starting price point all to hell, because those AIB manufacturers KNOW the FE is just a reference card.

you are missing the point. fe is not meant to be best or good value for money. it is meant to ensure availability.
fe does not skew the starting point, it has a different and deliberately higher msrp.
aib cards are available at the "normal" msrp as always.

 

[rtwjunkie]

you are missing the point. fe is not meant to be best or good value for money. it is meant to ensure availability.
fe does not skew the starting point, it has a different and deliberately higher msrp.
aib cards are available at the "normal" msrp as always.

Except the AIB's did not view it that way.

 

[yogurt_21]

Except the AIB's did not view it that way.

why would they? Their options usually have a better bundle and better coolers. Why should they sell for less money?

the fact remains that nvidia could have simply ensured proper stock levels prior to launch without a "special edition" card.

They chose to spin it as a 50-100$ early adopter fee...as if bugginess wasn't bad enough.

The odd thing is...instead of being wise consumers and shunning the concept, they sold like crazy!

so that's cool. Not like wise consumerism was the only thing keeping massive profiteering in check...oh wait.

 

[rtwjunkie]

why would they? Their options usually have a better bundle and better coolers. Why should they sell for less money?

the fact remains that nvidia could have simply ensured proper stock levels prior to launch without a "special edition" card.

They chose to spin it as a 50-100$ early adopter fee...as if bugginess wasn't bad enough.

The odd thing is...instead of being wise consumers and shunning the concept, they sold like crazy!

so that's cool. Not like wise consumerism was the only thing keeping massive profiteering in check...oh wait.

Exactly my point to the other guy. AIB's regarded it as a new starting point.

 

[Slizzo]

intel just announced a 24 core 48 thread cpu for almost $9,000, it's the same situation

Come now, that's a server part. Also, 24 cores on a single package is quite a bit.

 

[TheoneandonlyMrK]

intel just announced a 24 core 48 thread cpu for almost $9,000, it's the same situation

if the power draw goes down yet there's still a performance increase, that leaves room to make a bigger chip than what was previously possible & to make a smaller chip that's as fast or faster than what used to be the biggest chip, so would you suddenly price a new gen that beats everything old at a constantly lower price, making the tons of old stock obsolete e-waste?

meanwhile nv's laptops started using the same gpus as desktops, with the performance to show for it, unlike the past mess of model numbers

then we can forget about amd competing at all if 1060 was GP104 at $250... i dont even like nvidia but i'm not going to ignore the fact that price to performance is an important concept for everyone involved

not sure you understood the 'why should i care' part, i'm saying if you always pick the $300 tier, if it's always 30% faster than last gen, if eventually one gen moves from Gx00 to Gx04, but every other variable is identical, then there is no difference, you wouldnt even know about a bigger gpu if it wasnt public information or if they only kept the big one for professional compute, your upgrade schedule & wallet cost has not changed

with the exchange rate here, gtx1070 is like $700, of course that's too much, this tier is supposed to be $300-350 (400-460 converted), so i'm not into high prices, but not due to arbitrary gpu sizes

im an engineer for a living so it bites my ass a bit to see the Bom stay the same or be reduced yet the price goes up, thats market conditions grabbing by nvidia as you seam to agree and what im saying is i saw it happening and bought from elsewhere on Principal, performance be damned.

what your saying is that so long as its 30% better you dont mind and they have done enough to make you happy to pay twice the price, not me sir , feck helll no.

and a 24 core chip is not the same , its been binned with losses that need to be equalised, and due to yeilds its a rare and in demand chip hence price, it isnt a mid range spec asic made on mass and specced for the yields to fall within a certain performance and yeild range.

ones ultra top end chip ,nvidias is just not.

 

[kn00tcn]

im an engineer for a living so it bites my ass a bit to see the Bom stay the same or be reduced yet the price goes up, thats market conditions grabbing by nvidia as you seam to agree and what im saying is i saw it happening and bought from elsewhere on Principal, performance be damned.

what your saying is that so long as its 30% better you dont mind and they have done enough to make you happy to pay twice the price, not me sir , feck helll no.

and a 24 core chip is not the same , its been binned with losses that need to be equalised, and due to yeilds its a rare and in demand chip hence price, it isnt a mid range spec asic made on mass and specced for the yields to fall within a certain performance and yeild range.

ones ultra top end chip ,nvidias is just not.

i didnt say twice the price, i kept saying the same price for the same performance increase vs previous gen

'high end geforce' marketing model has been 'about 500-600' for several generations including before kepler's 'midrange' engineering model, the existence of a larger gpu (the first titan even took a while to appear late) did not change the value of the marketing model to the end user, what is the point of exaggerating twice the price?

7800gtx - $600
8800gtx - $600
gtx280 - $650 turned $500 thanks to amd
gtx480 - $500
gtx580 - $500 (not quite the expected perf boost, but same happened to amd, all due to tsmc's missing 32nm)
gtx680 - $500 (the first 'midrange' physical, but nothing else changed, still perf boost, same price, no resulting difference)
gtx780 - $500
gtx980 - $550
gtx1080 - $600

for the intel comparison, what do you mean nvidia's is not, i'm saying there are top consumer i7 cores & then there's the maximum cpu they've ever made... for nv there's the top consumer gpu, & now for a few years there's the additional maximum gpu they've ever made, but the game performance per watt or die size doesnt scale, the clocks cant be as high, etc

what is the point in limiting to a single chip for every use case? wide & slow has one use, lean & fast has another use, especially for most games... we need high clocked single thread with a few/enough cores on cpu & high clocked parallel enough gpus that dont end up 300 watts, this is separate from fiddling with turning $500 to $600

 

[TheoneandonlyMrK]

i didnt say twice the price, i kept saying the same price for the same performance increase vs previous gen

'high end geforce' marketing model has been 'about 500-600' for several generations including before kepler's 'midrange' engineering model, the existence of a larger gpu (the first titan even took a while to appear late) did not change the value of the marketing model to the end user, what is the point of exaggerating twice the price?

7800gtx - $600
8800gtx - $600
gtx280 - $650 turned $500 thanks to amd
gtx480 - $500
gtx580 - $500 (not quite the expected perf boost, but same happened to amd, all due to tsmc's missing 32nm)
gtx680 - $500 (the first 'midrange' physical, but nothing else changed, still perf boost, same price, no resulting difference)
gtx780 - $500
gtx980 - $550
gtx1080 - $600

for the intel comparison, what do you mean nvidia's is not, i'm saying there are top consumer i7 cores & then there's the maximum cpu they've ever made... for nv there's the top consumer gpu, & now for a few years there's the additional maximum gpu they've ever made, but the game performance per watt or die size doesnt scale, the clocks cant be as high, etc

what is the point in limiting to a single chip for every use case? wide & slow has one use, lean & fast has another use, especially for most games... we need high clocked single thread with a few/enough cores on cpu & high clocked parallel enough gpus that dont end up 300 watts, this is separate from fiddling with turning $500 to $600

No company is making just one chip for every use case but that is the point chips cost big bucks per iteration 12 layers triple patternated possibly unique masks plus processes is definitely not cheap so one design is very often multiple purpose with some stuff hidden to different use cases that's how they roll ,stop including me in your we I'd like it to be more then just good at optimised dx11 tasks and proprietary shit I'm sorry but that's my opinion.
And yeah all those were too dear in my opinion I bought none of them high end cards
Even though I know some are good ish.

In fact look back at the 480 -580 -680 gens the first two were shit ish so they pulled their thumbs out their proverbial then slapped every buyers proverbial ,not cool I remember the inflated prices.

 

[kn00tcn]

No company is making just one chip for every use case but that is the point chips cost big bucks per iteration 12 layers triple patternated possibly unique masks plus processes is definitely not cheap so one design is very often multiple purpose with some stuff hidden to different use cases that's how they roll ,stop including me in your we I'd like it to be more then just good at optimised dx11 tasks and proprietary shit I'm sorry but that's my opinion.
And yeah all those were too dear in my opinion I bought none of them high end cards
Even though I know some are good ish.

well now i'm curious... are you into the $300 tier instead? always get amd's top tier? did you know i had a 4870x2 when it launched in 2008 until it died in 2013? (i even met robert hallock in person at the amd extravalanza at ati HQ)

for a decade or more it's been said how radeon has well engineered hardware, also first to smaller nodes & new memory types (but all this isnt in a vacuum, there is still perf per watt per dollar relative to every other sibling/parent/competitor that's already out)

 

[TheoneandonlyMrK]

well now i'm curious... are you into the $300 tier instead? always get amd's top tier? did you know i had a 4870x2 when it launched in 2008 until it died in 2013? (i even met robert hallock in person at the amd extravalanza at ati HQ)

for a decade or more it's been said how radeon has well engineered hardware, also first to smaller nodes & new memory types (but all this isnt in a vacuum, there is still perf per watt per dollar relative to every other sibling/parent/competitor that's already out)

I mix it up I got a 5870x2 gtx 460x2 ,7970, r9390 then these but I fold so compute always mattered and it's a matter of integrity ,for me Nvidias isn't great after such shenanigans as the gtx 680 ( first little big card) or memory gate and others like Fe cards , plus I have owned a few Nvidia cards besides temperarily but mostly used for physx (hybrid).
It's specifically the better physical integrity and business decisions I partly am investing in but it's gotten play crisis right that and keep me rolling for a year or few.

 

[kn00tcn]

I mix it up I got a 5870x2 gtx 460x2 ,7970, r9390 then these but I fold so compute always mattered and it's a matter of integrity ,for me Nvidias isn't great after such shenanigans as the gtx 680 ( first little big card) or memory gate and others like Fe cards , plus I have owned a few Nvidia cards besides temperarily but mostly used for physx (hybrid).

the funny thing is, back in late 2007/2008 i was itching to build... thought about 1440x900, got set on 8800gts512, but then the failing mobile gpus scandal appeared, maybe some other scandals appeared, started getting irritated at how they handle themselves, needed performance for 1920x1200 which means a single gpu was not enough hence the only choice being ati, then more scandals kept happening with batman's nv locked AA, they continue to happen of course...

but still, when GK104 happened, after comparing the whole picture, i didnt arbitrarily scold them purely on the gpu itself, the gpus werent underdelivering in games, only compute regressed (i wouldnt want much compute in the first place, i liked intel's idea of a compute only card/architecure separate from cpus separate from gpus)

EDIT: reminds me of the first dx10 cards, i very specifically saw benchmarks that showed 8800 was an unusually large boost over 7800 compared to 7800's boost over 6800, then you look at 2900 vs x1900 & x1900 vs x800, ati's boosts were even... so what actually happened back then was nv jumped ahead too far, not that ati sucked (well, maybe the power sucked as always, but 3870 was a great idea... read the background engineering story of 3870 & 4870 on anand, 30 pages of amazing)

also funny, i happen to have a 660 non ti, the one with the asymmetric memory, the one that's supposed to be worse than 970's method, yet nobody complained until 970... yes i only see 1.5gb in afterburner during games, oh well

 

[Slizzo]

gtx780 - $500

GTX 780 was $650 at launch.

 

[kn00tcn]

GTX 780 was $650 at launch.

ah i see that now, had anand's 780ti+780 price drop numbers show up

 

[Slizzo]

ah i see that now, had anand's 780ti+780 price drop numbers show up

Yeah, it's difficult to remember that the 780 was out a few months before the 780Ti came out and supplanted it in the lineup.