[BitsAndChips]390X ready for launch - AMD ironing out drivers - Computex launch

[RussianSensation]

But I don't think multiplying the frequency with the cores is the right thing to do. Going by an other hardware.fr test, the diff between 290 and 290X is around 7%. That's far from the 18% when comparing CU*clocks.

The 290X vs. 290 scaling doesn't work because there is not a linear increase in ROPs/memory bandwidth. R9 390X should have more than double the effective memory bandwidth increase over R9 290X and more than 50% increase in shader, texture and pixel fill-rate. But even if we assume your 45.6% increase, that puts us 2-3% behind Titan X. At $700, that would already make it a better buy than the Titan X. Now imagine R9 390 nonX just 10% slower than an R9 390X --> we would end up with a chip 13-14% slower than the Titan X for $500. Say hello to 2 of those in CF! Good-bye Titan X! :biggrin:

As I said before, all the hype is centered around 390X but if you look back at the history of AMD (5850/6950 unlocked/7950/R9 290 nonX), it's those 2nd tier AMD cards, when overclocked, provide hands down the best value on the AMD high-end. Remember this article?

AMD tends to charge too high of a premium for the small difference for its flagship cards (remember X850XT PE vs. X800XT or 9800XT 256MB over 9800 Pro?) AMD can still entice gamers to step-up to the R9 390X at $700 over the $500 R9 390 nonX if the latter only has 4GB of VRAM+air cooling, but the former has 8GB+AIO CLC. That would provide a huge incentive for high-end gamers to step-up to the 390X.

 

[AtenRa]

But even if we assume your 45.6% increase, that puts us 2-3% behind Titan X. At $700, that would already make it a better buy than the Titan X. Now imagine R9 390 nonX just 10% slower than an R9 390X --> we would end up with a chip 13-14% slower than the Titan X for $500. Say hello to 2 of those in CF! Good-bye Titan X! :biggrin:

That will make the GTX980 to fall to $350 and it will only be 10-15% slower than the 390.

 

[VulgarDisplay]

That will make the GTX980 to fall to $350 and it will only be 10-15% slower than the 390.

Everyone knew the gtx980 was just a midrange card branded as high end anyways. Still shelled out $650+ for them.

I will NEVER understand customers that loyal to a brand. They know they are getting raped, and welcome it.

 

[RussianSensation]

That will make the GTX980 to fall to $350 and it will only be 10-15% slower than the 390.

I doubt NV will drop 980 to only $350. For me NV operates in its own world, with its own loyal fans.

XFX R9 290X = $290 USD
Gigabyte 980 = $530 USD for ~15% more performance

Given this, NV can easily sell a $429-449 GTX980 with 10-15% slower performance than an R9 390 $499-549. In fact, NV can sell GTX980 for $499 alongside a $499 R9 390 even if the R9 390 is 15% faster. Like I said, given the market trends in the last 3 years, it's becoming clear the market is separated by NV customers and brand agnostic customers primarily. NV customers would not consider R9 390, no matter the price/performance. Slap 8GB of VRAM on a 980, drop the price just $50 and it'll sell against a 10-15% faster $499 R9 390. As long as reviewers keep pushing perf/watt and ignore after-market AMD cards from AIBs, NV can charge huge premiums for its products via these marketing/PR tactics.

Everyone knew the gtx980 was just a midrange card branded as high end anyways. Still shelled out $650+ for them.

I will NEVER understand customers that loyal to a brand. They know they are getting raped, and welcome it.

What makes it worse is that during Kepler generation when NV launched 680 2GB for $500 and 4GB for $580, a lot of us called it -- mid-range @ high end prices. I bought 7970s but they paid for themselves via bitcoin mining which meant even if each 7970 cost me $999, it was essentially $0 that I spent on it. With 680, that was definitely not the case. Then after $1K Titan --> $650 780 --> $400 R9 290, it was becoming clearer that NV's goal was to delay the true flagship for as long as possible while milking all these in-between SKUs. Before Maxwell generation started, many called that NV would more or less repeat the same strategy. Not only did NV repeat it, but they raised 680's successor price yet again by $50 and they even managed to sell Titan X for $999 without DP -- the key parameter that justified the original Titan's massive premium in the first place! And to think that most of this forum ripped a $549 7970 apart after it absolutely leveled a 580 by 48-80%, while doubling its VRAM for $100 more. Oh, and btw, NV users who recommended/bought GTX680 are now peep quiet and benchmarks don't lie where HD7970Ghz is destroying a 680 by 15-25% in modern titles and approaching a $650 780! Yet, when many talked about GCN's long-term viability and its 3GB of VRAM vs. 2GB for 680, all of that was dismissed. It's hilarious how now a $1K Titan X's price is acceptable since paying for the fastest commands premiums but when 7970 OC destroyed 580 OC by 48-80%, the amount of criticism against it was immense. In hindsight, holly cow how biased some people on this forum are...

More shocking, is the amount of hype 970/980 launch garnered considering after-market cards like Sapphire Tri-X 290 were selling for $360-380 for 6 months before 970 even launched. Even I almost fell for the hype and had my trigger on Gigabyte GTX970s on order but then took a step back and realized wait a second it took NV a whole 10 months to release a card 5% faster than an after-market 290 for only $40-50 less than the market rate of cards like Sapphire Tri-X 290s by Spring 2014. I wouldn't be surprised if 970/980 owners will be upgrading to consumer GM200 6GB cards this year, yet again. NV's bifurcating a generation strategy has resulted in them getting 55%+ gross margins and getting their customers to buy not 1, but 2 GPUs in the same generation. JHH probably can't even himself believe this strategy actually worked! GP204 Pascal at $500-550 is pretty much a lock in.

 

[gamervivek]

I wasn't following the gpu scene at the time but the Titan hype was immense, so I thought it must be quite a card. Looked up its release benchmarks recently and it was only 31% faster than 7970Ghz while costing more than twice as much..

And now it's only around 17% faster...

Though there's the 6GB which has held it in good stead, the DP for some users and apparently it overclocks a tad better. So all is not lost, but still a 1000$ price for that, nvidia got away with daylight robbery.

 

[xthetenth]

Everyone knew the gtx980 was just a midrange card branded as high end anyways. Still shelled out $650+ for them.

I will NEVER understand customers that loyal to a brand. They know they are getting raped, and welcome it.

Some people do research before the fact to make the right decision. Some people do research after the fact to justify the decision they made.

 

[RussianSensation]

I wasn't following the gpu scene at the time but the Titan hype was immense, so I thought it must be quite a card. Looked up its release benchmarks recently and it was only 31% faster than 7970Ghz while costing more than twice as much..

And now it's only around 17% faster...

Though there's the 6GB which has held it in good stead, the DP for some users and apparently it overclocks a tad better. So all is not lost, but still a 1000$ price for that, nvidia got away with daylight robbery.

Barely more than a year since Titan's launch, one could buy 2x Titan level of performance in dual after-market R9 290s and have $200 to spare! 1.5 years after the Titan's launch, one could buy "3X Titans" in 970 Tri-SLI $990. Right now we are about 2 years since the Titan's launch and you can buy the equivalent of "4 Titans" in the form of 4x $250 R9 290s. I remember I said that the Titan was a card in 'no man's land' and I explained why. My reasoning was that it was too fast for older games, but not fast enough for future games. By the time next generation games would come out, we would be able to get Titan's performance for $400-500 or even lower. The rate at which Titan's performance became available at way cheaper prices was even quicker than I had anticipated.

Less than 1 year ago R9 295X2 was $1499 and today you can readily buy it for $650-660. History is going to repeat itself with the Titan X, but this time without DP performance, I feel that its depreciation will be steeper because if a $599 GM200 6GB card or a $650 R9 390X card is 90-95% as fast, who is going to want to buy a Titan X in the used market 15 months from now for more than $500? If R9 390X manages to beat the Titan X in performance, that would be even worse.

 

[Enigmoid]

Meh.

I just don't buy those kinds of high end cards.

 

[boxleitnerb]

And the idiotic comparison between SGPU and SLI/CF continues although people should know better that the perceived performance of AFR is less. Thus the whole performance/$ comparison is [false] in the first place. And if you count games where AFR doesn't work correctly at <=50%, it looks even worse...

Profanity isn't allowed in the technical forums.
--stahlhart

 

[3DVagabond]

Barely more than a year since Titan's launch, one could buy 2x Titan level of performance in dual after-market R9 290s and have $200 to spare! 1.5 years after the Titan's launch, one could buy "3X Titans" in 970 Tri-SLI $990. Right now we are about 2 years since the Titan's launch and you can buy the equivalent of "4 Titans" in the form of 4x $250 R9 290s. I remember I said that the Titan was a card in 'no man's land' and I explained why. My reasoning was that it was too fast for older games, but not fast enough for future games. By the time next generation games would come out, we would be able to get Titan's performance for $400-500 or even lower. The rate at which Titan's performance became available at way cheaper prices was even quicker than I had anticipated.

Less than 1 year ago R9 295X2 was $1499 and today you can readily buy it for $650-660. History is going to repeat itself with the Titan X, but this time without DP performance, I feel that its depreciation will be steeper because if a $599 GM200 6GB card or a $650 R9 390X card is 90-95% as fast, who is going to want to buy a Titan X in the used market 15 months from now for more than $500? If R9 390X manages to beat the Titan X in performance, that would be even worse.

The people who buy Titans, in any iteration, don't care about any of that. Where are the people who justified the price of Titan by it's DP performance now complaining about the Titan-X because it doesn't have it?

And the idiotic comparison between SGPU and SLI/CF continues although people should know better that the perceived performance of AFR is less. Thus the whole performance/$ comparison is [false] in the first place. And if you count games where AFR doesn't work correctly at <=50%, it looks even worse...

That wasn't really the point. It was just a comparison of theoretical perf/$ to draw a picture. It helps to illustrate Titans are incredibly over priced and people are wasting money by buying them, which was the gist of the post.

 

[RussianSensation]

And the idiotic comparison between SGPU and SLI/CF continues although people should know better that the perceived performance of AFR is less. Thus the whole performance/$ comparison is [false] in the first place. And if you count games where AFR doesn't work correctly at <=50%, it looks even worse...

You seem to not understand this simple concept:

1) When SLI/CF doesn't work, if there is a $500 GPU with 87-90% of the performance of a $1000 Titan X, that's HALF the price for 10-13% less performance.

2) When SLI/CF does work, that's 50-70% more performance for a similar price.

In other words, that's a win-win situation because 10-13% less performance will not make Titan X more playable in games, while having a 2nd GPU when CF/SLI works provides a huge improvement in performance.

No one advocates buying dual GPUs with similar performance to a single-chip flagship if the prices are similar. In other words, no one would advise someone to buy GTX760 SLI over 780Ti/R9 290X or GTX970 SLI over a $700 GM200 6GB / R9 390X (assuming it has 90% of the performance of the Titan X). However, it should be obvious that there are clear situations where dual videocard setups are hands down better than a single GPU option. 2 primary examples are $660 GTX970 SLI vs. $550-600 980 OR R9 295X2 $650 vs. $550-600 980 for 4K gaming. Your desire to always ignore multi-GPU setups as irrelevant dismisses specific cases where dual-GPU setups are simply the better option.

The minute there is a card with 87-90% of the performance of a Titan X for $500-550, the Titan X is history/irrelevant. You don't have to agree but many will. As 3DVagabond mentioned, the Titan X at $1K is grossly overpriced. It's way out of line based on the price/technology curve and it's obvious to anyone who follows the GPU industry. Even though there is a small group of enthusiasts who always buy the best, it doesn't change the facts that a cheaper priced GM200 6GB is inevitable, as well as a way superior price/performance AMD R9 390 option.

The people who buy Titans, in any iteration, don't care about any of that. Where are the people who justified the price of Titan by it's DP performance now complaining about the Titan-X because it doesn't have it?

Ya, I get that. There are going to be people buying 3-4 Titan Xs. However, in 3-6 months from now, there will be cards with similar performance for $600-700. If someone is in no rush (i.e., has a backlog of games), is still waiting for FreeSync vs. GSync monitor debate to settle down, and could care less about bragging rights, the Titan X is an easy skip. Again, many called that the original Titan a huge rip-off and a card that was launched at the wrong time for gaming; and that turned out to be 100% true. The Titan X's performance increase over the 980 is less than the original Titan's was over the 680 but the price is the same. That makes the Titan X's relative performance standing even worse, especially so since the rumoured specs for an R9 390X are simply beastly.

Don't forget that some gamers buy 2 GPUs when they upgrade. If for example I can get 87-90% of the performance of Titan X SLI in R9 390 nonX CF/GM200 6GB SLI for 40-50% less, it's worth waiting for 3-6 months. Some people disagree, but for me it's also the principle - I won't support $999 single GPU prices even though I can afford it because I think sending a signal to NV that these prices are OK does more damage to the GPU industry then it helps.

 

[5150Joker]

Wait did i miss something here? Which AMD gpu is within 10-15% of Titan X?

 

[RussianSensation]

Wait did i miss something here? Which AMD gpu is within 10-15% of Titan X?

This is a hypothetical discussion. In the past, a 2nd tier flagship card from AMD came within 15% of the fastest card from NV. HD6950 2GB unlocked vs. 580, HD7950 OC vs. 680 OC, R9 290 OC vs. 780Ti OC all come to mind. With the specs of R9 390X, and an inevitable GM200 6GB, it's more or less guaranteed that there will be some GPU from AMD or NV priced at $550-600 that will offer 85-90% of the Titan X performance. For those who can afford the Titan X and already bought one, this doesn't matter. For the rest of us, history repeats itself. Just 9 months after the original $1K Titan launched, AMD delivered that level of performance in a $400-550 after-market R9 290/X. Since this time the Titan X is even less impressive performance wise compared to the 980 than the original Titan was vs. the 680, there is an even greater possibility that R9 390/390X will offer far superior price/performance by June 2015, which is only 3 months away.

 

[DearLord]

This is a hypothetical discussion. In the past, a 2nd tier flagship card from AMD came within 15% of the fastest card from NV. HD6950 2GB unlocked vs. 580, HD7950 OC vs. 680 OC, R9 290 OC vs. 780Ti OC all come to mind. With the specs of R9 390X, and an inevitable GM200 6GB, it's more or less guaranteed that there will be some GPU from AMD or NV priced at $550-600 that will offer 85-90% of the Titan X performance. For those who can afford the Titan X and already bought one, this doesn't matter. For the rest of us, history repeats itself. Just 9 months after the original $1K Titan launched, AMD delivered that level of performance in a $400-550 after-market R9 290/X. Since this time the Titan X is even less impressive performance wise compared to the 980 than the original Titan was vs. the 680, there is an even greater possibility that R9 390/390X will offer far superior price/performance by June 2015, which is only 3 months away.

While I basically agree with you, the problem AMD has is that, in those 3 (more realistically 3-5) months, we've got 2 huge titles coming in Batman and Witcher 3 (likely the 2 biggest of the year). One of these is being offered as a latter day pack-in for the 900 series. So even those who would prefer to wait for the 390/x (or even the 980ti) might just decide not to. Certainly the college and lower aged gamers have an easier time whiling away the days gaming in summer also, meaning they'll be chomping at the bit for a state-of-the-art, affordable card that won't require updating their psu.

 

[3DVagabond]

While I basically agree with you, the problem AMD has is that, in those 3 (more realistically 3-5) months, we've got 2 huge titles coming in Batman and Witcher 3 (likely the 2 biggest of the year). One of these is being offered as a latter day pack-in for the 900 series. So even those who would prefer to wait for the 390/x (or even the 980ti) might just decide not to. Certainly the college and lower aged gamers have an easier time whiling away the days gaming in summer also, meaning they'll be chomping at the bit for a state-of-the-art, affordable card that won't require updating their psu.

There's no reason to think that the 390 will require a PSU upgrade any more than the 980ti. It's being reported now that the 980ti won't be out until after Summer though.

 

[DearLord]

There's no reason to think that the 390 will require a PSU upgrade any more than the 980ti. It's being reported now that the 980ti won't be out until after Summer though.

Yeah I was thinking of the 290/x vs the 970

 

[Dug]

I do not think it has anything to do with being an environmentalist. People just want to have something to complain about. If the only difference was one card used a phillips head screw while the other had flatheads that is what they would complain about. BTW, phillips all the way, anyone who says otherwise is not an enthusiast.

Sorry, I think a square recess is superior

 

[Silverforce11]

As for overclocked vs stock. I have no idea what your talking about. I bought the 980 and fully intended to overclock it.

I think you are missing the point.

If you have a $500 980 and OC to reach 15% to a Titan X stock, its still less than the potential $500 AMD card RS was predicting, $500 at STOCK within 15% of Titan X.

As such, OC, it ~= Titan X for $500.

Capiche?

ps. R290X with 10% OC ~= 980. For <$300. ???

 

[digitaldurandal]

While I basically agree with you, the problem AMD has is that, in those 3 (more realistically 3-5) months, we've got 2 huge titles coming in Batman and Witcher 3 (likely the 2 biggest of the year). One of these is being offered as a latter day pack-in for the 900 series. So even those who would prefer to wait for the 390/x (or even the 980ti) might just decide not to. Certainly the college and lower aged gamers have an easier time whiling away the days gaming in summer also, meaning they'll be chomping at the bit for a state-of-the-art, affordable card that won't require updating their psu.

Batman and Witcher 3 are the two biggest titles of the year?

Witcher 2 didn't come anywhere near biggest title, and Batman sales vs COD?

I know the new Batman is supposed to be somewhat of an upgrade visually and allow more enemies on the screen however the recommended system requirements I found online say 6950/660. I don't see this being a game pushing users to purchase a Titan X if they were thinking about waiting.

Additionally both games you name are Gameworks titles. If users are of the opinion that these are the two biggest releases of the year, I don't see them purchasing AMD.

 

[Dug]

You seem to not understand this simple concept:

1) When SLI/CF doesn't work, if there is a $500 GPU with 87-90% of the performance of a $1000 Titan X, that's HALF the price for 10-13% less performance.

2) When SLI/CF does work, that's 50-70% more performance for a similar price.

In other words, that's a win-win situation because 10-13% less performance will not make Titan X more playable in games, while having a 2nd GPU when CF/SLI works provides a huge improvement in performance.

No one advocates buying dual GPUs with similar performance to a single-chip flagship if the prices are similar. In other words, no one would advise someone to buy GTX760 SLI over 780Ti/R9 290X or GTX970 SLI over a $700 GM200 6GB / R9 390X (assuming it has 90% of the performance of the Titan X). However, it should be obvious that there are clear situations where dual videocard setups are hands down better than a single GPU option. 2 primary examples are $660 GTX970 SLI vs. $550-600 980 OR R9 295X2 $650 vs. $550-600 980 for 4K gaming. Your desire to always ignore multi-GPU setups as irrelevant dismisses specific cases where dual-GPU setups are simply the better option.

The minute there is a card with 87-90% of the performance of a Titan X for $500-550, the Titan X is history/irrelevant. You don't have to agree but many will. As 3DVagabond mentioned, the Titan X at $1K is grossly overpriced. It's way out of line based on the price/technology curve and it's obvious to anyone who follows the GPU industry. Even though there is a small group of enthusiasts who always buy the best, it doesn't change the facts that a cheaper priced GM200 6GB is inevitable, as well as a way superior price/performance AMD R9 390 option.



Ya, I get that. There are going to be people buying 3-4 Titan Xs. However, in 3-6 months from now, there will be cards with similar performance for $600-700. If someone is in no rush (i.e., has a backlog of games), is still waiting for FreeSync vs. GSync monitor debate to settle down, and could care less about bragging rights, the Titan X is an easy skip. Again, many called that the original Titan a huge rip-off and a card that was launched at the wrong time for gaming; and that turned out to be 100% true. The Titan X's performance increase over the 980 is less than the original Titan's was over the 680 but the price is the same. That makes the Titan X's relative performance standing even worse, especially so since the rumoured specs for an R9 390X are simply beastly.

Don't forget that some gamers buy 2 GPUs when they upgrade. If for example I can get 87-90% of the performance of Titan X SLI in R9 390 nonX CF/GM200 6GB SLI for 40-50% less, it's worth waiting for 3-6 months. Some people disagree, but for me it's also the principle - I won't support $999 single GPU prices even though I can afford it because I think sending a signal to NV that these prices are OK does more damage to the GPU industry then it helps.

But all you are talking about is future performance and prices. Most of us live right now. If you wait, yes there will be be something comparable in 3-6 months for 40-50% of the cost. You can keep doing that for the rest of your life. Why wait? Most of us will enjoy those 3-6 months of performance while you wait.

Point is, people buying a Titan X right now get to enjoy it. Once something comes out 3-6 months later at a cheaper price doesn't mean Titan X owners all of a sudden don't enjoy their purchase.

This is true in any industry. I buy a car in 2015 at certain price. In 2016 a new car comes out with more features at a better price. In 2017 it has even more features and the same or better price. At what point do you buy a car? Should I just keep waiting?

 

[soccerballtux]

am I the only one that has trouble believing they're 'just ironing out the drivers'? wait they don't have their own fabs they just contracted them so it's not like they'd be wasting downtime sitting around not fabbing more. But still, bad to have product sitting...expensive

 

[digitaldurandal]

am I the only one that has trouble believing they're 'just ironing out the drivers'? wait they don't have their own fabs they just contracted them so it's not like they'd be wasting downtime sitting around not fabbing more. But still, bad to have product sitting...expensive

They probably pushed the release to try to get some users to wait with the promise of a card around the corner.

 

[rgallant]

But all you are talking about is future performance and prices. Most of us live right now. If you wait, yes there will be be something comparable in 3-6 months for 40-50% of the cost. You can keep doing that for the rest of your life. Why wait? Most of us will enjoy those 3-6 months of performance while you wait.

Point is, people buying a Titan X right now get to enjoy it. Once something comes out 3-6 months later at a cheaper price doesn't mean Titan X owners all of a sudden don't enjoy their purchase.

This is true in any industry. I buy a car in 2015 at certain price. In 2016 a new car comes out with more features at a better price. In 2017 it has even more features and the same or better price. At what point do you buy a car? Should I just keep waiting?

doing what that todays cards can't do ? if in sli yes but 1x a 1k card not much imo.

 

[digitaldurandal]

But all you are talking about is future performance and prices. Most of us live right now. If you wait, yes there will be be something comparable in 3-6 months for 40-50% of the cost. You can keep doing that for the rest of your life. Why wait? Most of us will enjoy those 3-6 months of performance while you wait.

Point is, people buying a Titan X right now get to enjoy it. Once something comes out 3-6 months later at a cheaper price doesn't mean Titan X owners all of a sudden don't enjoy their purchase.

This is true in any industry. I buy a car in 2015 at certain price. In 2016 a new car comes out with more features at a better price. In 2017 it has even more features and the same or better price. At what point do you buy a car? Should I just keep waiting?

I do agree with your statement. It is a question of value for each user though. The question is, if you believe that a competitive card will be out in less than 6 months for 2/3 of the price. Is it worth $350 to you to have the performance for 6 months or less longer.

Of course we don't know for sure what performance AMD's card will bring or when the 980 ti will release.

I will say, if I knew I could wait a model year on my vehicle and get most of the same features for 2/3 the price I would wait . Then again we are talking about a lot more money.

I don't normally purchase the very best because they devalue so quickly. I would rather spend 600 bucks this year (2 cards) and 600 bucks in 1 or 2 years than spend 1200 and keep cards for longer.

 

[boxleitnerb]

The people who buy Titans, in any iteration, don't care about any of that. Where are the people who justified the price of Titan by it's DP performance now complaining about the Titan-X because it doesn't have it?


That wasn't really the point. It was just a comparison of theoretical perf/$ to draw a picture. It helps to illustrate Titans are incredibly over priced and people are wasting money by buying them, which was the gist of the post.

But that doesn't work at all since he blatantly ignores that there is no comparable performance between these solutions. 40 fps on Titan X should feel like approximately 60 fps in SLI 980 for instance. Because of AFR.

Then there is input lag, graphic glitches, features not working with MGPU (DSR+Gsync+sli), certain AA modes (crysis 2) and the wait for profiles.

All these factors are important and have to be considered which he doesn't because he has no first hands experience with MGPU like I and others do.
Yet he posts about this topic again and again. :thumbsdown: