What does AMD have to fight the 780?

[hyrule4927]

Was on Alienware notebooks since going to med school and now that I'm almost done, I'm back on desktop. Is gaming on a DTR somehow inferior in your eyes?

Considering the amount of money you threw away on those Alienware machines you could have easily had a solid desktop and a much more mobile (but still capable of gaming) laptop the entire time.

 

[NIGELG]

No they don't, the want AMD right where they have them.

And you know this how??

 

[omeds]

And you know this how??

Because it benefits them to not have AMD go under, and are currently in a very strong position.

 

[tential]

I don't follow the logic here. Yes the titan is clearly better than the 7970. It might only be the top card for ~6-9 months. When the 9970 launches it will likely be better (we have no idea if it's a mammoth die, or just a refined 7970). The titan successor is likely a year out or possibly 2 if it's as delayed as the titan. Yes the titan architecture is better than the 7970, but the crown will switch.

It's not like cpus where intel is the 100% of the time (since i7) leader. They don't lose at each refresh, actually they hardly refresh their cpus. (+10% wow) Anyways, who cares about cpus when talking about gpus.

/ot

So you follow perfectly then. I said the crown switches, no company is going to have it all the time since both companies dont release cards at the exact same time. I said exactly what you just said.

 

[NIGELG]

Because it benefits them to not have AMD go under, and are currently in a very strong position.

That is just your speculation.

But I understand why many Nvidians think like you do.


''We can kill and bury you....if we wanted to but we are a nice company so we would not do that even though you are a rival in a capitalist market....you are a pathetic rival not worth the effort to finish off''

 

[SiliconWars]

Because it benefits them to not have AMD go under, and are currently in a very strong position.

You really don't know what you're talking about.

 

[omeds]

But I understand why many Nvidians think like you do.

Must you always sink so low with name calling? For the record I prefer AMD as a company, and have owned far more AMD products than Nvidia.

''We can kill and bury you....if we wanted to but we are a nice company so we would not do that even though you are a rival in a capitalist market....you are a pathetic rival not worth the effort to finish off''

Who said they can kill and bury them?

You really don't know what you're talking about.

Don't I?

 

[Shmee]

Back on topic please.

 

[Lepton87]

Unfortunately for AMD even if they have a cheaper and faster card people still buy nv's cards. Most of the time for a given budget AMD's offering is a better value yet NV still sells a hell of a lot more cards then AMD.This generation they have been selling a mid-range card for last generation's high-end's price and pretended that this is their high-end card and a lot of people even on this forum fell for it, yet it was painfully apparent that's not the case. There were many arguments on this forum about this. I would very like AMD to compete in this high-end market but I realize it's not economically feasible for them. Even if they matched Titan and sell that card for 780's price, people would still buy more 780s and even more Titans. On the other hand it's quite depressing to see that not only AMD stopped competing with High-end Intel CPUs but it seems like they are set on doing the same with their GPUs. There was no point in countering the Titan with it's ridiculous 1000$ price tag it was like an Intel Extreme Edition CPU, but 780 isn't priced sky-high, it's quite in-line with previous generation high-end cards. It's a bit more expensive but when you consider inflation it's not really that much more expensive. The best counter I see is mastering CF, if CF worked every time and the fps it delivered would feel just as good as a single GPU card's FPS then that would be awesome. TRI-way 7950 CF would mop the floor with a 780 and Titan for not much more then 780 costs. Even two would be faster.

 

[SiliconWars]

Last time I heard that AMD had stopped competing with Nvidia on the high end, they released the 4870 which nearly matched the gtx 280 at much smaller die size, TDP and half the price, then they obliterated Nvidia with the 4870x2. After that the 5-series crushed Fermi on all fronts for almost 18 months.

History has this weird way of repeating itself especially for those who didn't learn from their last mistake.

 

[Lepton87]

Last time I heard that AMD had stopped competing with Nvidia on the high end, they released the 4870 which nearly matched the gtx 280 at much smaller die size, TDP and half the price, then they obliterated Nvidia with the 4870x2. After that the 5-series crushed Fermi on all fronts for almost 18 months.

History has this weird way of repeating itself especially for those who didn't learn from their last mistake.

It was so bad that NV gave rebates to early adopters of GTX280. They didn't have to do that. What's funny right now they did just that themselves with 780. It's practically as fast as a Titan at 40% lower price point. Pretty much exactly the same scenario as with GTX280 and 4870.

 

[omeds]

After that the 5-series crushed Fermi on all fronts for almost 18 months.

Ehem, refer to your previous post...

You really don't know what you're talking about.

 

[Lepton87]

fter that the 5-series crushed Fermi on all fronts for almost 18 months.

5 series didn't crush Fermi on all fronts but it came out a whole 6 months later so AMD had no competition for half a year. I had 5870 at the time, it was not worth it to upgrade it to GTX480 for 15% better performance and hugely lower performance per watt. So 5-series crushed fermi on time to market and performance per watt and the flagship was just a bit slower. For raw performance GTX480 was better then 5870 but for that additional 15% you had to pay heavily in terms of noise and power consumption. GTX580 made performance per watt more reasonable.

 

[SiliconWars]

5-series was AMD's finest moment. That was when they overtook Nvidia on market share, eyefinity was the thing to have and Nvidia could only match the 5870 with the 560 Ti 15 months later (with similar die sizes/tdp).

 

[amenx]

Last time I heard that AMD had stopped competing with Nvidia on the high end, they released the 4870 which nearly matched the gtx 280 at much smaller die size, TDP and half the price, then they obliterated Nvidia with the 4870x2.

For a short while... until the GTX295 arrived which obliterated it.

 

[amenx]

5-series was AMD's finest moment. That was when they overtook Nvidia on market share, eyefinity was the thing to have and Nvidia could only match the 5870 with the 560 Ti 15 months later (with similar die sizes/tdp).

Its pretty easy to beat a competitor in market share when they dont have their product line out. I recall those days, 5870 owners gleefully linking to benches where their cards beat the older gen gtx285 . The 560 vs 5870 comparison is not really proper, unless you think die size/tdp are the only things enthusiasts care about. As if the GPU race was only about that.

 

[SiliconWars]

For a short while... until the GTX295 arrived which obliterated it.

Yeah I remember how well it stood up at 1600p with its 768mb memory.

 

[SiliconWars]

Its pretty easy to beat a competitor in market share when they dont have their product line out. I recall those days, 5870 owners gleefully linking to benches where their cards beat the older gen gtx285 . The 560 vs 5870 comparison is not really proper, unless you think die size/tdp are the only things enthusiasts care about. As if the GPU race was only about that.

I remember 5870 buyers enjoying Eyefinity gaming for over a year while Nvidia struggled to get the laughable 480 out months later before they finally got their own bad version of Eyefinity another 6 months later, yeah.

 

[Granseth]

Unfortunately for AMD even if they have a cheaper and faster card people still buy nv's cards. Most of the time for a given budget AMD's offering is a better value yet NV still sells a hell of a lot more cards then AMD.This generation they have been selling a mid-range card for last generation's high-end's price and pretended that this is their high-end card and a lot of people even on this forum fell for it, yet it was painfully apparent that's not the case. There were many arguments on this forum about this. I would very like AMD to compete in this high-end market but I realize it's not economically feasible for them. Even if they matched Titan and sell that card for 780's price, people would still buy more 780s and even more Titans. On the other hand it's quite depressing to see that not only AMD stopped competing with High-end Intel CPUs but it seems like they are set on doing the same with their GPUs. There was no point in countering the Titan with it's ridiculous 1000$ price tag it was like an Intel Extreme Edition CPU, but 780 isn't priced sky-high, it's quite in-line with previous generation high-end cards. It's a bit more expensive but when you consider inflation it's not really that much more expensive. The best counter I see is mastering CF, if CF worked every time and the fps it delivered would feel just as good as a single GPU card's FPS then that would be awesome. TRI-way 7950 CF would mop the floor with a 780 and Titan for not much more then 780 costs. Even two would be faster.

People seems to have the notion that gk104 is a midrange chip and Tahiti is a high end chip when Tahiti is only 17 percent bigger. And while the compute abilities makes the Tahiti more professional oriented it's not that much difference, and not unreasonable to think that Tahitis 17 % has largely gone towards compute abilities, and a wider memory bus.
I think nVidia learned from Fermi and new that they wouldn't be able to bring out "big" Kepler from the start, I think they where able to guess at AMD making a 300-350mm2 chip and hit fairly well on what they needed to compete with that.

I agree with you on that since apparently AMD is probably not going to bring out new chips, the next best thing would be to put work into drivers, and continue supporting game companies to put compute shaders into games since they do very well in those games. It might not be a hard sell either since XB1 and PS probably will use that as well to bring better effects into new games.

 

[amenx]

Yeah I remember how well it stood up at 1600p with its 768mb memory.

I wonder how many people were using 2560x1600 screens in 2009 . Not only that, but different sites show different results, even at that res.

http://www.anandtech.com/show/2708/4

 

[amenx]

I remember 5870 buyers enjoying Eyefinity gaming for over a year while Nvidia struggled to get the laughable 480 out months later before they finally got their own bad version of Eyefinity another 6 months later, yeah.

Yeah, 5870 users had eyefinity. Well, the 1% of them who could run it by using the power of 2 cards or more to run it properly. Then they had to deal with Crossfire issues which were much worse than the poor state they are in today. :whiste:

 

[Lepton87]

People seems to have the notion that gk104 is a midrange chip and Tahiti is a high end chip when Tahiti is only 17 percent bigger. And while the compute abilities makes the Tahiti more professional oriented it's not that much difference, and not unreasonable to think that Tahitis 17 % has largely gone towards compute abilities, and a wider memory bus.
I think nVidia learned from Fermi and new that they wouldn't be able to bring out "big" Kepler from the start, I think they where able to guess at AMD making a 300-350mm2 chip and hit fairly well on what they needed to compete with that.

I agree with you on that since apparently AMD is probably not going to bring out new chips, the next best thing would be to put work into drivers, and continue supporting game companies to put compute shaders into games since they do very well in those games. It might not be a hard sell either since XB1 and PS probably will use that as well to bring better effects into new games.

That GTX680 is a mid-range chip is vindicated by the release of 780. Titan is just a halo product. I'd call GTX680 very fast clocked big mid-range chip and Tahiti small high-end chip that is clocked more conservatively. 384 bit bus has a lot to do with that, if Tahiti had 256bit bus I would probably also consider it very upper mid-range, but focusing so much on compute would suggest otherwise. Tahiti FP64 capabilities also has a lot to do with considering it high-end. GK104 is a lean mean gaming chip without additional wasteful FP64 and other compute capabilities like dynamic scheduler etc. Otherwise I agree.

 

[bystander36]

If AMD is going to compete, they need to change their reputation, and that takes time. Their performance is fine right now, but they still have a bad reputation because of drivers. For most the life of the 7000 series, they have experience stuttering in single GPU setups, and still microstutter issues in multi-GPU, and until the end of 2012, they were slower.

If you are buying a single Radeon now, it makes since, but if you bought it a year ago, you would have had an inferior experience. If you go crossfire, you'd still have an inferior experience. I imagine most people have purchased a least a couple AMD/ATI cards in the past, and also Nvidia products and have noticed the difference. I know I have. But after you've experience both sides, most people tend to prefer Nvidia, and then just stay with them.

AMD has improved a lot, but they are still behind, and when they finally catch up, if they do, it will still take a few years before people give them a fair shot, due to their reputation in the past.

 

[SolMiester]

wow...not everyone needs the fastest and greatest....the 7990 is a wickedly fast card, it just needs some tweaks..
AMD have a great card in the 7970 at a price most can stretch too.

 

[Granseth]

That GTX680 is a mid-range chip is vindicated by the release of 780. Titan is just a halo product. I'd call GTX680 very fast clocked big mid-range chip and Tahiti small high-end chip that is clocked more conservatively. 384 bit bus has a lot to do with that, if Tahiti had 256bit bus I would probably also consider it very upper mid-range, but focusing so much on compute would suggest otherwise. Tahiti FP64 capabilities also has a lot to do with considering it high-end. GK104 is a lean mean gaming chip without additional wasteful FP64 and other compute capabilities like dynamic scheduler etc. Otherwise I agree.

That Tahiti is more pro oriented is nothing I disagree with you about, and that Tahiti is more universal`in what it can do. But pretending nVidia hadn't planed on using gk104 as their top chip for gaming seems strange. And isn't it high end if it isn't their top chip?
I think nVidia with a few exceptions is very good at doing business, and found out that they wanted to do it like this quite early on, especially after gf100.

And the reason I argue about this is because people makes it seem like Tahiti should be that much better at gaming because it's supposed to be high-end when in reality it's not that much difference in the number of transistors they have to use on graphics.
Now of course it's different when they have made their refresh, gk110, that is their new high-end chip.