R260/270/280/290/290x Review thread

[VulgarDisplay]

Agree no big dies on 20nm before a year.
Nv will spend a year trailing amd on the high end for both perf and mm2 eff. But 104 is best mm2 perf vs tahiti for the performance segment.
Everyone and his phone brother will pay for 20nm for their a57 derivatives and i think nv and amd will slow down adoption of new nodes for economic reason for the big dies for tvat reason. Mobile could probably be different.
I dont think we will see the same pattern for 20nm that we have seen for prior nodes.

Nvidia will not trail AMD if they don't want to. If they decide to swallow their profit margins and release a full kepler die at a reasonable price for gamers they will take back the lead.

Just have to wait and see if they will actually do it.

 

[2is]

It isn't loud in default configuration, I linked Anandtech's noise measurements. It's a blower so it already has that working against it in terms of noise.

Yes, it is. I've read the AT review. It's one of, if not THE loudest card in their lineup. The card throttles a lot even in Uber mode.

 

[Will Robinson]

If only you could count on it staying at that speed longer than a couple minutes during gaming....

And yet just pages ago you were telling how you were rushing out to buy two...😉

 

[Ajay]

Ocuk has a Prolimatech MK-26 cooler dropping temps by about 30 degrees C compared to the stock cooler.

http://forums.overclockers.co.uk/showthread.php?p=25177410

So $750 US (We don't pay VAT?) and 4 PCIe slots! They got it to run @ 1200 MHz and 1.4 v - we must be talking early GTX 480 O/C power numbers here D:. Asus should be able to make something pretty nice, though less aggressive, with their 3 x PCIe wide Matrix line.

I think it's a great card for extreme users who can afford the necessary PSUs and aftermarket GFX cooling. Releasing the 290X so cheap on a reference cooler is a bonus for such users. It seems like a bit of a disservice to those with more modest budgets - but hey, it's a free country. I remember running a pair of K7s for F@H on heatsinks with 60 mm Delta fans - talk about ear bleed - I only ran them at night in my office (at the other end of the house) with both the office and bedroom doors closed :twisted:

 

[Will Robinson]

But you are smarter than all of them right? Even though they have evidence and you don't.

Yup hes out thought them all

Originally Posted by ShintaiDK View Post
There will be no APU in PS4 and Xbox720.

:whiste:

Your posting privileges are going up for staff review. I've had it with your incessant troublemaking.
-- stahlhart

 

[AdamK47]

Nvidia will not trail AMD if they don't want to. If they decide to swallow their profit margins and release a full kepler die at a reasonable price for gamers they will take back the lead.

Just have to wait and see if they will actually do it.

They won't be releasing a full Kepler with Maxwell right around the corner. If anything, this may prompt nVidia to hurry the release of Maxwell.

 

[GaiaHunter]

So $750 US (We don't pay VAT?) and 4 PCIe slots! They got it to run @ 1200 MHz and 1.4 v - we must be talking early GTX 480 O/C power numbers here D:. Asus should be able to make something pretty nice, though less aggressive, with their 3 x PCIe wide Matrix line.

I think it's a great card for extreme users who can afford the necessary PSUs and aftermarket GFX cooling. Releasing the 290X so cheap on a reference cooler is a bonus for such users. It seems like a bit of a disservice to those with more modest budgets - but hey, it's a free country. I remember running a pair of K7s for F@H on heatsinks with 60 mm Delta fans - talk about ear bleed - I only ran them at night in my office (at the other end of the house) with both the office and bedroom doors closed :twisted:

£490 - £439 card+£51 for the cooler.

 

[Black Octagon]

I hope nv does NOT rush Maxwell, personally. I think AMD and NV have taken 28nm pretty dang far, looking at the 290x. I hope the next gen is based on 20nm and done properly.

Anyway, right now my attention is on details about the 785 Ti. Then, it will be on aftermarket 290x cards, and then it will be on how the 290x performs 'in real life' with Mantle utilised.

Then, I'm all for getting hot and sweaty about 20nm-based Maxwell

 

[Erenhardt]

I hope nv does NOT rush Maxwell, personally. I think AMD and NV have taken 28nm pretty dang far, looking at the 290x. I hope the next gen is based on 20nm and done properly.

Anyway, right now my attention is on details about the 785 Ti. Then, it will be on aftermarket 290x cards, and then it will be on how the 290x performs 'in real life' with Mantle utilised.

Then, I'm all for getting hot and sweaty about 20nm-based Maxwell

Yep, the way they release card now, there is always something on the horizon

 

[ICDP]

The 290 has incredible potential. Give credit where it is due. But that cooler makes me just want to line up the AMD engineering staff and smack the dumb looks off their faces. To throw a cooler like that on an engineering feat like the 290? It's criminal. How does something like this get through? What are they thinking?

LOL, so true. Someone must have told them this was a bad idea, yet here we are. I wonder if the AMD engineers are sitting around wondering why people are saying the cooler sucks... er blows... you know what I mean. :biggrin:

 

[Vesku]

LOL, so true. Someone must have told them this was a bad idea, yet here we are. I wonder if the AMD engineers are sitting around wondering why people are saying the cooler sucks... er blows... you know what I mean. :biggrin:

It's clearly a cost decision, it took a lot of man hours to come up with the updated powertune to let it beat the 780 on that cooler.

 

[thilanliyan]

Nvidia will not trail AMD if they don't want to. If they decide to swallow their profit margins and release a full kepler die at a reasonable price for gamers they will take back the lead.

Oh yeah, I forgot, nVidia's REAL high end isn't here yet.

Just wait...it's coming:

Their GK110 die must be magical!! Sprouting CUDA cores left and right to stave off the evil, power hungry (pun intended lol) 290X.
😀

 

[ozzy702]

They won't be releasing a full Kepler with Maxwell right around the corner. If anything, this may prompt nVidia to hurry the release of Maxwell.

They can't, not within the next year anyway. Anyone that thinks that "big die" 20nm is coming in 2014 is fooling themselves. GK104 style chips on 20nm will be all we see until the node is matured and with that we won't see huge jumps in performance from these big die 28nm chips.

 

[JDG1980]

The 290 has incredible potential. Give credit where it is due. But that cooler makes me just want to line up the AMD engineering staff and smack the dumb looks off their faces. To throw a cooler like that on an engineering feat like the 290? It's criminal. How does something like this get through? What are they thinking?

They probably figured that since almost no one buys the reference cards after the first couple months, it would be cheaper to just let Asus, MSI, and Gigabyte handle the cooling problems.

Keep in mind that the Titan/780 cooler costs about $60, and that's just manufacturing costs, not including how much it cost Nvidia to design it in the first place. AMD wants to hit a very aggressive price point on the R9 290X and they succeeded.

 

[ShintaiDK]

They won't be releasing a full Kepler with Maxwell right around the corner. If anything, this may prompt nVidia to hurry the release of Maxwell.

Maxwell comes with 20nm. 20nm is roughly a year away. Both AMD and nVidia hungers for 20nm. But this is the new reality. 3 year node cycles.

 

[OCGuy]

Maxwell comes with 20nm. 20nm is roughly a year away. Both AMD and nVidia hungers for 20nm. But this is the new reality. 3 year node cycles.

People keep ignoring TSMC updates.........

AMD had 7900 ready to go in Jan '12 on a very young 28nm.

TSMC said this week 20nm will ramp 30% faster than 28nm, and 2014 20nm revenue will be greater than 2012 28nm.

Now this doesn't mean the REAL Hawaii or Maxwell will be in our Easter baskets, but the predictions of Q4 are probably just as inaccurate.

 

[SlowSpyder]

Maxwell comes with 20nm. 20nm is roughly a year away. Both AMD and nVidia hungers for 20nm. But this is the new reality. 3 year node cycles.

I wonder if this generation's release cycle will be the new norm as there is more time between new manufacturing nodes being ready. That is the high end parts for the first year or so will be smaller GPU's and not until the process matures and the costs of manufacturing come down a bit will we see the bigger dies.

 

[Keysplayr]

They probably figured that since almost no one buys the reference cards after the first couple months, it would be cheaper to just let Asus, MSI, and Gigabyte handle the cooling problems.

Keep in mind that the Titan/780 cooler costs about $60, and that's just manufacturing costs, not including how much it cost Nvidia to design it in the first place. AMD wants to hit a very aggressive price point on the R9 290X and they succeeded.

Well congratulations to AMD for reaching whatever price point. But I think we all know this costs them in the long run. Brand recognition doesn't come easy, and this is a sure way to deter that from happening. If you're going to be smart (design an excellent GPU) then be smart the whole way. Don't walk up 20 inches from the finish line and cut your own achilles tendons. It's completely not self serving and AMD needs to be self serving if they want the appreciation, not condemnation, of their consumers.

 

[Vesku]

2014 20nm revenue will be greater than 2012 28nm.

Node discussion should be in it's own thread but I will point out that this part of TSMCs statement is easily true simply from increased cost to their customers for 28nm wafers. Higher cost = greater revenue even with equal or less wafers sold.

 

[Vesku]

Well congratulations to AMD for reaching whatever price point. But I think we all know this costs them in the long run. Brand recognition doesn't come easy, and this is a sure way to deter that from happening. If you're going to be smart (design an excellent GPU) then be smart the whole way. Don't walk up 20 inches from the finish line and cut your own achilles tendons. It's completely not self serving and AMD needs to be self serving if they want the appreciation, not condemnation, of their consumers.

Their AIBs will most certainly be delivering custom solutions. It's my guess that AMD has these reference only launch restrictions because there needs to be some civility among their AIBs at launch or some might go rogue and offer price cuts, in the form of custom solutions at reference prices. Look at the blood spilled when MSI and EVGA were going at it over custom Kepler voltage control.

 

[OCGuy]

Well congratulations to AMD for reaching whatever price point. But I think we all know this costs them in the long run. Brand recognition doesn't come easy, and this is a sure way to deter that from happening. If you're going to be smart (design an excellent GPU) then be smart the whole way. Don't walk up 20 inches from the finish line and cut your own achilles tendons. It's completely not self serving and AMD needs to be self serving if they want the appreciation, not condemnation, of their consumers.

I think the delay/low availability/removing base clock of 290x all have to do with engineering, QC, and marketing trying to land on the right final specs. The cooler may have looked much better with a lower clock target.

When 290x was going to launch previously, Newegg had it listed higher (not the $9,999.00 place holder).

I would be interested in the die statistics and a breakdown of just what this card actually costs AMD to give to an AIB to sell for $550. 4GB of VRAM, 512Mbit interface, and obviously beefy circuitry are not something to sneeze at.

They are playing it smart. Halo cards never make up more than 1% of the market, so taking a hit to your margin on your new product in order to increase volume on your high-margin, high-market-share mature products just makes sense.

 

[VulgarDisplay]

People keep ignoring TSMC updates.........

AMD had 7900 ready to go in Jan '12 on a very young 28nm.

TSMC said this week 20nm will ramp 30% faster than 28nm, and 2014 20nm revenue will be greater than 2012 28nm.

Now this doesn't mean the REAL Hawaii or Maxwell will be in our Easter baskets, but the predictions of Q4 are probably just as inaccurate.

Yes they are ramping up faster, but phone makers are buying up all the capacity early on.

 

[Zargon]

And of course... cheaper.


But you know... no G-synch.

Some of these guys remind me of the Mike Meyers "Scottish" SNL sketch.

#1 reason I am not buying one right there

need to see how g-sync plays out, esp since my 7950's should be fine for what I am doing.

if it werent for gsync I would be seriously considering ordering a new case and a ref card, and then selling my 7950s to buy a second

 

[krumme]

Well congratulations to AMD for reaching whatever price point. But I think we all know this costs them in the long run. Brand recognition doesn't come easy, and this is a sure way to deter that from happening. If you're going to be smart (design an excellent GPU) then be smart the whole way. Don't walk up 20 inches from the finish line and cut your own achilles tendons. It's completely not self serving and AMD needs to be self serving if they want the appreciation, not condemnation, of their consumers.

True. It shouldnt be difficult to release a non blower ref design and get 5-10% performance for free at lower noise levels. It dont get it. Its more or less the same cost. Making an effective blower design with this kind of performance looks counter productive.

The sad thing is we knew it in advance. I wrote they would release it with bad cooler and lower clocks. They did, they always do.

The surpricing thing is this gpu is so damn fast. Its got huge legs and i didnt expect them to go so far.

 

[ozzy702]

Yes they are ramping up faster, but phone makers are buying up all the capacity early on.

Bingo. AMD and NVIDIA are at the back of the bus. I'd be surprised if we see any new 20nm parts until the very end of the year and nothing super exciting (IE big die 780, 290X style) until well into 2015.