Comments Locked

49 Comments

Back to Article

  • DanNeely - Wednesday, September 18, 2013 - link

    Since the new GPU is coming out in Q4 of this year it still being 28nm isn't that surprising.

    I'm wondering what it says about their plans though. Earlier in the year when the 8xxx oem rebadging was announced it was said we wouldn't be getting a real new GPU until mid next year when TSMC 20nm was expected to become available.

    That has me wondering if this launch is an indication that TSMC's 20nm process is having problems and will be delayed and this is is what they've chosen to launch instead of waiting. Alternately, was the size of the performance bump nVidia got with GF110 and the 7xx refresh significantly larger than expected causing them to release this as a stopgap instead of yielding the performance crown for another half year.
  • Jumangi - Wednesday, September 18, 2013 - link

    Oh I think is plainly obvious TSMC is having more of the same problems getting 20nm going just like it took them forever to get 28nm up to respectable levels. As these processes get lower it looks like it will be longer between jumps.
  • Wreckage - Wednesday, September 18, 2013 - link

    They may have spent most of their resources developing for the consoles. This may be a hint at their exit from the high end similar to their CPUs. I doubt they will have a GK110 killer on their hands with 28nm.
  • Goty - Wednesday, September 18, 2013 - link

    Just like GK110 wasn't a big step over GK104 just because it's on the same manufacturing process.

    Oh, wait.
  • Hector2 - Wednesday, September 18, 2013 - link

    TSMC has never met their predicted process schedules and new production-worthy processes always come out at least a year behind schedule. It's just a marketing attempt to get people to wait for something that supposedly "just around the corner". It's always much easier to change a number on a Powerpoint slide than to actually change the process --- especially now that we're below 22nm
  • Frenetic Pony - Wednesday, September 18, 2013 - link

    Well, TSMC 20nm already doesn't offer much of anything in improvement in the way of power/performance and isn't any cheaper. The only improvement would then be more silicon per mm squared. We already know Apple signed a big contract for 20nm, so my guess is AMD decided they'd rather go 28nm and get it out in time for Christmas than wait 4+ months for little gain.
  • Da W - Wednesday, September 18, 2013 - link

    A wild guess here, we can expect GCN 1.1 with may be 5-10% performance improvement. Some sort of architecture optimisation kinda like we saw from fermi to keppler.
    With a 23% smaller die than Titan, maturing process and probably a 1.1Ghz clock, this would put Hawaii at par or slightly above GTX780. My guess is that AMD went for higher power consumption improvement rather than pure performance.

    Anyhow, this will be the best card for an eyfinity setup, without microsuttering issues.
  • purerice - Wednesday, September 18, 2013 - link

    Agreed, though you may have dropped a t in microstuttering. I hopefully won't have to upgrade for a year or two but it's nice to see this. I am not a fan of either AMD or nVidia so I want both to keep pushing the other, even if in incremental improvements.
  • Frenetic Pony - Wednesday, September 18, 2013 - link

    It's been 2 years, I've heard they're calling this an actual new architecture (GCN 2.0) rather than a refresh. Which means their plans would have changed a lot since earlier this year, which is what this article seems to imply anyway.
  • Sabresiberian - Thursday, September 19, 2013 - link

    Above GK110? I hardly think so, unless there is a lot more improvement in the "maturing process".

    Don't get me wrong, I'd love for it to be true - we need the competition to stay alive in the GPU space just as it was good for us in the CPU space - but I have serious doubts. AMD's current single-GPU solution is far below the GTX 780, and it is going to take more than a 10% bump to make them competitive at the top end.
  • nathanddrews - Wednesday, September 18, 2013 - link

    Any indication that they are moving toward some sort of modular design with each tier getting more GCN "cores"?

    Perhaps all this investigation into stuttering and such has led to a better method of on-PCB Crossfire, perhaps a method that makes the process transparent to the application? Combined with HSA, that would be sweet...
  • MrSpadge - Wednesday, September 18, 2013 - link

    GPUs have already been highly modular since a long time. If you need more performance you essentially simply scale the number of building blocks.
  • nathanddrews - Wednesday, September 18, 2013 - link

    For lack of better terms to describe it, I'm referring to building up to a larger chip rather than scaling down from a larger chip.
  • Cow86 - Wednesday, September 18, 2013 - link

    For this to be really as exciting and efficient as AMD is touting, this has to beat GK104 in performance/die size...and also performance/watt. I'm wondering however if they just mean they come close to Titan with 23% less die size...

    There are still far too many unknowns though, like how much compute Hawaii would have (GK110 is a compute monster afterall, which is also a big reason for the large die size). Here's me hoping AMD wins this round, even if it isn't by much...Can't imagine it being a huge win if they're staying on 28 nm afterall, but I'll take a marginal one.
  • The Von Matrices - Friday, September 20, 2013 - link

    Remember, some of Titan and GTX 780 is disabled from the full GK110 die, so just comparing die sizes isn't the full story. You need to consider the amount of die space that is actually enabled on those chips, and then you'll find that AMD is comparable.
  • HighTech4US - Wednesday, September 18, 2013 - link

    Why is AMD only comparing die size and not performance?
    Who cares about die size it performance that counts.
  • Ortanon - Wednesday, September 18, 2013 - link

    I doubt they'd be comparing the die size if performance wasn't comparable.
  • JarredWalton - Wednesday, September 18, 2013 - link

    I think that's a stretch. AMD and NVIDIA always have different strengths in terms of gaming, but I don't think AMD is really planning on eclipsing Titan. More likely is that they're aiming at the GTX 780, which of course has 384 fewer cores (two disabled SMXes) than Titan and at stock speeds ends up being about 10% slower. The law of diminishing returns is of course in full effect, so getting that last 10% (e.g. with Titan) costs a lot, and I don't think AMD is planning on going after the $1000 GPU market. Hopefully AMD comes in at around $500 again and brings NVIDIA's GTX Titan/780 prices down to a more reasonable range.
  • superjim - Wednesday, September 18, 2013 - link

    A second coming of the 5870 (or even 9700 Pro) would be nice. $380 for current-$500 performance.
  • imaheadcase - Wednesday, September 18, 2013 - link

    Do you guys have one in house yet? Or do you normally get them just a few days before release?
  • dylan522p - Wednesday, September 18, 2013 - link

    NDA. They can't say shit.
  • imaheadcase - Thursday, September 19, 2013 - link

    Yes they can, they have said before if they have one. NDA is for performance related stuff.
  • JlHADJOE - Thursday, September 19, 2013 - link

    An NDA can be for anything. It really depends only on the two parties involved.
  • surt - Sunday, September 22, 2013 - link

    However, after the first time they say that, the next NDA says you can't disclose the existence of the NDA.
  • MrSpadge - Wednesday, September 18, 2013 - link

    Because it's neither a launch nor an official introduction yet, just a small tease in some interview.

    And die size does count, because it determines how low the price of a chip can be. If AMD offers similar performance with a smaller chip than nVidia expect some agressive pricing. The other way around - not.
  • haukionkannel - Wednesday, September 18, 2013 - link

    Yep. we don't actually need another 1000$ GPU to compete with Titan. So if this is about same size as 7970, then the price will be guite the same when 7970 did come out 500-600$ Dependin on how good it will be compared to Nvidia 780.
  • eanazag - Wednesday, September 18, 2013 - link

    Die size and process technology are important to Forbes readers (investors). If it was to be a GPU performance talk, why go to Forbes?
    That being said; smaller die likely means less power and heat which is a good thing to me for a GPU. AMD just needs to break even on performance or even be negative in 3% or less range with the 780 and it's a win. It could end up being a play like Nvidia did previously by dropping the compute strength down.

    No matter what. It is time to release something even if it is just a little better than 7970. I can't stand rebranding. It irks me.
  • Ammaross - Wednesday, September 18, 2013 - link

    @HighTech4US: "Who cares about die size it performance that counts."

    Because die size is directly related to chip costs, and thus, retail pricing.
  • Rick83 - Thursday, September 19, 2013 - link

    Well, not really. Pricing is mostly determined by how many units you can sell, with the hope of recouping R&D and making good profits.

    The influence of die size is that you can have more chips ready for a launch in less time, as the time to process a wafer is roughly constant, and especially on a fresh process, smaller dies means more dies that aren't broken (of course by disabling broken bits on a GPU you can relatively cheaply reduce the impact on the total number of functioning dies).

    But yeah, at this point die size isn't such a critical metric with regards to cost. On the other hand, it is an interesting metric with regard to total current through the chip, which can be larger, the larger your chip is. Current correlates with power requirements and output, so it's an interesting measure in that aspect.

    Just look at GK104 for an example of small die does not equal low retail pricing.
  • B3an - Wednesday, September 18, 2013 - link

    Maybe my next upgrade... I've often got the latest GPU's but lately i've stuck with two 6970's because even at 2560x1600 they run almost everything maxed out still, because almost everything is a console port these days.

    This isn't a good thing to me though. I want something to push my system and want amazing graphics, so if it means i have to upgrade my GPU's then so be it. Hopefully the new consoles will push PC graphics and texture resolutions up significantly. I hope these new AMD cards have 4GB RAM.
  • Da W - Wednesday, September 18, 2013 - link

    New consoles are 7870 caliber. I doubt they will push PC gaming. PC gaming will push PC gaming.
  • DanNeely - Wednesday, September 18, 2013 - link

    Programming closer to the bare silicon always gives consoles an advantage over an equivalent PC GPU; so they still might be competitive with the best PC graphics next year. Regardless the big gains in PC gaming graphics will probably come in a few years when the XB360 and PS3 are deprecated and developers are able to put all the resources they previously needed to scale down to a 2005 era GPU into maxing out 2015 or 2016 models.
  • jasonelmore - Wednesday, September 18, 2013 - link

    wtf? not even close. it's mobile SOC gpu vs full fledge dedicated, not even close
  • dylan522p - Wednesday, September 18, 2013 - link

    How is it a mobile SOC? Just because they use 8 mobile cores doesn't make it a mobile SOC.
  • nunomoreira10 - Wednesday, September 18, 2013 - link

    new consoles will definitely push pc gaming, there are several games that run 1080p and 30fps on the new consoles, push that to 120fps or 1440p and 60fps and you will need 4 times the power, rise some effects like dynamic lights, shadows, aliasing, draw distance and it will kick the hell of the best pc you can make
  • superjim - Wednesday, September 18, 2013 - link

    The PS4 is actually closer to a 7850 and I agree they won't push PC gaming even it had a GTX 780 in it.
  • B3an - Wednesday, September 18, 2013 - link

    "PC gaming will push PC gaming" ... no it wont. Look at PC games over the past few years, pretty much all console ports with minor graphical tweaks, if any. Yet the PC version requires superior PC hardware to get the same frame rates as the console version. Theres no dedicated PC games that really push things forward. It's not like when the first Crysis came out on PC and blew everything else away because it was totally designed for PC hardware - those days are gone. We now have to rely on consoles to push things forward.

    With all the RAM available to games on the new consoles (around 5GB) i'm atleast expecting to see some very high res textures on both the consoles and the PC ports. Over the next year we will atleast see a big increase in texture res on PC games and the recommended VRAM will shoot up. Look at BF4 for instance, they recommend 3GB VRAM.
  • airmantharp - Wednesday, September 18, 2013 - link

    Hilarity, all around. The PC versions are the same games running on engines that have been reworked for years to make use of the extra resources available.

    But you are right about the textures/in-game assets. If a console game can use 4GB of RAM for graphics, expect to need 6GB minimum per GPU just to turn the details up- and that's before MSAA or anything else that might jump up the per-pixel memory requirements. I'm not getting a card with less than 8GB/GPU to replace my 2GB GTX670s. I'm shooting for 4k, full details enabled, in a year or two, probably on Haswell-E as well.
  • Peanutsrevenge - Wednesday, September 18, 2013 - link

    Look harder.
    Star Citizen will be PC only and designed for high end hardware (dx11 only for starters).
  • Peanutsrevenge - Wednesday, September 18, 2013 - link

    Something like Star Citizen you mean? :-)
  • Kevin G - Wednesday, September 18, 2013 - link

    Good timing. I just had a Radeon 7970 die due to shipping issues (thank you TSA/US Customs).
  • jljaynes - Wednesday, September 18, 2013 - link

    You have to remember that some of the most popular games currently are free to play/cheap that don't take top-notch hardware to run - think LoL or Dota 2
  • ezjohny - Wednesday, September 18, 2013 - link

    Hopefully this 28nm GPU could work with the latest AMD 7000 series graphic cards!
  • ezjohny - Wednesday, September 18, 2013 - link

    This is not the APU, I'm wrong.
  • Laststop311 - Thursday, September 19, 2013 - link

    Price is going to be the key thing here. It's pretty much a given its not going to beat the 1000 dollar titan But the Gtx780 is only a tiny step behind titan in performance and it's only 650-700 dollars. It's pretty safe to say this will not beat gtx 780 in performance either. I think AMD is going for a GTX 770 killer. At 400-450 for the GTX 770 if AMD can get a nice 10-15% boost over GTX 770 and price it at the same we will have a real winner on our hands. I don't think we are gonna see the GTX 780 topped tho even though that would be nice. I would love to see the prices fall.

    I'm on the cusp of purchasing a GPU for my silent PC build and was gonna pull the trigger on an MSI gamer N770 or MSI Gamer N780 for it's low noise larger fans. If AMD can price these right I may switch to team Red as long as they can get the power consumption under control the 7970 was a freakin hot as hell running hog that was loud.
  • gostan - Thursday, September 19, 2013 - link

    still cant play GTA V :(
  • Da W - Thursday, September 19, 2013 - link

    If you look at bench there's not that much difference between 7870 Ghz edition and GTX 780, except in a few specific games like total war of Civ. Plus power consumption is alike, temp and noise is higher on Nvidia.
    Add to that that GCN is first gen architecture and Keppler is 2nd Gen architecture (fermi improvment). I think with GCN 2.0, smaller but undisabled die, maturing process and cranking up the clock beyong 1Ghz, AMD can beat GTX 780.
  • evolucion8 - Tuesday, September 24, 2013 - link

    I am certainly sure that the new Radeon will beat Titan. Why? As like Da W stated, the HD 7970GHz offers around 10-15% of the GTX 780 in average scenarios while there are scenarios where it can even match or outperform Titan in games like GRID 2, Dirt Showdown, Splinter Cell Conviction etc. So a GPU that will bring over 37% more shaders, with architecture improvements, will be able to be at least 30% or more faster than the HD 7970GHz, putting it on Titan territory or more, specially on general purpose computing aka GPGPU where nVidia still lacking even today with Titan, plus more games are coming with a general purpose computing approach to calculate stuff like hair, which is more suitable with the computing oriented AMD's GCN.
  • shinkueagle - Wednesday, September 25, 2013 - link

    LOL! FANBOY SPEAK MUCH....

    YOU ARE SO WRONG....

    - http://www.techpowerup.com/191453/amd-gpu14-event-...

Log in

Don't have an account? Sign up now