Jump to content

AMD might reveal the first 20nm GPUs this year, or they might not :<

KakaoDj

WCCF article

Kitguru article has a different conclusion

 

Updated the post with additional info, check bottom part of OP.

 

 

I don't hink we will be seeing R9 23xx this year  :(

They will use a FinFET process, which TSMC just released some info on here: TSMC reveals new 16nm FinFET+ process, vows to start 10nm production in Q4 2015

 

 

Lisa Su – SVP and General Manager of Global Business Units

 

Sure, Chris. So let me take that and give you a little bit of our thinking.

So in terms of product and technology selection, certainly we need to be at the leading-edge of the technology roadmap. So what we’ve said in the past is certainly this year all of our products are in 28-nanometer across both, you know, graphics client and our semi-custom business. We are, you know, actively in the design phase for 20-nanometer and that will come to production. And then clearly we’ll go to FinFET. So that would be the progression of it.

 

Lisa Su – SVP and General Manager of Global Business Units

 

UPDATE

 

 

 

KitGuru interprets the quote differently.

it is logical to assume that the first AMD products to be made using 20nm fabrication process are graphics processing units (GPUs).

While the senior vice president of AMD did not officially confirm that the first 20nm AMD Radeon GPUs are set to emerge in 2014, her comment regarding coming “to production” may hint on the plan to reveal such GPUs already this year.

AMD-20nm.jpg

Link to comment
Share on other sites

Link to post
Share on other sites

Well then f u amd i'm definitely buying 760 now  -_-

Location: Kaunas, Lithuania, Europe, Earth, Solar System, Local Interstellar Cloud, Local Bubble, Gould Belt, Orion Arm, Milky Way, Milky Way subgroup, Local Group, Virgo Supercluster, Laniakea, Pisces–Cetus Supercluster Complex, Observable universe, Universe.

Spoiler

12700, B660M Mortar DDR4, 32GB 3200C16 Viper Steel, 2TB SN570, EVGA Supernova G6 850W, be quiet! 500FX, EVGA 3070Ti FTW3 Ultra.

 

Link to comment
Share on other sites

Link to post
Share on other sites

So re-branding here we come

Unless they be like Nvidia and do anything to change their architecture to be more efficient/faster/etc.

.

Link to comment
Share on other sites

Link to post
Share on other sites

Introducing the AMD time machine!

"No new gpus coming out this year!"

...releases r9 290/290x

Link to comment
Share on other sites

Link to post
Share on other sites

No clue what smaller dies mean but this is going to be interesting!

More performance, less energy consumption and higher prices basically

Link to comment
Share on other sites

Link to post
Share on other sites

Was expected. Nvidia too has been unable to get 20nm out the door.

 

Yes, but there is an expected window of Q3 for 20nm maxwell production and a possible Q4 release. Hearing that AMD will have nothing until 2015 is a bit disappointing. I prefer Nvidia, but I desperately want AMD to kick them around and force down prices. Those chunks of silicon cant possibly cost more than $300 to produce and charging $500 for them is acceptable, hence the GTX 780 & R9 290. As a hardware enthusiast, the $700 GTX 780 ti & R9 290X makes me want to cry.

Quote

Ignis (Primary rig)
CPU
 i7-4770K                               Displays Dell U2312HM + 2x Asus VH236H
MB ASRock Z87M Extreme4      Keyboard Rosewill K85 RGB BR
RAM G.Skill Ripjaws X 16GB      Mouse Razer DeathAdder
GPU XFX RX 5700XT                    Headset V-Moda Crossfade LP2
PSU Lepa G1600
Case Corsair 350D
Cooling Corsair H90             
Storage PNY CS900 120GB (OS) + WD Blue 1TB

Quote

Server 01Alpha                                       Server 01Beta                            Chaos Box (Loaner Rig)                Router (pfSense)
CPU
 Xeon X5650                                      CPU 2x Xeon E5520                    CPU Xeon E3-1240V2                     CPU Xeon E3-1246V3
MB Asus P6T WS Pro                               MB EVGA SR-2                             MB ASRock H61MV-ITX                 MB ASRock H81 Pro BTC
RAM Kingston unbuffered ECC 24GB  RAM G.Skill Ripjaws 16GB         RAM Random Ebay RAM 12GB    RAM G.Skill Ripjaws 8GB
GPU XFX R5 220                                       GPU EVGA GTX 580 SC               GPU Gigabyte R9 295x2                GPU integrated
PSU Corsair CX430M                               PSU Corsair AX1200                   PSU Corsair GS700                         PSU Antec EA-380D
Case Norco RPC-450B 4U                      Case Rosewill  RSV-L4000C        Case Modified Bitfenix Prodigy   Case Norco RPC-250 2U
Cooling Noctua NH-U9S                        Cooling 2x CM Hyper 212 Evo  Cooling EVGA CLC 120mm           Cooling stock
Storage PNY CS900 120GB (OS)           Storage null                                 Storage PNY CS900 120GB (OS)  Storage Fujitsu 150GB HDD
               8x WD Red 1TB in Raid 6                                                                                WD Black 1TB    
               WD Green 2TB

 

Link to comment
Share on other sites

Link to post
Share on other sites

Those chunks of silicon cant possibly cost more than $300 to produce and charging $500 for them is acceptable, hence the GTX 780 & R9 290. As a hardware enthusiast, the $700 GTX 780 ti & R9 290X makes me want to cry.

 

Yeah, especially as you could buy a GTX 590 for $600.

Intel i7 5820K (4.5 GHz) | MSI X99A MPower | 32 GB Kingston HyperX Fury 2666MHz | Asus RoG STRIX GTX 1080ti OC | Samsung 951 m.2 nVME 512GB | Crucial MX200 1000GB | Western Digital Caviar Black 2000GB | Noctua NH-D15 | Fractal Define R5 | Seasonic 860 Platinum | Logitech G910 | Sennheiser 599 | Blue Yeti | Logitech G502

 

Nikon D500 | Nikon 300mm f/4 PF  | Nikon 200-500 f/5.6 | Nikon 50mm f/1.8 | Tamron 70-210 f/4 VCII | Sigma 10-20 f/3.5 | Nikon 17-55 f/2.8 | Tamron 90mm F2.8 SP Di VC USD Macro | Neewer 750II

Link to comment
Share on other sites

Link to post
Share on other sites

Yes, but there is an expected window of Q3 for 20nm maxwell production and a possible Q4 release. Hearing that AMD will have nothing until 2015 is a bit disappointing. I prefer Nvidia, but I desperately want AMD to kick them around and force down prices. Those chunks of silicon cant possibly cost more than $300 to produce and charging $500 for them is acceptable, hence the GTX 780 & R9 290. As a hardware enthusiast, the $700 GTX 780 ti & R9 290X makes me want to cry.

It's not just AMD, I have my doubts whether nvidia can get 20nm Maxwell out the door within this year. If they do I expect it will be the end of the year, but more likely 2015 for them too.

Link to comment
Share on other sites

Link to post
Share on other sites

Yes, but there is an expected window of Q3 for 20nm maxwell production and a possible Q4 release. Hearing that AMD will have nothing until 2015 is a bit disappointing. I prefer Nvidia, but I desperately want AMD to kick them around and force down prices. Those chunks of silicon cant possibly cost more than $300 to produce and charging $500 for them is acceptable, hence the GTX 780 & R9 290. As a hardware enthusiast, the $700 GTX 780 ti & R9 290X makes me want to cry.

actually they yield much more money than that for gpu's. amd is just terrible at marketing and managing volume.

Link to comment
Share on other sites

Link to post
Share on other sites

No clue what smaller dies mean but this is going to be interesting!

Lower power consumption and also a small performance boost due to the electrical circuits being shorter. We are only a couple of short years away from seeing HD 7850 level performance on a card that requires no external power. The further they get with that, the more performance they can also cram into low powered units (laptops, cell phones, etc).

Link to comment
Share on other sites

Link to post
Share on other sites

Wccf D: Though isn't the article a general idea anyways? 20nm Maxwell is going to be out just before 2015, or possibly in 2015 or something like that anyways?

Link to comment
Share on other sites

Link to post
Share on other sites

same node =/= rebrand

 

Yes but look how physically big GK110 is. Without shrinking the transistors down from 28nm, you can't fit anymore on the die, thus we aren't going see any massive performance increases, and will likely see tons of rebranding if another 28nm GPU range is launched. 

Link to comment
Share on other sites

Link to post
Share on other sites

Not AMD but Globalfoundries dictates the node size of AMD products...  They are getting close to silicon minimal tranzistor size so they have to keep their business up and running until graphene becomes available.

FX8320 4.2Ghz@1.280v& 4.5 Ghz Turbo@1.312v Thermalright HR-02/w TY-147 140MM+Arctic Cooling 120MMVRM cooled by AMD Stock Cooler Fan 70MM 0-7200 RPM PWM controlled via SpeedfanGigabyte GA990XA-UD3Gigabyte HD 7970 SOC@R9 280X120GiBee Kingston HyperX 3K2TB Toshiba DT01ACA2001TB WD GreenZalman Z11+Enermax 140MM TB Apollish RED+2X Deepcool 120MM and stock fans running @5VSingle Channel Patriot 8GB (1333MHZ)+Dual Channel 4GB&2GB Kingston NANO Gaming(1600MHZ CL9)=14GB 1,600 Jigahurtz 10-10-9-29 CR1@1.28VSirtec High Power 500WASUS Xonar DG, Logitech F510Sony MDR-XD200Edifier X220 + Edifier 3200A4Tech XL-747H 3600dpiA4Tech X7-200MPdecent membrane keyboardPhilips 236V3LSB 23" 1080p@71Hz .

               
Sorry for my English....

Link to comment
Share on other sites

Link to post
Share on other sites

Yes but look how physically big GK110 is. Without shrinking the transistors down from 28nm, you can't fit anymore on the die, thus we aren't going see any massive performance increases, and will likely see tons of rebranding if another 28nm GPU range is launched. 

shrinking isnt the only way to increase performance .Shrinking and making clocks higher is an old strategy that stopped during Pentium's  late days

 

also sticking to 28nm doesnt mean rebranding if they are using an entirely new architecture they will not use the GK110

rebranding means they are using the exact same architecture , the same node with a different name

 

architectural changes is where the big differences are

pretty much every company does intel's tick tock method or a variation of it 

https://en.wikipedia.org/wiki/Intel_Tick-Tock

 

an example would be intel's 45nm penryn to intel's 45nm Nehalem  Or pentium's netburst to Core all on 65nm

 

shrinking doesnt make many big improvements

an example would be 8xxx to 9xxx or 4xx to 5xx

but architectural  changes make big improvements like from 9xxx to 2xx or 5xx to 6xx

If your grave doesn't say "rest in peace" on it You are automatically drafted into the skeleton war.

Link to comment
Share on other sites

Link to post
Share on other sites

So re-branding here we come

Actually it opens up more flexibility with a new architecture by sticking with 28nm. While it won't be as 'jam packed' and could just be that, rebranding of similar products, they may end up pulling a rabbit out of their hats and surprising everyone with something very, very good. The flexibility comes with less pressure to shrink, already having 28nm architectures to modify may assist with something that's a step forward outside of size. Also, more resources to dedicate.

Edit: Then again, this could be all heresay. WCCFTech blows.

Link to comment
Share on other sites

Link to post
Share on other sites

Actually it opens up more flexibility with a new architecture by sticking with 28nm. While it won't be as 'jam packed' and could just be that, rebranding of similar products, they may end up pulling a rabbit out of their hats and surprising everyone with something very, very good. The flexibility comes with less pressure to shrink, already having 28nm architectures to modify may assist with something that's a step forward outside of size. Also, more resources to dedicate.

Edit: Then again, this could be all heresay. WCCFTech blows.

I do agree but I think it is time for AMD to change things up. But I do also understand that they don't have the money for R&D.

\

Link to comment
Share on other sites

Link to post
Share on other sites

Not AMD but Globalfoundries dictates the node size of AMD products...  They are getting close to silicon minimal tranzistor size so they have to keep their business up and running until graphene becomes available.

 

The next set of AMD chips are going to be made by TSMC, globalfoundries are likely to be making the low-end stuff and power saving chips.

 

http://www.xbitlabs.com/news/other/display/20131017231002_AMD_to_Tape_Out_First_20nm_14nm_FinFET_Chips_Within_Next_Two_Quarters.html

Intel i7 5820K (4.5 GHz) | MSI X99A MPower | 32 GB Kingston HyperX Fury 2666MHz | Asus RoG STRIX GTX 1080ti OC | Samsung 951 m.2 nVME 512GB | Crucial MX200 1000GB | Western Digital Caviar Black 2000GB | Noctua NH-D15 | Fractal Define R5 | Seasonic 860 Platinum | Logitech G910 | Sennheiser 599 | Blue Yeti | Logitech G502

 

Nikon D500 | Nikon 300mm f/4 PF  | Nikon 200-500 f/5.6 | Nikon 50mm f/1.8 | Tamron 70-210 f/4 VCII | Sigma 10-20 f/3.5 | Nikon 17-55 f/2.8 | Tamron 90mm F2.8 SP Di VC USD Macro | Neewer 750II

Link to comment
Share on other sites

Link to post
Share on other sites

shrinking isnt the only way to increase performance .Shrinking and making clocks higher is an old strategy that stopped during Pentium's  late days

 

also sticking to 28nm doesnt mean rebranding if they are using an entirely new architecture they will not use the GK110

rebranding means they are using the exact same architecture , the same node with a different name

 

architectural changes is where the big differences are

pretty much every company does intel's tick tock method or a variation of it 

https://en.wikipedia.org/wiki/Intel_Tick-Tock

 

an example would be intel's 45nm penryn to intel's 45nm Nehalem  Or pentium's netburst to Core all on 65nm

 

shrinking doesnt make many big improvements

an example would be 8xxx to 9xxx or 4xx to 5xx

but architectural  changes make big improvements like from 9xxx to 2xx or 5xx to 6xx

 

Sure, not rebranding in a technical and true sense like you described, I think the masses feel it's a rebrand when a slightly cheaper product is almost the same as a last gen one people see it as a rebrand (GTX770 not technically a rebrand but close vs R9 280X which is identical to an HD7970 GHz).

 

I would have said that the Kepler Refresh is the 'tick' tbh, look at the gains from GTX570 --> GTX670 --> GTX770. Of course shrinking doesn't automatically give more performance, but it gives more space for more cores etc. 

 

People will feel the 800 series is a crap shoot from NVidia if we get <20% performance gains. Tweaking/improving general things with the architecture isn't going to give anyone a performance increase like what you would see with 50% more cores.

Link to comment
Share on other sites

Link to post
Share on other sites

same node =/= rebrand

 

Not necessarily, but come on you know they'll rebrand most of the cards and only release maybe 1 or 2 new flagships

-------

Current Rig

-------

Link to comment
Share on other sites

Link to post
Share on other sites

Yes, but there is an expected window of Q3 for 20nm maxwell production and a possible Q4 release. Hearing that AMD will have nothing until 2015 is a bit disappointing. I prefer Nvidia, but I desperately want AMD to kick them around and force down prices

 

This: you guys need to realize how damaging this is for overal growth: Now Nvidia has no fucking reason whatsoever to release top of the line Maxwell cards. They might release gimped cards as we expected (256 bit memory) but they can chill out in that relaxed state without actually pushing maxwell for the entirety of 2015 it seems.

-------

Current Rig

-------

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×