DX12 on fermi

Seriously, why should i upgrade? Because you say so? I put my money where ever i want to. I make my judgement about the firm act and what is the best option for me. I love Fermi , i love the 3rd party tools but hate what is nvidia doing in the past 6 months. A company not gonna force me to do something what is against my will. Most users are thinking the same way, even they love the product/s , it is principle .

 

Last time i checked Techpowerup isn't geforce.com or a subsidiary of nvidia

http://www.geforce.com/hardware/notebook-gpus/geforce-920m/specifications

though they seems to be more on top of things.

Ending driver support maybe worse , but at least its not a lie.

 

AMD never publicly claimed DirectX 12 support for Terascale GPUs...

 

Last time I checked, you posted the link to Techpowerup to illustrate some point. Whoops.

 

last time i checked i linked the GTX 580 in the highlighted geforce.com text

and last time i checked i was pointing out that the gf117 is still being produced where it says production active in that sentence

anyhow no point trying to derail thread further.

 

That's exactly what I'll do, nvidia won't get any cent from me from now on. This 970 of mine will be the last green card for me.

Hope polaris & vega really deliver this time.

 

The idiocy in this is that you don't even demand support for things that were promised. At least AMD said from the beginning that DX12 was GCN only. NVIDIA released a bad driver for Fermi and you actually approve of that. That's pathetic.

 

No, the idiocy is in a Radeon owner visiting a GeForce drivers section to teach us poor GeForce owners how to use our products while the company which made his card straight up stopped supporting anything older than 7800/7900 cards which were launched two years after the release of Fermi GPUs.

And again, I very much prefer NV to spend the resources on the cards which actually matter. No, Fermi cards do not matter at this point of time in pretty much the same way as GTX8800 doesn't matter and Riva TNT doesn't matter. So I'm perfectly fine with no DX12 support for obsolete Fermi GPUs.

 

Fermi GPUs have the same feature set as Kepler, they are not obsolete, in fact, a 580 has the same compute power 670, 760. In my opinion, the things is, Nvidia said that DX12 would have support on Fermi, and if they don't give us support as promised, it will probably have consequences. There is not such thing as "spend the resources" for developing the driver as you think, we paid for this, it's not a courtesy, look at Microsoft, they have to support Windows 7 users, some time ago even Windows XP users.

 

Nvidia has claimed that Fermi users are less than 10% of their install base, as such there will be no Vulkan support - will this likely translate across to the DX12 API? Who knows, but it has been a long time since we have heard anything about a driver release target in terms of either driver branch or cut-off date...

Fermi consists of approximately 135 products, the Quadro users should be the ones shouting loudest as they paid a premium for a product that is about to lose future support. Next in line are Fermi SLI users (such as myself) who are still using GTX580/570/560/480/470/465/460 without issue.

Below is a list of Fermi products that are about to become obsolete;

Code:

List of Nvidia Fermi products that currently do not have DX12 or Vulkan support

135 products currently not supported!

Note: Tesla M2070 (Device ID 06D2), Tesla T20 (Device ID 06DE), Tesla M2070-Q (Device ID 06DF), Tesla M2090 (Device ID 1091), Tesla M2075 (Device ID 1094) are not included in this list due to not having any display-out ports. 
Also GeForce 920M/910M has been left off as I cannot find the hardware ID for the Fermi SKU.

GF100 (3)

GeForce GTX480 (Device ID 06C0)
GeForce GTX470 (Device ID 06CD)
GeForce GTX465 (Device ID 06C4)

GF100M (1)

GeForce GTX 480M (Device ID 06CA)

GF100GL (5)

Tesla C2050 / C2070 (Device ID 06D1)
Quadro Plex 7000 (2 x Quadro 6000) (Device ID 06DC)
Quadro 6000 (Device ID 06D8)
Quadro 5000 (Device ID 06D9)
Quadro 4000 (Device ID 06DD)

GF100GLM (1)

Quadro 5000M (Device ID 06DA)

GF104 (3)

GeForce GTX 460 (and X2) (Device ID 0E22)
GeForce GTX 460 OEM (Device ID 0E24)
GeForce GTX 460 SE (Device ID 0E23)

GF104M (2)

GeForce GTX 485M (Device ID 0E31)
GeForce GTX 470M (Device ID 0E30)

GF104GLM (2)

Quadro 4000M (Device ID 0E3B)
Quadro 3000M (Device ID 0E3A)

GF106 (3)

GeForce GTS 450 (Device ID 0DC4, 0DC6)
GeForce GTS 450 OEM (Device ID 0DC5)
GeForce GT 440 OEM (Device ID 0DC0)

GF106M (5)

GeForce GT 555M (Device ID 0DCD, 0DCE)
GeForce GT 550M (Device ID 0DD6)
GeForce GTX 460M (Device ID 0DD1)
GeForce GT 445M (Device ID 0DD2)
GeForce GT 435M (Device ID 0DD3)

GF106GL (2)

Quadro 2000D (Device ID 0DD8) 
Quadro 2000 (Device ID 0DD8)

GF106GLM (1)

Quadro 2000M (Device ID 0DDA)

GF108 (11)

GeForce GT 730 DDR3 (Device ID 1287)
GeForce GT 630 (Device ID 0F00)
GeForce GT 620 (Device ID 0F01)
GeForce GT 610 OEM (Device ID 0DE7)
GeForce GT 530 OEM (Device ID 0DE5)
GeForce GT 520 OEM (Device ID 0DE4)
GeForce GT 440 (Device ID 0DE0)
GeForce GT 430 (Device ID 0DE1)
GeForce GT 430 OEM (Device ID 0DE1)
GeForce GT 420 (Device ID 0DE2)
GeForce GT 420 OEM (Device ID 0DE2)

GF108M (16)

GeForce GT 635M (Device ID 0DE3)
GeForce GT 630M (Device ID 0DE9)
GeForce GT 620M (Device ID 0DE8)
GeForce 615M (Device ID 0DEA)
GeForce 610M (Device ID 0DEA)
GeForce GT 555M (Device ID 0DEB)
GeForce GT 550M (Device ID 0DF6)
GeForce GT 540M (Device ID 0DF4)
GeForce GT 525M (Device ID 0DEC)
GeForce GT 520M (Device ID 0DED)
GeForce GT 435M (Device ID 0DF2)
GeForce GT 425M (Device ID 0DF0)
GeForce GT 420M (Device ID 0DF3)
GeForce GT 415M (Device ID 0DEE)
NVS 5400M (Device ID 0DEF)
NVS 5200M (Device ID 0DFC)

GF108GL (1)

Quadro 600 (Device ID 0DF8) 

GF108GLM (2)

Quadro 1000M (Device ID 0DFA)
Quadro 500M (Device ID 0DF9) 

GF110 (7)

GeForce GTX 590 (Device ID 1088)
GeForce GTX 580 (Device ID 1080, 1089, 108B)
GeForce GTX 570 Rev. 2 (Device ID 1086) 
GeForce GTX 570 (Device ID 1081)
GeForce GTX 560 Ti 448 Cores (Device ID 1087)
GeForce GTX 560 Ti OEM 256 bit (Device ID 1082)
GeForce GTX 560 OEM 320 bit (Device ID 1084)

GF110GL (3)

Quadro 7000 (Device ID 109B)
Tesla C2090 (Device ID 108E)
Tesla C2050 / C2075 (Device ID 1096)

GF110GLM (1)

Quadro 5010M (Device ID 109A)

GF114 (9)

GeForce GT 645 OEM (Device ID 1207)
GeForce GTX 560 Ti (and X2) (Device ID 1200)
GeForce GTX 560 Ti OEM (Device ID 1202)
GeForce GTX 560 (Device ID 1201)
GeForce GTX 560 SE (Device ID 1208)
GeForce GTX 555 (Device ID 1206)
GeForce GTX 555 OEM (Device ID
GeForce GTX 460 v2 (Device ID 1205)
GeForce GTX 460 SE v2 (Device ID 1203) 

GF114M (4)

GeForce GTX 675M (Device ID 1212)
GeForce GTX 670M (Device ID 1213)
GeForce GTX 580M (Device ID 1211)
GeForce GTX 570M (Device ID 1210)

GF116 (7)

GeForce GT 640 OEM Rebrand (Device ID 124B)
GeForce GTX 550 Ti (Device ID 1244)
GeForce GT 545 GDDR5 OEM (Device ID 1241)
GeForce GT 545 DDR3 (Device ID 1243)
GeForce GT 530 (Device ID 0DE5)
GeForce GTS 450 DDR3 Rev 3 144 Core (Device ID 1249)
GeForce GTS 450 GDDR5 Rev 2 192 Core (Device ID 1245) 

GF116M (3)

GeForce GT 635M (Device ID 124D)
GeForce GTX 560M (Device ID 1251)
GeForce GT 555M (Device ID 124D)

GF117M (19)

GeForce 820A (28nm) (Device ID 1140
GeForce 820M (28nm) (Device ID 1140)
GeForce GT 820M (28nm) (Device ID 1140)
GeForce 810M (28nm) (Device ID 1140)
GeForce 800M (28nm) (Device ID 1140)
GeForce 720A (28nm) (Device ID 1140)
GeForce 720M (28nm) (Device ID 1140)
GeForce GT 720A (28nm) (Device ID 1140)
GeForce GT 720M (28nm) (Device ID 1140)
GeForce 710A (28nm) (Device ID 1140)
GeForce 710M (28nm) (Device ID 1140)
GeForce GT 710M (28nm) (Device ID 1140)
GeForce 705M (28nm) (Device ID 1140)
GeForce GT 630M (28nm) (Device ID 1140)
GeForce 620M (28nm) (Device ID 1140)
GeForce GT 625M (28nm) (Device ID 1140)
GeForce 620M (28nm) (Device ID 1140)
GeForce GT 620M (28nm) (Device ID 1140)
GeForce 610M (40nm) (Device ID 1140)

GF119 (17)

GeForce GT 710 (Device ID 104D)
GeForce GT 705 OEM (Device ID 104C)
GeForce GT 625 OEM (Device ID 104B)
GeForce GT 620 OEM (Device ID 1049)
GeForce GT 610 (Device ID 104A)
GeForce GT 610 OEM (Device ID 104A)
GeForce GT 610 PCIe x1 (Device ID 104A)
GeForce GT 610 PCI (Device ID 104A) 
GeForce 605 OEM (Device ID 1048)
GeForce GT 520 (Device ID 1040)
GeForce GT 520 OEM (Device ID 1040)
GeForce GT 520 PCIe x1 (Device ID 1040) 
GeForce GT 520 PCI (Device ID 1040)  
GeForce 510 (Device ID 1042)
GeForce 510 OEM (Device ID 1042)
NVS 315 (Device ID 107C)
NVS 310 (Device ID 107D)

GF119M (7)

GeForce 705M (Device ID 105B)
GeForce 610M (Device ID 1058, 1059, 105A)
GeForce GT 520MX (Device ID 1051)
GeForce GT 520M (Device ID 1050, 1052)
GeForce 410M (Device ID 1054, 1055)
Quadro NVS 4200M (Device ID 1057)
NVS 4200M (Device ID 1056)

 

To run it on the new driver you need to install the Insider preview build as mentioned at the start of the post.

 

apart from the well known hostility & fighting attitude of dr_rus towards anyone who has not "GTX > 980" written in their infobox there is another thing which is amusing about this decision to leave fermi on DX11 as well:

notice the irony in denying those cards dx12/vulkan...
because it's exactly those cards that would profit the most from it to achieve smooth framerates in the recent triple A titles and make them pretty playable.

probably NVIDIA announced it for fermi while underestimating (or not knowing) its performance gains so they feared sales would drop => they silently backed out.

 

also everyone keeps talking about of EOL for Fermi but i haven't found anything official or even a rumor other then it being posted here by Dr_RUS

 

AMD dropped 6000 series for the same reason. They got DX12 support then realized those cards would perform their tflop equivalents, putting them on par with GCN, so they killed support for the entire series.

Or you know, we can just not fabricate stories for the sake of furthering our narrative.

 

To me, there are two features that I want. I don't care about new lightning models, gameworks features, whatever, I just want DX12 for less CPU Overhead. Testing with API Overhead, I can't pass 1,200,000 score on DX11, while on DX12 I can achieve 3,400,000 easily, with just one card, and I have 2, and it is the other feature that I care about, other multi-gpu rendering better than AFR, their are calling this new DX12 feature "Explicit Multi-Adapter" with two modes, Linked and Unlinked GPUs.

And even if the SLI thing can't be possible, I care the most about the overhead, and it I think that is already working, otherwise the beta driver won't be able to work. Anyway, I'm hoping something on the next weeks, with the release of the Redstone update for mainstream W10 builds, maybe Nvidia is waiting for this update, I don't know...

 

1.) you say "AMD dropped 6000 for the same reason" & you say it's fabricated?
how does that fit together?

2.) AMD never dropped 6000, they just never announced that it would support DX12 API. so there was never anything to drop.

these are the official AMD cards to support DX12 (status W10 insider Build 10041... thats now almost 2 years ago, source: http://www.amd.com/en-us/innovations/software-technologies/directx12 )

Code:

Once you’ve installed Windows® 10 Technical Preview Build 10041 (or later) and obtained the latest driver from Windows Update, these AMD components are DirectX® 12-ready:
AMD Radeon™ R9 Series graphics
AMD Radeon™ R7 Series graphics
AMD Radeon™ R5 240 graphics
AMD Radeon™ HD 8000 Series graphics for OEM systems (HD 8570 and up)
AMD Radeon™ HD 8000M Series graphics for notebooks
AMD Radeon™ HD 7000 Series graphics (HD 7730 and up)
AMD Radeon™ HD 7000M Series graphics for notebooks (HD 7730M and up)
AMD A4/A6/A8/A10-7000 Series APUs (codenamed “Kaveri”)
AMD A6/A8/A10 PRO-7000 Series APUs (codenamed “Kaveri”)
AMD E1/A4/A10 Micro-6000 Series APUs (codenamed “Mullins”)
AMD E1/E2/A4/A6/A8-6000 Series APUs (codenamed “Beema”)

Nvidia in contrary announced DX12 API support for Fermi, and now you hear nothing of it anymore.
I don't plan on buying a Fermi card so it's not really relevant for me. But for others who own one it may be essential. As we can read in these forums...

 

http://docs.nvidia.com/nsight-visual-studio-edition/5.1/Content/Release_Notes_Main.htm

"Starting with the next release of NVIDIA Nsight, Fermi GPUs will no longer be supported. "

This is why fermi is likely EOL , fermi is being dropped from nsight which is used for software development and debugging.

No software development support = effective death

 

k that would make sense as no one in the right mind would develop software using older gpus. the Fermi Quadro GPUs mostly any Quadro GPU should EOL faster than the consumer product mostly because you would want Top of the line Quadro cards in your setup for programing new games i guess.

 

There is 1.58 Tflops compute power on the gtx580. Its more powerful than Xbox one. Its a shame to throw into the garbage becouse its has "missing feature set" for next api.

They promissed and didnt deliver. I was very surpriced when i heard that there is a older driver that can run dx12 api overhead test while latest one cant. What a nonsense.

 

a solution is nvidia open source code driver nvidia to all and users create driver dx12 for fermi

why nvidia not open source code of your drivers?