
NVIDIA has officially launched its new high-end graphics cards, the GeForce GTX 980 and GTX 970, both powered by the second iteration of the Maxwell GPU. Many details (specs, benchmarks, slides, etc.) are already available in this article, so let’s recap the main and official specifications:
GeForce GTX 980 specs
:
- GPU: GM204-400 (Second Generation Maxwell architecture), 1126MHz base clock, 1216MHz boost clock, 28nm, 5.2-billion transistor
- CUDA Cores: 2048
- SMM (Streaming Multiprocessor Maxwell): 16 (one SMM = 128 CUDA cores)
- FP32: 4981 GFLOPS (see comp table HERE)
- FP64: 1/32 FP32 (see comp table HERE)
- ROPS: 64
- TMUs: 128
- Memory; 4GB GDDR5 256-bit @ 1750MHz real speed or 7000MHz effective speed
- TDP: 165W
- Power connectors: two 6-pin
- Price: USD $550
- 3D APIs: OpenGL 4.5, Direct3D 12
- GPU Computing: OpenCL 1.1, CUDA, PhysX
- Output connectors: 3 x DisplayPort 1.2, 1 x HDMI, 1 x DVI


source
The GTX 980 official hompage is here: GeForce GTX 980 @ NVIDIA.
Gallery






A small part of the backplate can be removed for better cooling when several GTX 980 are used in SLI:
Scores
TessMark score:

source

source
Video Reviews
Reviews
- techreport.com
- techpowerup.com
- anandtech.com
- hardware.fr
- hardwareheaven.com
- hexus.net
- diy.pconline.com.cn
- pugetsystems.com
Articles
GeForce GTX 970 specs
- GPU: GM204-200 (Second Generation Maxwell architecture), 1050MHz base clock, 1178MHz boost clock, 28nm, 5.2-billion transistor
- CUDA Cores: 1664
- SMM (Streaming Multiprocessor Maxwell): 13 (one SMM = 128 CUDA cores)
- FP32: 3920 GFLOPS (see comp table HERE)
- FP64: 1/32 FP32 (see comp table HERE)
- ROPS: 64
- TMUs: 104
- Memory; 4GB GDDR5 256-bit @ 1750MHz real speed or 7000MHz effective speed
- TDP: 145W
- Power connectors: two 6-pin
- Price: USD $330
- 3D APIs: OpenGL 4.5, Direct3D 12
- GPU Computing: OpenCL 1.1, CUDA, PhysX
- Output connectors: 3 x DisplayPort 1.2, 1 x HDMI, 1 x DVI
The ASUS Strix GTX 970 OC GPU-Z screenshot:

GeForce GTX 970 official homepage: GeForce GTX 970 @ NVIDIA
Hmm.. finally this sound interesting and reasonably priced.
256-bit mem – 64 ROP – 104 TMU. I don’t like it, it seems NV made another “top” GF680. I’m waiting for real top not that half-top.
“I’m waiting for real top not that half-top.”
I agree, but what would one expect with a lack of competition from AMD?
NVIDIA is going to milk its cow slowly. Ship the “full” architecture as Titan/Tesla first, then ship the consumer version. It worked well for them the first time around, didn’t it?
http://www.youtube.com/watch?v=sjiJ_B0yIMw
http://www.youtube.com/watch?v=O9y_AVYMEUs
VXGI is really cool.
http://blog.icare3d.org/2014/09/maxwell-gm204-opengl-extensions.html
Definitely best nVidia card since 8800/9800 or best chip since G8x/G9x series
http://www.youtube.com/watch?v=b57f0nR58RI
“GAME24 Keynote: NVIDIA CEO Jen-Hsun Huang Introduces Maxwell, GeForce GTX 980 and 970”
Does GeForce GTX 970 support vray adv and vray rt in 3DS max?
I’ve got vray 3 rt and Gygabite’s Gtx 980. It don’t work in CUDA mode, (i’ve tried with vray 3.00.03 and 3.00.07 in max 2014 and 2015).
OpenCL mode works perfect.
I don’t know if is something common or just me…
We will have to wait newer versions i guess!
http://www.geforce.com/whats-new/articles/geforce-gtx-900m-laptops-available-now
http://www.geforce.com/hardware/notebook-gpus/geforce-gtx-980m
http://www.geforce.com/hardware/notebook-gpus/geforce-gtx-970m
I have the same problem than Ender (with v-ray 3.00.03 and 3.00.07 also tested)
I wish I read your comment before, cos I bought my new gtx980 yesterday and doesn’t work on vray RT (CUDA) and I bought it for that 🙁
Vray 3.0 doesn’t support 970/980 series because these GPUs use a different way to compile data.
Probably you’ll see support in the upcoming SP.
Didnt yet tried with 3.00.08 if they introduced this support but from changelog it doesn’t seem to be (a demo is available on chaosgroup website).
I cant believe this I just bought 3x GTX 970 and you are telling me that it doesent support the rt render.. Fuck sake… it really doesent work for me either what a fucking waste of money
http://www.nvidia.com/coolstuff/demos#!/apollo-11
Maxwell Apollo 11 Demo