NVIDIA is reportedly preparing faster mobile GPUs based on Maxwell, with a GTX 980MX and 970MX on the way.
The new GTX 980MX would sit between the GTX 980M and the laptop version of the full GTX 980, with 1664 CUDA cores (compared to 1536 with the 980M), 104 Texture Units (up from the 980M's 96), a 1048 MHz core clock, and up to 8 GB of GDDR5. Memory speed and bandwidth will reportedly be identical to the GTX 980M at 5000 MHz and 160 GB/s respectively, with both GPUs using a 256-bit memory bus.
The GTX 970MX represents a similar upgrade over the existing GTX 970M, with CUDA Core count increased from 1280 to 1408, Texture Units up from 80 to 88, and 8 additional raster devices available (56 vs. 48). Both the 970M and 970MX use 192-bit GDDR5 clocked at 5000 MHz, and available with the same 3 GB or 6 GB of frame buffer.
WCCFtech prepared a chart to demonstrate the differences between NVIDIA's mobile offerings:
| Model | GeForce GTX 980 Laptop Version | GeForce GTX 980MX |
GeForce GTX 980M |
GeForce GTX 970MX | GeForce GTX 970M | GeForce GTX 965M |
GeForce GTX 960M |
|
Architecture |
Maxwell | Maxwell | Maxwell | Maxwell | Maxwell | Maxwell | Maxwell |
| GPU | GM204 | GM204 | GM204 | GM204 | GM204 | GM204 | GM107 |
| CUDA Cores | 2048 | 1664 | 1536 | 1408 | 1280 | 1024 | 640 |
| Texture Units | 128 | 104 | 96 | 88 | 80 | 64 | 40 |
| Raster Devices | 64 | 64 | 64 | 56 | 48 | 32 | 16 |
| Clock Speed | 1218 MHz | 1048 MHz | 1038 MHz | 941 MHz | 924 MHz | 950 MHz | 1097 MHz |
| Memory Bus | 256-bit | 256-bit | 256-bit | 192-bit | 192-bit | 128-bit | 128-bit |
| Frame Buffer | 8 GB GDDR5 | 8/4 GB GDDR5 | 8/4 GB GDDR5 | 6/3 GB GDDR5 | 6/3 GB GDDR5 | 4 GB GDDR5 | 4 GB GDDR5 |
| Memory Frequency | 7008 MHz | 5000 MHz | 5000 MHz | 5000 MHz | 5000 MHz | 5000 MHz | 5000 MHz |
| Memory Bandwidth | 224 GB/s | 160 GB/s | 160 GB/s | 120 GB/s | 120 GB/s | 80 GB/s | 80 GB/s |
| TDP | ~150W | 125W | 125W | 100W | 100W | 90W | 75W |
These new GPUs will reportedly be based on the same Maxwell GM204 core, and TDPs are apparently unchanged at 125W for the GTX 980MX, and 100W for the 970MX.
We will await any official announcement.
I will take improvements
I will take improvements anywhere I can, but can we have more products that are like the laptop variant of GTX 980.
But why, we don’t need more
But why, we don’t need more models at the higher end. It just adds more confusion and market segmentation.
Why would nvidia launch these
Why would nvidia launch these cards? Aren’t they about to launch a new series of GPU’s? This just seems odd to me, unless they are having problems with Pascal.
Yes, nVidia might be having
Yes, nVidia might be having problems with Pascal, and/or they felt threatened by AMD/RTG showing off Polaris running Star Wars Battlefront. Those chips are likely to end up in laptops)
I don’t think so, man. I
I don’t think so, man. I think they’re just milking Maxwell as much as possible. Which is exactly what they should be doing.
For Nvidia, its more like
For Nvidia, its more like cutting off their nose to spite their face.
Ever notice how when either
Ever notice how when either Nvidia or AMD has something come out in the media(whether leaked or officially released) that the other one releases something very shortly there after? Basic marketing tactics as a way of taking some of the spotlight back.
They could be replacing the
They could be replacing the current mobile parts because yields are sufficient that they have a lot of stock built up with the higher number of units without defects. With the old 28 nm process, yields could be quite high. Alternatively, they could just keep all of them around and try to make more money by charging a significantly higher price for these GPUs with more units activated.
The GPU market has lower margins partially because of the limited number of products made from each device. If you look at Intel Xeons, they might get 20 or more different products out of the same die. If they make an 18 core chip, they would then sell it as a 16, 14, 12, and maybe even 10 core part. Add varying clock speeds and varying amounts of cache activated and you have a huge number of variants. They don’t have to sell them as either an 18 core part or a 10 core part with nothing in between. For desktop GPUs, they may get 2 or 3 variants out of a single GPU die variant. They take a big hit for selling the salvaged part, since it might be cut down significantly (like half). Mobile can add a few extra variants by having different number of functional units and different clock speeds. This will allow them to sell more of their salvaged die at a higher price.
I don’t think that this necessarily indicates any issues with pascal. We are still probably at least 6 months from wide availability of 14 nm GPUs, so there is quite a bit of time to sell these parts. Nvidia probably already had these salvaged parts stocked up. Most people are not going to buy a high end GPU right now. Enthusiast mostly already should have them and are waiting for 14 nm parts. Laptops will probably still keep on selling though, so they might as well try to unload these parts as mobile GPUs. I need a new laptop because my old one is broken beyond repair. I am not going to wait 6 months without a laptop.
My guess is these will
My guess is these will replace the 970M/980M’s. They are, after all failed full fledged GM204 cores with groups of cores disabled. My guess is yields are high enough on the cards to up the amount cores they can leave on every chip. This is mature silicon since the 980 has been around for sometime. I think they should just go all out full fledged 980’s and remove the M/MX versions. Having the same graphics card in both Desktop/laptop would be great. I know this is currently being done but in only very few top tier laptops. But I guess they have to fill all parts of the market, so these different variations make sense. The more options they have, the more cores they can get rid of that dont cut the snuff as a full fledged GM204 cores.
Are these the Maxwells Nvidia
Are these the Maxwells Nvidia tried to pass as Pascals in their Drive X2.
Nvidia never claimed they
Nvidia never claimed they were. There was speculation that the cards were Pascal-based, for all of the 5-10 minutes until someone actually looked at the cards and went “hey, those are identical to the 980 MXM modules”.
the 970 mx has an interesting
the 970 mx has an interesting ROP count of 56, does this mean it has an extra “half” memory controller?
Why would anyone buy 28 nm
Why would anyone buy 28 nm mobile gpus when 14/16 nm with hbb 2.0 should be out this year which will probably get you the same performance levels as mid level at significantly way lower power draw. Amd showed of polaris with gddr5 imagine it with hbm 2.0 and 100 watt tdp.
Hopefully we get more HBM
Hopefully we get more HBM enabled GPUs this year. It requires a different design; the low end parts coming out first (Polaris 10?) will almost certainly be GDDR5. The higher end part (Polaris 11?) might be HBM. It is somewhat unclear whether the higher end part that is supposed to come out this year is the one that will replace the Fury. Given the massive process upgrade, even a mid-range size 14 nm GPU may be able to outperform the large 28 nm parts, so it could be a replacement for the Fury. The Fury is close to 600 square mm being sold for $650. This would be much more expensive if it wasn’t an old 28 nm, high yield process. I don’t know if we will ever see such a large GPU on 14 nm. The yields just might be too low.
What do the sales curves look
What do the sales curves look like for high end GPUs? I would expect that those that want them would mostly already have bought them and current sales would just be a trickle. This may just be a way to try to get rid of built up stock.
So….will GeForce GTX 980MX
So….will GeForce GTX 980MX specification got ‘miscommunicated’ again?
64 ROP? Or is it 56 ROP? 4GB 256 bit or segmented 3.5GB 224 bit + 0.5GB 32 bit