Nvidia GeForce GTX 1070 Graphics Cards: Price List

The GTX 1070 is based on a GP104 die with 25% of the SMs disabled. It also uses more traditional GDDR5 memory as opposed to the GDDR5X vRAM used on the GTX 1080 and GTX 1080 Ti. Nvidia opted to keep the GPU back-end fully enabled on the GTX 1070, however, which gives it a full 256-bit memory bus and 64 ROPs.

MORE: Best Deals

Nvidia GeForce 10-series "Pascal" GPUs

Swipe to scroll horizontally

GPU	Nvidia GeForce GTX 1080 Ti	Nvidia GeForce GTX 1080	Nvidia GeForce GTX 1070	Nvidia GeForce GTX 1060	Nvidia GeForce GTX 1050 Ti	Nvidia GeForce GTX 1050	Nvidia GeForce GT 1030
Core	GP102	GP104	GP104	GP106	GP107	GP107	GP108
SMs	28	20	15	10	6	5	3
Shader Units	3584	2560	1920	1280 (6GB) / 1152 (3GB)	768	640	384
Texture Units	224	160	120	80	48	40	24
ROPs	88	64	64	48	32	32	16
Transistor Count	12 Billion	7.2 Billion	7.2 Billion	4.4 Billion	3.3 Billion	3.3 Billion	1.02 Billion
Base Clock / Boost Clock	1480MHz / 1582MHz	1607MHz / 1733MHz	1506MHz / 1683MHz	1506MHz / 1708MHz	1290MHz / 1392MHz	1354MHz / 1455MHz	1227MHz / 1468MHz
Memory	Up To 11GB GDDR5X @ 11Gbps, 352-Bit	Up To 8GB GDDR5X @ 10Gbps, 256-Bit	Up To 8GB GDDR5 @ 8Gbps, 256-Bit	Up To 6GB GDDR5 @ 8Gbps, 192-Bit	Up To 4GB GDDR5 @ 7Gbps, 128-Bit	Up To 2GB GDDR5 @ 7Gbps, 128-Bit	Up To 2GB GDDR5 @ 6Gbps, 64-Bit
TDP	250W	180W	150W	120W	75W	75W	30W
Process Node	16nm	16nm	16nm	16nm	14nm	14nm	14nm

MORE: Best Graphics Cards

Asus

EVGA

Gigabyte

MSI

Zotac

TOPICS

Michael Justin Allen Sexton is a Contributing Writer for Tom's Hardware US. He covers hardware component news, specializing in CPUs and motherboards.

22 Comments Comment from the forums

real810thug

Ok, not really an "article" but more of a list of video cards with referral links. Maybe tomshardware can start a new section called clickbait and put it there, as to not confuse readers looking at articles since this is posted under "articles" on the main page. Nice quick money grab though. I'm sure there will be at least a few suckers who will click those links for you.
Reply
Poozle

Meh. Its an article to me, a person who is interested in the increasing prices of Nvidia mid range cards such as this. Personally, it tells me a lot about the cards and how well they are selling as well as the mining issue and how long until Volta is released. The face that Pascal is no longer being manufactured to the amount to keep up with pricing tells me that volta is under 6 months away for sure, and I would bet 4 months until its revealed.
Reply
michaelzehr

I have no complaints about it in part because the headline is completely accurate and the contents were exactly what I expected. If it had said review, or roundup, or something other than prices at the end I might feel differently.
Reply
bit_user

Could use some plots of price over time. pcpartpicker has those, but only for specific models and not overall trends.
Reply
derekullo

How dare you place self updating prices for the Nvidia Geforce GTX 1070 in an article about Nvidia Geforce GTX 1070 prices

That handy chart showing the technical details of all of the Nvidia cards was also completely out of line

In all seriousness I didn't know the Geforce 1050 and the 1030 were made on a smaller process node than the others.

Seems odd they would use totally different manufacturing methods for the low end and not just use the worst of the 1060s with the defected areas disabled, but I'm sure they have their reasons.
Reply
cknobman

No issues with article, title clearly explained its intent.

Plus I just used these types of articles last month when I built my new rig for price comparisons.
Reply
spdragoo

Huh, I was thinking those prices seemed kind of high for a 1070...

Until I saw the prices for Micro Center (http://www.microcenter.com/search/search_results.aspx?N=4294966937+4294840904&NTX=&NTT=&NTK=all&page=1&sortby=pricelow)...

I guess semi-good news, they actually have RX 570 & 580 cards in stock, & their prices are back to being comparable (RX 570 = 1060 3GB, RX 580 = 1060 6GB). No Vega 56 cards, though, just the Vega 64 air-cooled models (single fan?!? & not on AMD reference cards, but on the partner model!?!)... & at their prices, you can get a 1080TI for the same price (or, get a 1080 for the price of a 1070).
Reply
IInuyasha74

20123319 said:
How dare you place self updating prices for the Nvidia Geforce GTX 1070 in an article about Nvidia Geforce GTX 1070 prices

That handy chart showing the technical details of all of the Nvidia cards was also completely out of line

In all seriousness I didn't know the Geforce 1050 and the 1030 were made on a smaller process node than the others.

Seems odd they would use totally different manufacturing methods for the low end and not just use the worst of the 1060s with the defected areas disabled, but I'm sure they have their reasons.

I'm glad to see that the majority of readers see value in these articles and understand their purpose. We have been creating several of these to assist people with new PC builds, by making it easy to find the best prices on parts online. We hastened production of these GPU price lists due to the shortage, and it does help in keeping an eye on the GPU shortage as well.

@derekullo: It was surprising when Nvidia announced the use of 12nm transistors in its lower-end 1000 series GPUs. I am curious if the 14nm process offered notable advantages in power consumption and die area over the 16nm process. I suspect it might have, and Nvidia made the decision to go with 14nm on those lower-end GPUs to better serve laptops. I don't have anything to support that idea, but just a thought. It is noteworthy that the 14nm parts do have lower clock speeds than the 16nm parts. Just a general observation.
Reply
bit_user

20125530 said:
I am curious if the 14nm process offered notable advantages in power consumption and die area over the 16nm process. I suspect it might have, and Nvidia made the decision to go with 14nm on those lower-end GPUs to better serve laptops.
Power efficiency (in GFLOPS/W) of the GTX 1050's is actually worse than most of AMD's GPUs, while the GTX 1030 and 1060 match the bottom of their curve at about 32. The 1050 delivers 23 GFLOPS/W, while the 1050 Ti achieves 26. The 1080 is far more efficient, at about 46 GFLOPS/W, though it's usually the bigger GPUs that are most efficient. Vega 56 is AMD's best, at about 40 GFLOPS/W.

Anyway, this leads me to suspect that the reason for using 14 nm was cost-based, and because GPUs this small aren't so much limited by power or thermals (except in laptops).
Reply
IInuyasha74

It certainly could be price.
Reply

Show more comments