Introduction
Nvidia supercharged PC graphics by releasing RTX 30 Series GPUs over the last few months. Aiming to solidify dominance in the premium end of the market, the green team has gone for aggressive recommended pricing in light of rival AMD's RX 6800 Series. Purported cost is relatively academic as all high-end cards sell out as soon as meagre levels of stock lands.
Nevertheless, the next piece in the Ampere jigsaw is the GeForce RTX 3060 Ti, aiming to entice more users to upgrade from older cards. Priced from $399 (£369), Nvidia adopts a familiar tack in developing this card. Let's take a look.
Ampere and Turing Compared | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|
RTX 3090
|
RTX 3080
|
RTX 3070
|
RTX 3060 Ti
|
RTX 2080 Ti
|
RTX 2080 Super
|
RTX 2080
|
RTX 2070 Super
|
RTX 2060
|
||
Launch date |
Sep 2020
|
Sep 2020 |
Oct 2020 |
Dec 2020 |
Sep 2018 |
July 2019 |
Sep 2018 |
July 2019 |
Jan 2019 |
|
Codename |
GA102
|
GA102 |
GA104 |
GA104 |
TU102 |
TU104 |
TU104 |
TU104 |
TU106 |
|
Architecture |
Ampere
|
Ampere |
Ampere |
Ampere |
Turing |
Turing |
Turing |
Turing |
Turing |
|
Process (nm) |
8
|
8 |
8 |
8 |
12 |
12 |
12 |
12 |
12 |
|
Transistors (bn) |
28.3
|
28.3 |
17.4 |
17.4 |
18.6 |
13.6 |
13.6 |
13.6 |
10.8 |
|
Die Size (mm²) |
628.4
|
628.4 |
392.5 |
392.5 |
754 |
545 |
545 |
545 |
445 |
|
PCIe |
4.0
|
4.0 |
4.0 |
4.0 |
3.0 |
3.0 |
3.0 |
3.0 |
3.0 |
|
Base Clock (MHz) |
1,400
|
1,440 |
1,500 |
1,500 |
1,350 |
1,650 |
1,515 |
1,605 |
1,365 |
|
Boost Clock (MHz) |
1,695
|
1,710 |
1,725 |
1,665 |
1,545 |
1,815 |
1,710 |
1,770 |
1,680 |
|
Founders Edition Clock (MHz) |
1,695
|
1,710 |
1,725 |
1,665 |
1,635 |
1,815 |
1,800 |
1,770 |
1,680 |
|
Shaders |
10,496
|
8,704 |
5,888 |
4,864 |
4,352 |
3,072 |
2,944 |
2,560 |
1,920 |
|
GFLOPS |
35,581
|
29,768 |
20,314 |
16,197 |
13,448 |
11,151 |
10,068 |
9,062 |
6,221 |
|
Founders Edition GFLOPS |
35,581
|
29,768 |
20,314 |
16,197 |
14,231 |
11,151 |
10,598 |
9,062 |
6,221 |
|
Tensor Cores |
328
|
272 |
184 |
152 |
544 |
384 |
368 |
320 |
240 |
|
RT Cores |
82
|
68 |
46 |
38 |
68 |
48 |
46 |
40 |
30 |
|
Memory Size |
24GB
|
10GB |
8GB |
8GB |
11GB |
8GB |
8GB |
8GB |
6GB |
|
Memory Bus |
384-bit
|
320-bit |
256-bit |
256-bit |
352-bit |
256-bit |
256-bit |
256-bit |
192-bit |
|
Memory Type |
GDDR6X
|
GDDR6X |
GDDR6 |
GDDR6 |
GDDR6 |
GDDR6 |
GDDR6 |
GDDR6 |
GDDR6 |
|
Memory Clock |
19.5Gbps
|
19Gbps |
14Gbps |
14Gbps |
14Gbps |
15.5Gbps |
14Gbps |
14Gbps |
14Gbps |
|
Memory Bandwidth |
936
|
760 |
448 |
448 |
616 |
496 |
448 |
448 |
336 |
|
ROPs |
112
|
96 |
96 |
80 |
88 |
64 |
64 |
64 |
48 |
|
Texture Units |
328
|
272 |
184 |
152 |
272 |
192 |
184 |
160 |
120 |
|
L2 cache (KB) |
5,120
|
5,120 |
4,096 |
4,096 |
5,632 |
4,096 |
4,096 |
4,096 |
3,072 |
|
SLI |
Yes
|
No |
No |
No |
Yes |
Yes |
Yes |
Yes |
Yes |
|
Power Connector (FE) |
12-pin
|
12-pin |
12-pin |
12-pin |
8-pin + 8-pin |
8-pin + 6-pin |
8-pin + 6-pin |
8-pin + 6-pin |
6-pin |
|
TDP (watts) |
350
|
320 |
220 |
200 |
250 |
250 |
215 |
215 |
160 |
|
Founders Edition TDP (watts) |
350
|
320 |
220 |
200 |
260 |
250 |
225 |
215 |
160 |
|
Suggested MSRP |
$1,499
|
$699 |
$499 |
$399 |
$999 |
$699 |
$699 |
$499 |
$349 |
|
Founders Edition MSRP |
$1,499
|
$699 |
$499 |
$399 |
$1,199 |
$699 |
$799 |
$499 |
$349 |
GeForce RTX 3060 Ti Analysis
Building this GPU from the guts of RTX 3070 FE makes sense. Though occupying the fourth rung of the Ampere ladder, GA104 GPU is still a beast, comprising over 17 billion transistors. Putting that into context, 2017's champion card, GeForce GTX 1080 Ti, harnesses only 12bn transistors, albeit on a larger, less efficient 16nm process. Point is, don't let RTX 3060 Ti's positioning fool you into thinking this is strictly a mid-range solution. It is not.
Cleaving away parts of the RTX 3070 design to achieve a desired price-to-performance ratio is a balancing act. Nvidia opts to lose top-end muscle whilst keeping memory bandwidth the same. Specifically, speaking of GA104 implementations, RTX 3060 Ti drops from 46 SMs to 38. That move has instant ramifications for total shaders, RT Cores and Tensor Cores, which all drop by that 17.4 per cent. Nvidia also reins in frequencies a touch, leading to around 20 per cent less shading potential. This approach is normal when slicing up a GPU for a lower-specced card.
Nvidia, however, keeps the 8GB of GDDR6 memory humming along at 14Gbps, so back-of-the-envelope calculations suggest RTX 3060 Ti will benchmark at between 85-90 per cent of RTX 3070. A fair compromise for the $100 snip. Running at lower clocks and having to keep fewer shaders and cores busy enables Team GeForce to reduce board power to 200W. It is reasonable to think RTX 3060 Ti is built from dies that don't make the requisite 3070 grade.
It's becoming increasingly difficult to compare performance between generations as GPU architects focus on boosting performance in ways that are not immediately obvious on the spec sheet. Looking back at RTX 20 Series, Nvidia alludes to this card performing along the same lines as the RTX 2080 Super, which, remember, was introduced at a $699 price point last year in Founders Edition form. In effect, Nvidia is asking $300 less for similar framerate, at less power. Ordinarily such a move would be met with glee, but things have moved on in the graphics space this year.
The introduction of this GPU puts rival AMD into a tight spot for now. Its Big Navi cards don't yet scale down to the $399 (£369) price point, so the fiscal competition is actually from last year's RX 5700 Series. AMD's not duly worried as even those older cards are selling out like hotcakes, according to our sources. The state of the world, eh?
A little brother to RTX 3070 opens up Ampere to a few more punters lucky enough to pick up a card this week. Looking like a great solution at QHD and passable at UHD, with class-leading ray tracing and framerate-boosting DLSS technology in tow, you're going to want to see benchmarks before ripping the credit card from the wallet.