Introduction
Launching cutting-edge products during acute ongoing stock shortages is an interesting move. Nvidia has done precisely that with the arrival of the GeForce RTX 3080 Ti and GeForce RTX 3070 Ti premium graphics cards announced yesterday.
Built to offer an additional bump in performance compared to non-Ti cards debuting last September, today we're putting the RTX 3080 Ti through its paces on our new benchmark suite. Let's evaluate if it's worthy of an update as we roll out the specification table.
Ampere and Turing Compared | |||||||||
|---|---|---|---|---|---|---|---|---|---|
|
RTX 3090
|
RTX 3080 Ti
|
RTX 3080
|
RTX 3070
|
RTX 2080 Ti
|
RTX 2080 Super
|
RTX 2080
|
RTX 2070 Super
|
||
| Launch date |
Sep 2020
|
June 2021 |
Sep 2020 |
Oct 2020 |
Sep 2018 |
July 2019 |
Sep 2018 |
July 2019 |
|
| Codename |
GA102
|
GA102 |
GA102 |
GA104 |
TU102 |
TU104 |
TU104 |
TU104 |
|
| Architecture |
Ampere
|
Ampere |
Ampere |
Ampere |
Turing |
Turing |
Turing |
Turing |
|
| Process (nm) |
8
|
8 |
8 |
8 |
12 |
12 |
12 |
12 |
|
| Transistors (bn) |
28.3
|
28.3 |
28.3 |
17.4 |
18.6 |
13.6 |
13.6 |
13.6 |
|
| Die Size (mm²) |
628.4
|
628.4 |
628.4 |
392.5 |
754 |
545 |
545 |
545 |
|
| PCIe |
4.0
|
4.0 |
4.0 |
4.0 |
3.0 |
3.0 |
3.0 |
3.0 |
|
| Base Clock (MHz) |
1,400
|
1,365 |
1,440 |
1,500 |
1,350 |
1,650 |
1,515 |
1,605 |
|
| Boost Clock (MHz) |
1,695
|
1,665 |
1,710 |
1,725 |
1,545 |
1,815 |
1,710 |
1,770 |
|
| Founders Edition Clock (MHz) |
1,695
|
1,665 |
1,710 |
1,725 |
1,635 |
1,815 |
1,800 |
1,770 |
|
| Shaders |
10,496
|
10,240 |
8,704 |
5,888 |
4,352 |
3,072 |
2,944 |
2,560 |
|
| GFLOPS |
35,581
|
34,099 |
29,768 |
20,314 |
13,448 |
11,151 |
10,068 |
9,062 |
|
| Founders Edition GFLOPS |
35,581
|
34,099 |
29,768 |
20,314 |
14,231 |
11,151 |
10,598 |
9,062 |
|
| Tensor Cores |
328
|
320 |
272 |
184 |
544 |
384 |
368 |
320 |
|
| RT Cores |
82
|
80 |
68 |
46 |
68 |
48 |
46 |
40 |
|
| Memory Size |
24GB
|
12GB |
10GB |
8GB |
11GB |
8GB |
8GB |
8GB |
|
| Memory Bus |
384-bit
|
384-bit |
320-bit |
256-bit |
352-bit |
256-bit |
256-bit |
256-bit |
|
| Memory Type |
GDDR6X
|
GDDR6X |
GDDR6X |
GDDR6 |
GDDR6 |
GDDR6 |
GDDR6 |
GDDR6 |
|
| Memory Clock |
19.5Gbps
|
19Gbps |
19Gbps |
14Gbps |
14Gbps |
15.5Gbps |
14Gbps |
14Gbps |
|
| Memory Bandwidth |
936
|
912 |
760 |
448 |
616 |
496 |
448 |
448 |
|
| ROPs |
112
|
112 |
96 |
96 |
88 |
64 |
64 |
64 |
|
| Texture Units |
328
|
320 |
272 |
184 |
272 |
192 |
184 |
160 |
|
| L2 cache (KB) |
5,120
|
5,120 |
5,120 |
4,096 |
5,632 |
4,096 |
4,096 |
4,096 |
|
| SLI |
Yes
|
No |
No |
No |
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 |
|
| TGP (watts) |
350
|
350 |
320 |
220 |
250 |
250 |
215 |
215 |
|
| Founders Edition TGP (watts) |
350
|
350 |
320 |
220 |
260 |
250 |
225 |
215 |
|
| Suggested MSRP |
$1,499
|
$1,199 |
$699 |
$499 |
$999 |
$699 |
$699 |
$499 |
|
| Founders Edition MSRP |
$1,499
|
$1,199 |
$699 |
$499 |
$1,199 |
$699 |
$799 |
$499 |
|
GeForce RTX 3080 Ti Analysis
There's nothing radically new here. RTX 3080 Ti uses the same GA102 die as the RTX 3090 and RTX 3080. For those that need a refresher on what the Ampere architecture brings to the table, head over here. The specs intimate it is closer to the RTX 3090 in composition, so the best method of discerning potential is to compare it to the flagship GPU.
Whereas RTX 3090 uses an 82-SM approach yielding 10,496 Cuda cores, RTX 3080 Ti trims this down to 80. Crunching the 128-cores-per SM reveals a 10,240-shader GPU endowed with a commensurate ratio of Tensor and RT Cores. As the table shows, the Ti variant is comfortably more powerful than the non-Ti, but overall performance is not as clear cut as the Ti ships with marginally lower clocks. Do be aware that no GeForce uses the full might of the GA102 architecture; that honour is left to the workstation-optimised RTX A6000 housing the full complement of 10,752 cores.
The similarities to RTX 3090 persist through Nvidia choosing to maintain a 384-bit memory bus, albeit with two key differences. The memory footprint is halved to 12GB, which feels stingy, and speed reduced by an effective 500MHz. The other difference is a lack of NV-Link support on the newest member of the Ampere clan, but who runs SLI nowadays anyway?
All the numbers point to the RTX 3080 Ti benchmarking at around 95 per cent of the RTX 3090's levels, perhaps even closer. Nvidia therefore plumbs this GPU with the same TGP of 350W. It's easy to think of this as repurposed silicon that didn't quite make the higher RTX 3090 grade.
Let's be clear. This is an RTX 3090 Lite more than an upgraded RTX 3080, and Nvidia knows this. We can liberally discard MSRPs these days, but for what it's worth, the RTX 3080 Ti comes in at $1,199 (£1,049), further underlining its RTX 3090 Lite credentials. Is this a value-focussed part? Of course not, but when regular RTX 3080 boards are selling for $2,000-plus on popular auction sites, due to a combination of stock shortages and inflated cryptocurrency interest, talking about price is verging on pointless.
Speaking of crypto, like very recent RTX GPUs, Nvidia is limiting the hash-rate potential of this card. There's no mention of this anywhere in the specifications, but if you're thinking of buying one to mine, think again, and we'll show you just how artificially castrated it is later on.
In summary, before we go on a benchmarking journey, know that GeForce RTX 3080 Ti 12GB feels like an interesting but largely unnecessary addition to the high-end gaming fold during times of constraint.