The GTX 1060 3GB ($200) card's existence is curious. The card was initially rumored to exist prior to the 1060 6GB's official announcement, and was quickly debunked as mythological. Exactly one month later, nVidia did announce a 3GB GTX 1060 variant – but with one fewer SM, reducing the core count by 10%. That drops the GTX 1060 from 1280 CUDA cores to 1152 CUDA cores (128 cores per SM), alongside 8 fewer TMUs. Of course, there's also the memory reduction from 6GB to 3GB.
The rest of the specs, however, remain the same. The clock-rate has the same baseline 1708MHz boost target, the memory speed remains 8Gbps effective, and the GPU itself is still a declared GP106-400 chip (rev A1, for our sample). That makes this most the way toward a GTX 1060 as initially announced, aside from the disabled SM and halved VRAM. Still, nVidia's marketing language declared a 5% performance loss from the 6GB card (despite a 10% reduction in cores), and so we decided to put those claims to the test.
In this benchmark, we'll be reviewing the EVGA GTX 1060 3GB vs. GTX 1060 6GB performance in a clock-for-clock test, with 100% of the focus on FPS. The goal here is not to look at the potential for marginally changed thermals (which hinges more on AIB cooler than anything) or potentially decreased power, but to instead look strictly at the impact on FPS from the GTX 1060 3GB card's changes. In this regard, we're very much answering the “is a 1060 6GB worth it?” question, just in a less SEF fashion. The GTX 1060s will be clocked the same, within normal GPU Boost 3.0 variance, and will only be differentiated in the SM & VRAM count.
For those curious, we previously took this magnifying glass to the RX 480 8GB & 4GB cards, where we pitted the two against one another in a versus. In that scenario, AMD also reduced the memory clock of the 4GB models, but the rest remained the same.
Today will remain focused strictly on GTX 1060 cards. That means the charts will not have the dozen other cards we've tested on them, for sake of ease and a focused test. If you want to know the results of an RX 480 vs. GTX 1060, or of all the other cards out this generation, check here:
-
RX 480 Review (prior to 1060 launch)
-
GTX 1080 Hybrid vs. Sea Hawk review (high-end cards)
GTX 1060 3GB vs. 6GB Specs Differences
NVIDIA Pascal vs. Maxwell Specs Comparison | |||||
GTX 1080 | GTX 1070 | GTX 1060 3GB | GTX 1060 6GB | GTX 960 | |
GPU | GP104-400 Pascal | GP104-200 Pascal | GP106 Pascal | GP106 Pascal | GM204 |
Transistor Count | 7.2B | 7.2B | 4.4B | 4.4B | 2.94B |
Fab Process | 16nm FinFET | 16nm FinFET | 16nm FinFET | 16nm FinFET | 28nm |
CUDA Cores | 2560 | 1920 | 1152 | 1280 | 1024 |
GPCs | 4 | 3 | 2 | 2 | 2 |
SMs | 20 | 15 | 9 | 10 | 8 |
TPCs | 20 | 15 | 9 | 10 | - |
TMUs | 160 | 120 | 72 | 80 | 64 |
ROPs | 64 | 64 | 48 (?) | 48 | 32 |
Core Clock | 1607MHz | 1506MHz | 1506MHz | 1506MHz | 1126MHz |
Boost Clock | 1733MHz | 1683MHz | 1708MHz | 1708MHz | 1178MHz |
FP32 TFLOPs | 9TFLOPs | 6.5TFLOPs | 3.85TFLOPs | 3.85TFLOPs | 2.4TFLOPs |
Memory Type | GDDR5X | GDDR5 | GDDR5 | GDDR5 | GDDR5 |
Memory Capacity | 8GB | 8GB | 3GB | 6GB | 2GB, 4GB |
Memory Clock | 10Gbps GDDR5X | 4006MHz | 8Gbps | 8Gbps | 7Gbps |
Memory Interface | 256-bit | 256-bit | 192-bit | 192-bit | 128-bit |
Memory Bandwidth | 320.32GB/s | 256GB/s | 192GB/s | 192GB/s | 115GB/s |
TDP | 180W | 150W | 120W | 120W | 120W |
Power Connectors | 1x 8-pin | 1x 8-pin | 1x 6-pin | 1x 6-pin | 1x 6-pin |
Release Date | 5/27/2016 | 6/10/2016 | August, 2016 | 7/19/2016 | 01/22/15 |
Release Price | Reference: $700 MSRP: $600 |
Reference: $450 MSRP: $380 |
MSRP: $200 | Reference: $300 MSRP: $250 |
$200 |
Before getting to the heart of the content, just a quick note to clear things up: We've seen some comments asking why this isn't a “4GB card instead of a 3GB card.” It's not possible. At least, not with this current 1060 setup, and not in a way that makes sense.
Because of the way the GTX 1060 builds its memory subsystem, including use of a 192-bit interface, it is not possible to run a 4GB solution. Generally, when you see a multiple of 3 – like a 6GB card – that card will only ship in other multiples of 3, like a 3GB card. The same is true for 4GB and 8GB units – you'll never see a 4GB card with a 3GB alternative; it will always be 4GB or 2GB, or 4GB or 8GB, and so forth.
It's not quite so easy to put memory modules of new capacities or with a different channel count onto a board, then tell a GPU (which is built for another memory spec) to work with that setup. This wasn't a case of “nVidia had an option to do 4GB, but did 3GB for cost,” it was a case of not being possible to do a 4GB setup on the GP106-400 GPU architecture.
Comparing the Silicon
Above: The GTX 1060 FE GPU
Above: The small 3GB card torn-down. Same GPU.
Above are two photos of the torn-down cards under test: The GTX 1060 6GB FE and the EVGA GTX 1060 3GB. The GPU, plainly, is the same. GP106-400-A1 for each. Here's a block diagram of the GP106-400 GPU:
Imagine this with one less SM, and you've got the GTX 1060 3GB.
Continue to the next page for testing methodology.