There's a new budget GPU in town, the 3GB version of the GeForce GTX 1050, and this boy is a weird beast.
We recently looked at the disaster, which is the DDR4 version of the GT 1030, a terribly misleading product that shouldn't exist. We expect this to be one of the worst graphics cards in history.
We don't expect this GTX 1050 3GB version to be nearly as bad, in fact it may not be that bad at all. However, we are bringing the GT 1030 into memory because it is an unnecessary, odd and somewhat misleading version.
Before May 2018, there was only one version of the GTX 1050 (code name GP107-300) with a 2 GB memory buffer. This is the same GPU we've seen in retail since October 2016 when we first tested it. However, in May, Nvidia quietly updated the GTX 1050 page on its website to include a 3 GB model that should be used between the existing 2 GB 1050 and 4 GB 1050 Ti, although this is a bit more complicated.
The 3 GB 1050 is not just a 1050 with an additional gigabyte of VRAM. Instead, it has 768 CUDA cores, but on a smaller 96-bit memory bus. This is due to a change in the backend configuration. Ok, so it's a GTX 1050 Ti with a gigabyte less VRAM … no, not that either.
In this table you can see that while it has the same number of CUDA cores as the 1050 Ti, it has 25% fewer ROPs and that means it has 25% less than the original 1050. Although you have the same 7 Gbps GDDR5 Memory received, the memory bus is only 96 bits due to the reduction in ROPs in which one of the 4 memory controllers is deactivated. The result is only 84 GB / s memory bandwidth, compared to 112 GB / s with 2 GB 1050 and 4 GB 1050 Ti.
It also means that the L2 cache capacity has been reduced from 1MB to 768KB. Nvidia clocks the cores higher to compensate for this. However, we believe that in some cases this will not help overcome the 25% reduction in memory bandwidth, ROPs, and cache.
Why is Nvidia selling a 3 GB version of the GTX 1050 that probably doesn't work like a 1050 with an additional gigabyte of VRAM?
Given the specs, we expect it to be mostly between 2GB 1050 and 1050 Ti, but there will undoubtedly be cases where it is actually slower than even 2GB 1050. The most likely explanation is that they had a pool of defects For example, GP107 dies from a faulty memory controller. Nvidia saves these parts by creating the 3 GB 1050. It's a little strange to see this so late in the product cycle, but that's not really the problem here. We again see a problem in offering multiple GPU configurations under the same name.
At least the 3 GB version should be called 1050 SE or LE or something other than just 1050 3 GB. It's no different than the 3GB and 6GB GTX 1060, and I expect a similar result.
We have the EVGA GeForce GTX 1050 SC Gaming on hand for testing. It comes with a small factory overclocking of 5% and currently costs $ 140. This is basically what you can expect for a 2 GB model. This makes it around $ 30 cheaper than the GTX 1050 Ti. Our test consists of 23 games, all of which were tested with different quality settings at 1080p. However, we will only discuss the highlights for 8 of the games and then jump into our performance breakdown that looks at the whole picture.
Our test bench based on the Corsair Crystal Series 570X was used. Inside is a Core i7-8700K with 5 GHz and 32 GB DDR4-3200 memory. We are aware that this hardware configuration is overkill for these GPUs. As usual, the idea is to eliminate potential system bottlenecks that could affect GPU performance. On the good things …
Based on the Battlefield 1 results, we see that the 3 GB 1050 is 9% faster than the 2 GB version and 16% slower than the 1050 Ti. This is a great result for the 3GB version as it is actually faster than the original 1050 in this title. Nevertheless, it is still a bit slower than the 1050 Ti due to the downgrades mentioned earlier.
Despite the positive performance just seen, we see a completely different picture when testing with Wreckfest. Here the 3 GB 1050 was actually slower than the original 1050, actually 13% slower, which meant it was 21% slower than the 1050 Ti. An unsettling result, so let's go ahead.
When testing with Rainbow Six Siege, we found that the 3 GB 1050 is 22% faster than the original 2 GB model and only 11% slower than the 1050 Ti. As with Battlefield 1, this is a very good result for the new 3 GB model.
Okay, we are certainly seeing very mixed results here. This time the 3GB 1050 was a few frames slower than the 2GB model when tested with Far Cry 5, making it 5% slower. This also meant that it was 15% slower than the 1050 Ti.
Fortnite tests show similar performance between the 2GB and 3GB 1050 models, making them both slightly slower than the 1050 Ti.
We also see comparable performance when testing with Total War Saga: Thrones of Britannia. Here the 3 GB model was 3% faster than the original 2 GB version.
Interestingly, when we play ARMA 3, we see an 11% drop in performance for the 3 GB 1050, and as a result, the 1% low result fell below 60 fps. So another bad result for the new 3GB model.
Finally, we'll look at the results of Warhammer Vermintide 2. Here we see a relationship between the 2GB and 3GB versions of the 1050, making them 13% slower than the 1050 Ti.
On the next page, we'll look at the power consumption and full benchmark results for games in 23 games for a more complete picture, as well as our analysis of where the 3 GB GeForce GTX 1050 is located.