On hand is an overclocked model from MSI, taking the 1050 Ti out of the box I am immediately shocked by just how small it is. This is another reason why I have the MSI 1080 on hand, just look a the size difference here.
Now that we have taken a quick look at MSI’s GTX 1050 Ti, let’s rip off that cooler to get a better look at the PCB, while we are at it we might even take a peek at that cheeky little heatsink.
The entire card weighs just 329 grams which is almost 3x less than most mid-range to high-end graphics cards. Including the fan, plastic shroud and tiny aluminium heatsink the cooler totals 193 grams. The heatsink measures just 95 mm long, 90 tall, 18mm thick and tips the scales at just 118 grams, I’ve seen tablets with bigger heatsinks.
Like the standard 1050 the Ti model doesn’t require an external 6-pin PCIe connector and MSI hasn’t felt the need to include one, which I quite like. They have also stuck with a basic 3+1 power phase design though this should be more than enough to keep this 75-watt TDP card happy.
Unlike the standard 1050, the Ti model, uses the full GP107 silicon and this means 12 working SM units allowing it to pack 20% more CUDA cores, for a grand total of 768. There are also 48 TMUs and 32 ROPs. Nvidia has set the default clock specifications at 1290 MHz for the base clock and 1392 MHz for the boost clock. MSI has beefed up these frequencies at tad by pushing the base to 1341 MHz resulting in a boost frequency of 1455 MHz. This is a 5% overclock so we aren’t expecting huge improvements here.
All 1050 Ti cards will come fitted with 4GB of GDDR5 memory clocked at 1750 MHz and MSI hasn’t tinkered with the operating frequency here. Armed with what seems like an inadequate 128-bit wide memory bus the 1050 Ti grabs data from the memory at a rate of 112 GB/s. This is the same bandwidth available to the RX 460 which also boasts a 128-bit wide memory bus.
Getting back to the MSI card we find at the I/O end of the card a DisplayPort 1.4, HDMI 2.0 and Dual-Link DVI output. These outputs should have most users well covered and all three can be used simultaneously if need be.
Alright so at this point we should have pretty well covered all the important stuff. Before jumping to the benchmarks a few quick notes on testing. Many of you have requested we use the Skylake Core i3 test machine for this review, this is something we have done in the past for sub-$200 cards reviews and again we feel it makes more sense than the beefy Core i7 machine for these entry-level graphics cards.
The latest AMD and Nvidia drivers were used for testing and while the results for other cards such as the RX 460 have been carried over from previous test we did go through the results with the updated drivers and made changes as need be. For the most part though, things stayed much the same.
Okay finally the moment you guys have been waiting for, the benchmarks, let’s get to it…
Where else to start but Battlefield 1. Yes, we did only test using DX11, but no we aren’t trying to make Nvidia look good, put those pitchforks down. In our opinion the DX12 implementation is a little too buggy at the moment and both AMD and an Nvidia GPUs suffer from regular dips in performance.
So, this is why we are sticking with DX11 for now. The game is extremely well optimized using DX11 and runs like a treat on hardware from both teams. The GTX 1050 just managed to edge out the 950 which wasn’t bad but here we see much more impressive performance from the Ti model. The 18% boost in performance took the average frame rate from 49 fps to 58 fps. This meant the 1050 Ti was 5% faster than the GTX 960 and not a great deal slower than the R9 380, quite an impressive result then.
Deus Ex: Mankind Divided
For those of you who simply couldn’t forgive us for testing Battlefield 1 using the DX11 API, maybe we can find some favor back by including some Deus Ex Mankind Divided results. Here the 1050 Ti was 15% faster than the vanilla 1050, but more importantly matched the RX 460 4GB card. Even with the medium quality settings enabled, performance here wasn’t exactly stellar and a massive step forward can be seen when moving to the R9 380.
Doom with Vulkan, yep we are trying to keep everyone happy here. Doom really loathes 2GB graphics cards and for that reason the standard 1050 struggles a little with the minimums. That said we are gaming at 1080p with the game virtually maxed out so a 47 fps minimum is still very acceptable. Still this makes the 1050 Ti’s 73 fps minimum even more impressive and despite using Vulkan it was still able to best the RX 460 4GB card, so a big win for Nvidia here.
Gears of War 4
The 1050 Ti might have only been 11% faster on average when compared to the base model 1050 but that was enough of a boost to reach a 60 fps average. It also meant that the Ti was 25% faster than the RX 460. That said it was still 7% slower than the GTX 960 and 12% slower than the R9 380.
Even I was wondering why we bothered testing Mafia III. Really the only reason being that this is a newly released game another other than the shocking optimization it is quite fun to play. Using the low preset, I know low, freaking hell, we averaged just 43 fps with the 1050 Ti. This placed it on par with the R9 380X and as such meant it was 30% faster than the RX 460.
Mirror’s Edge Catalyst
Mirror’s Edge Catalyst is a great quality game and here the 1050 Ti was 12% faster than the standard 1050. Rendering 66 fps on average also meant the Ti was 32% faster than the RX 460 and just 4% slower than the R9 380. Interestingly it was also 6% faster than the GTX 960.
I realize these new GPUs aren’t going to have any issues with Overwatch. That said as a super competitive first person shooter gamers crave high frame rates in this title so being able to push well over 60 fps using the ultra quality settings at 1080p is a nice luxury. The 1050 Ti certainly delivers here averaging 113 fps and never dipping below 99 fps. This made it 55% faster than the 4GB RX 460.
Finally before getting to the overclocking, power consumption and temps we are going to take a look at the frame rate performance in Quantum Break. This is another controversial title to test with and since we are using the Steam version we are limited to DX11 testing. Here the 1050 Ti averaged 75 fps using the medium quality preset and this was obviously very playable. Shockingly this also meant that the 1050 Ti was 60% faster than the RX 460.
As far as I can tell this is a bit like the Battlefield 1 situation. The AMD graphics cards deliver much the same performance using DX11 with the Steam version as they do with the DX12 enabled Windows Store version. Meanwhile the GeForce graphics cards suck in the DX12 version, but are much stronger using Steam’s DX11 version.
Overclocking: Gears of War 4
Using MSI Afterburner we pushed the core to a base clock of 1552 MHz which gave us a boost clock of 1665 MHz. The memory went from 1750 MHz to 1927 MHz. This saw the 1050 Ti operating at between 1860 to 1900 MHz when gaming.
As a result we saw a mild 6% performance gain in Gears of War 4 taking the average frame rate from 60 fps to 64 fps. Not an earth shattering result but a free performance bump all the same.
Overclocking: Battlefield 1
Testing with Battlefield 1 saw an 8% performance gain from our overclock and that was enough to match the GTX 960, so not a bad result then.
I have been very keen to check out the power consumption figures and the 1050 Ti doesn’t disappoint here. Our entire Core i3 rig consumed just 118 watts when paired with the 1050 Ti, and that is gaming consumption folks. So basically you are looking at similar power consumption to the already efficient GTX 750 Ti. The 1050 Ti doesn’t make the RX 460 look to bad here but then again it was consistently faster while allowing the system to consume almost 20% less power.
The MSI GTX 1050 4G OC features a heatsink with less surface area than my smartphone. Even though it is only tasked with keeping 75-watt TDP rated GPU cool, it’s still very small. Nevertheless operating temperatures were kept in check and the card never exceed 72 degrees in our game tests. More importantly it was barely auditable and certainly couldn’t be heard over the existing case fans. Overclocked temperatures only increased by a few degrees and the fan only spun up ever so slightly.
Right so for now we are considering this as more of a preview, the big 20 plus game performance break down is still to come. Soon I will be comparing the 1050 2GB head to head against the RX 460 2GB and then of course the 1050 Ti 4GB vs RX 460 4GB and I might even sneak the RX 470 into that comparison as well.
Cost Per Frame
For now let’s take a look at the cost per frame data based on the small sample of games just tested. Fully aware that some of these games didn’t come out in AMD favors, in fact quite the opposite is true so again this won’t be our final conclusion. That said the RX 470 which we have priced at the new $170 MSRP comes out at a rather impressive $1.88 per frame making it one of the better value options. The GTX 1060 3GB also looks quite good as well.
The GTX 1050 Ti is on par with the 3GB 1060 so that’s good news, particularly given it seems to have obliterated AMD’s RX 460 range, even with the updated price cuts taken into account. Worse still for AMD the vanilla 1050 comes in at a much lower cost per frame while delivering more frames overall. In terms of cost per frame the standard 1050 is very similar to the RX 470 in terms of value.
For those of you who don’t find the scatter plot that easy to digest we have hacked together a bar graph to try and illustrate the situation in a more simplified manner. This isn’t really the correct way to show this data but we are aware that quite a few of you prefer this method. Anyway you can plainly see that the RX 460 series costs more per frame while delivering less frames overall.
Having now seen how the new 1050 series performs it is clear why AMD responded by slashing prices over the weekend. Granted knocking $10 off the MSRP for the RX 460 and 470 cards isn’t a game changer, it does help to strengthen their position. Based on our testing so far it does look as though AMD could shave a bit more off the RX 460 pricing. That said the RX 470 looks very attractive at the new $170 MSRP, we just hope it won’t be too difficult to pick up partner cards at this price.
Given the RX 470 was on average 34% faster than the GTX 1050 Ti we have to wonder if it isn’t worth just spending the $30 to get the more powerful AMD card, basic calculations would suggest that it is.
As for MSI’s cute little 1050 Ti OC graphics card, we rather like it. The fact that it can still be overclocked to 6 to 8% more performance without the need for an external power connector is awesome. It also ran at very reasonable temperatures and generated very little noise. For those rocking older systems with slightly dodgy power supplies this new graphics card might be the answer.
Alright guys that’s all from me for now. I hope you enjoyed this review and be sure to keep an eye out those upcoming head to head comparisons, they ought to be good!