Geforce GTX 1050 / 1050 Ti Launch Thread ($109 / $139 - October 25th)

[happy medium]

We aren't the audience for these cards anyway. And while (desktop) perf/watt is typically overrated by whatever side is ahead in perf/watt, when you reach the threshold where you don't need to use a six pin power connector that actually is quite important. I expect the 1050 to be in a boatload of cheap OEM systems.

That said, it really is a shame that the 1050 only comes with 2GB of memory. It makes it very difficult to recommend over the 460 unless you absolutely can't use a six-pin. And the 1050 Ti is difficult to recommend because it's priced to high relative to the significantly higher performance 470.

A 4GB 1050 at around the same price would be awesome, and an instant recommendation. And of course, the chip promises to be one of the all time best laptop GPUs ever, but I can't help but be disappointed in these particular products.

A 460/ 1050 class gpu will rarely use more than 2gb of vram before it falls on its face and is almost unplayable. They are as slow as a 7950 or gtx670.

 

[frozentundra123456]

You can run a 470 system with a 300W PSU no sweat, even an aftermarket 470.

https://www.computerbase.de/2016-08/radeon-rx-470-test/5/

At most, the entire system consumes 250W (83% of 300W or 300W is 20% more) and this is in a system with a 4.5GHz i7.

*Maybe* on a good quality 300 watt psu, although I personally still would not try it. In any case, I certainly would not jerry rig a molex adapter to a poor quality OEM "300 watt" psu and attempt to run a 470 on it.

 

[SPBHM]

A 460/ 1050 class gpu will rarely use more than 2gb of vram before it falls on its face and is almost unplayable. They are slower than a 7950 or gtx770.

I think it's not that simple, I mean some games can run OK on those GPUs and with some settings, like texture settings clearly benefit from more ram, but, on most games with a 1050/460 you are probably better reducing a lot of settings due to GPU speed, so also keeping the vram limitation in mind is probably not a huge deal right now.... I think it's acceptable for those cards to exist with 2GB, but I don't see 4GB for them as a complete waste, it's not like a r7 240/GT 740 4GB DDR3 or something like that.

 

[happy medium]

I think it's not that simple, I mean some games can run OK on those GPUs and with some settings, like texture settings clearly benefit from more ram, but, on most games with a 1050/460 you are probably better reducing a lot of settings due to GPU speed, so also keeping the vram limitation in mind is probably not a huge deal right now.... I think it's acceptable for those cards to exist with 2GB, but I don't see 4GB for them as a complete waste, it's not like a r7 240/GT 740 4GB DDR3 or something like that.

I agree, but very few games 4gb's makes a difference with slow cards like this....and they are usually the most demanding games.

 

[f2bnp]

A 460/ 1050 class gpu will rarely use more than 2gb of vram before it falls on its face and is almost unplayable. They are as slow as a 7950 or gtx670.

Not this again. What does this even mean? You can lower settings as you will, that's why you're gaming on a PC. You can lower lighting and shadow settings, but you can keep Texture Quality on the highest setting when using a 4GB card for no performance penalty.

Your GTX 960 4GB is also about as fast as a 7950, why didn't you pick the 2GB version?

 

[frozentundra123456]

Not this again. What does this even mean? You can lower settings as you will, that's why you're gaming on a PC. You can lower lighting and shadow settings, but you can keep Texture Quality on the highest setting when using a 4GB card for no performance penalty.

Your GTX 960 4GB is also about as fast as a 7950, why didn't you pick the 2GB version?

You have any benchmarks to show that you can run highest texture quality without any performance hit on weak cards like the 460 and 1050? I mean if there is no performance penalty for high textures, why do they even make the setting adjustable? Just make one max setting only. As far as the 960 goes, despite the general consensus that the 4gb card is much faster, the latest test I saw showed very little difference between the two cards except in one or two isolated games. For instance this test

 

[Erenhardt]

You have any benchmarks to show that you can run highest texture quality without any performance hit on weak cards like the 460 and 1050? I mean if there is no performance penalty for high textures, why do they even make the setting adjustable? Just make one max setting only. As far as the 960 goes, despite the general consensus that the 4gb card is much faster, the latest test I saw showed very little difference between the two cards except in one or two isolated games. For instance this test

Welcome back to the year 2015 - a demise of 2GB GPUs:

4GB 960 23% faster than 2GB version in 1080p. On average across whole indie crap filled 2015 year of testing. Give me a break!
2GB cards have no place in 2016 and beyond. They were dead in 2015 already. Why do we have to witness this zombie of a GPU apocalypse? Let the dead be dead.

Join me in lighting up the candles for 2GB cards!
[*] - RIP 2GB GPUs

Also, FYI I always adjusted settings on my 2GB 7870 to maximize visual quality to the limits of the card, which was basically trying to not spill from 2GB VRAM. Texture quality had close to no perf impact as long as I had enough memory. From low to high textures, the performance barely moved (5% if even that). Very high would often cause a large drop and introduce stutter.

 

[AtenRa]

*Maybe* on a good quality 300 watt psu, although I personally still would not try it. In any case, I certainly would not jerry rig a molex adapter to a poor quality OEM "300 watt" psu and attempt to run a 470 on it.

Majority of OEM systems are using the 35-65W Core i3/5s. RX 470 4GB is 120W TDP, that means you dont really need more than 160-200W from the wall.
Most OEM PSUs are 85+ because they need to be ENERGY STAR qualified. That means you can even use 220-250W 85+ PSUs with a 65W TDP CPU + 120W TDP dGPU easily.

Now, there is only a niche market for someone to choose the GTX 1050Ti over the RX 470. That is if you have a very low wattage PSU lower than 200W and your System chassis is small that will only fit the GTX 1050Ti. In that case yes the GTX 1050Ti is the faster card you can buy. Otherwise, even if you dont have a 6-pin connector the RX 470 4GB is the faster card and way better choice over the GTX 1050Ti.

Edit:

For comparison, Newegg has some refurbished old Core i3 systems

This one is a Lenovo with Core i3 530, it has 280W 85% efficiency or 320W 88% efficiency PSU.
Even the SFF model has a 240W 85% Efficiency PSU.

Specs link.

This one is a HP Core i3 2100 SandyBridge, it has a 240W 90% Efficiency at 50% Load or 87% efficiency at 100% Load PSU.

Specs link

 

[coercitiv]

Doesn't look that way to me.

How does it look then? Tom's Hardware saw a custom 1050 Ti use 67W@1800Mhz in a typical gaming workload: while this is probably still worse than what an underclocked 1060 can offer in terms of absolute performance, the 1050Ti might actually get close enough efficiency wise once you bring clocks down into the same frequency domain of 1400-1500Mhz.

Before reviews all we knew was the card might use 75W @ 1500Mhz, after reviews we know the card can use 67W@1800Mhz or 75W@1900Mhz which is quite a different story. This is why people on the forum recommended that you wait and see how they stack up. It doesn't make the product a winner, but it does give it a place in the price/performance/power balance sheet for people with mini boxes such as yourself. Keep in mind that when power efficiency is concerned, the bigger chip will always have the upper hand - a small 1070 properly underclocked will likely eat 1060 for breakfast. But there's a price to pay for bigger chips, isn't it?

 

[mohit9206]

You can run a 470 system with a 300W PSU no sweat, even an aftermarket 470.

https://www.computerbase.de/2016-08/radeon-rx-470-test/5/

At most, the entire system consumes 250W (83% of 300W or 300W is 20% more) and this is in a system with a 4.5GHz i7.

Just because you can run doesn't mean you should especially not on an oem prebuilt pc psu. Putting a huge risk of it causing fire and killing every other component.
Do you think my Antec BP450 would be able to run an overclocked rx470 with i5 cpu ?

A 460/ 1050 class gpu will rarely use more than 2gb of vram before it falls on its face and is almost unplayable. They are as slow as a 7950 or gtx670.

7950 happily used 3gb vram without falling flat on its face and 1050Ti is faster than 7950. Its as fast as a 7970 so by that simple logic, it should have no trouble fully utilizing 4gb vram before it runs out of gpu power.

I agree, but very few games 4gb's makes a difference with slow cards like this....and they are usually the most demanding games.

Slow card? Is 7970/R9 380X performance considered slow these days? These cards are little monsters. Fully capable of using 4gb vram.
Even rx460 is capable of using more than 2gb without falling flat on the face as some of the vram heavy game benchmarks have shown us.
Anything rx460 and above class of cards are powerful enough to utilize entire 4gb frame buffer WITHOUT running out of gpu horsepower.

 

[mohit9206]

Also another small point that will be attacked by some of you but its still necessary to point it out.
Let's compare $140 1050ti vs $170 rx470.
Considering average 46% advantage for only 21% more money makes rx470 a better choice than 1050Ti as pointed out by swilli89 above..
Now add in the cost of power bill. According to Guru3D 1050Ti review,
Cost per Year 5 days week / 4 hrs day / would cost $24.31 on 1050Ti. So RX470 consuming double power would cost $50 so a difference of $25 which would mean actual cost of rx470 over a period of one year would be $170+25=$195 versus $140 for 1050Ti.
Now doing the math, you get 46% more performance for $39% more money. Not such big difference now is it?
Outright purchase cost makes rx470 a better buy at msrp prices but ownership over a period of one year shows that both are equally good value.
Of course keeping the card for 2-3 yrs would would shift the advantage towards the 1050Ti...
Also before RussianSensation comes in saying you can make up that cost and actually even make profit by mining with RX470, yeah i agree maybe you can but mining is a different thing altogether so its irrelevant to this discussion.

 

[DamZe]

Also another small point that will be attacked by some of you but its still necessary to point it out.
Let's compare $140 1050ti vs $170 rx470.
Considering average 46% advantage for only 21% more money makes rx470 a better choice than 1050Ti as pointed out by swilli89 above..
Now add in the cost of power bill. According to Guru3D 1050Ti review,
Cost per Year 5 days week / 4 hrs day / would cost $24.31 on 1050Ti. So RX470 consuming double power would cost $50 so a difference of $25 which would mean actual cost of rx470 over a period of one year would be $170+25=$195 versus $140 for 1050Ti.
Now doing the math, you get 46% more performance for $39% more money. Not such big difference now is it?
Outright purchase cost makes rx470 a better buy at msrp prices but ownership over a period of one year shows that both are equally good value.
Of course keeping the card for 2-3 yrs would would shift the advantage towards the 1050Ti...
Also before RussianSensation comes in saying you can make up that cost and actually even make profit by mining with RX470, yeah i agree maybe you can but mining is a different thing altogether so its irrelevant to this discussion.

No one is going to feel the power consumption difference between those two cards unless you have them full throttled 24/7. The actual difference in a real-world scenario for the average user is miniscule! The 1050Ti is an underpowered/overpriced modern day 960 4GB edition, for a little more you get a true 1080p gaming card in the RX470/480(4GB). The 1050 cards have no legs to stand on, there is nothing to be gained from saving a few bucks on electricity in the next 2-3 years while having to suffer through horrible performance in games.

 

[f2bnp]

You have any benchmarks to show that you can run highest texture quality without any performance hit on weak cards like the 460 and 1050? I mean if there is no performance penalty for high textures, why do they even make the setting adjustable? Just make one max setting only. As far as the 960 goes, despite the general consensus that the 4gb card is much faster, the latest test I saw showed very little difference between the two cards except in one or two isolated games. For instance this test

No, I don't have any at hand. I speak from personal experience, this has been an issue since the dawn of 3D accelerators.

Voodoo2 SLI vs Voodoo3 (8MB texture memory vs 16MB unified, both cards have roughly equal performance otherwise)
FX 5900 XT vs FX 5900/Ultra (128MB vs 256MB)
GeForce 6800 LE vs 6800 GT (6800 LE unlocked to 6800 GT speeds/config, but 128MB limited options)
7800 GTX 256MB (this was insanely lame, I used to own this card and know it first hand, 6 months down the line we got the 7800 GTX 512MB)
8800 GT 256MB vs 512MB
3850 256MB vs 512MB
4870 512MB vs 1GB

And the list goes on and on. We're not talking silly cards like a GT 630 with 4GB VRAM nor are we talking about a case similar to the GeForce4 Ti 4200 128MB vs 64MB (the 64MB memory was clocked higher, thus was faster). Stuttering makes games unplayable IMO, the closest I can compare it with is uncapped games on consoles with pretty bad frametimes, but even that just doesn't do it justice.
In order to get rid of it, you have to lower Texture Quality and that makes a huge difference in IQ.

Heck, for the past couple of days I've been fooling around with a dual Pentium III 1.4GHz and an FX 5900 Ultra 256MB. That card was pretty lame even when it was brand new. It performs terribly if you use higher settings, to the point where some games see 100% better performance if you use older shaders (1.1 instead of 2.0). If anyone's interested in why, you should take a look at what happened with the GeForce FX cards during the Shader Model 2 era (DirectX 9a and 9b) and the Radeon 9700 Pro.
However, the 256MB VRAM instead of 128MB + nice memory bandwidth for the time allow it to use maximum textures in 2004-2005 and even some 2006 games. For example, Dawn of War and FEAR run like crap if you turn most settings on, but texture quality is set to the maximum since it doesn't really affect performance. And that really really helps IQ.

 

[coercitiv]

Do you think my Antec BP450 would be able to run an overclocked rx470 with i5 cpu ?

Your PSU is rated for 400W on the 12V rail. Typical gaming load for aftermarket RX470 + i5 is under 150W+50W = 200W which is exactly the sweet spot for your PSU in terms of average load, leaving safe room for other components and even additional overclocking.

 

[Sweepr]

Radeon RX 470 4GB vs Geforce 1050 Ti 4GB:
Hexus: 470 69% faster

62% - by testing only 6 games, 3 of which are Gaming Evolved titles + messed up Doom results where RX 470 is 25% faster than a 1060

Eurogamer: 470 50% faster

45%

PCGames Hardware: 470 52% faster

51%

Didn't bother to look at the others, but I assume you messed up the calculations as well. Add these to the random list of averages: 21% @ Hardware.info, 31% @ SweClockers, 35,5% @ HardwareCanucks (not counting GTA V where it's actually slower than a 1050 Ti). So no, 45-50% is not the conservative gain expected.

Even compared to $140, a 470 at $169 is 21% more expensive. At this super low price point, who wouldn't spend an extra 21% for 46% more performance? That's a 2:1 performance to price increase! Nvidia would have to drop the 1050ti down to $110 even for this to begin even break even.

It is the fastest 'pinless' card out there, has better media support than the competition and at $139 (like Zotac Mini) offers very reasonable performance per $ (almost matching discounted RX 470), as pointed by reviewers. OEMs will love it and now that we know GTX 1050 (non Ti) is 10% faster than RX 460 we can conclude the Ti has no direct competitor (even full P11 would be slower), which is why 232mm² (larger and more power hungry) Polaris 10 has to do the job.

https://tpucdn.com/reviews/MSI/GTX_1050_Ti_Gaming_X/images/perfdollar_1920_1080.png
http://www.techspot.com/articles-info/1269/bench/Scatter.png

 

[happy medium]

y
my post got screwed up.

 

[antihelten]

Sorry but no I don't accept that summary from one site. Here is a completely random selection from your original post list of reviews. I just chose 6 sites at random and compiled the indexes for each. I'm more than happy to upload my spreadsheet to Google Sheets if anyone wants to see it. What I did to make sure its a fair comparison: I used the best of either DX11, DX12, or Vulkan (DOOM) for all cards. Its not fair to include DX12 results for the 470 when the 1050ti actually performance worse in some cases so I chose the best of each. In the real world a gamer would not use a new API just to get worst performance so I basically included each cards best score when a site presented multiple APIs. Here are the results at 1080p:

Radeon RX 470 4GB vs Geforce 1050 Ti 4GB:

Lanoc Reviews: 470 46% faster
Tweaktown: 470 42% faster
Hot Hardware: 470 47% faster
Computerbase: 470 34% faster
Hexus: 470 69% faster
Eurogamer: 470 50% faster
PCGames Hardware: 470 52% faster
TechPowerUp: 470 25% faster

Average: 46% faster

On Newegg the MSI 1050ti that is referenced in many of these reviews is $169, and here is a great quality custom 470 for $165 as well: http://www.newegg.com/Product/Product.aspx?Item=N82E16814131696

So obviously at this price point there is absolutely no discussion, the 470 is without a doubt a much, much better purchase. Its a different class of performance and for some games is actually enough to bump from 1080p to 1440p and for most games this is doable with reduced settings.

Now let's give the 1050ti the benefit of the doubt and say custom cards come down to $140 (out of 14 there is literally 1 at this price point on Newegg, vast majority are $150+).

Even compared to $140, a 470 at $169 is 21% more expensive. At this super low price point, who wouldn't spend an extra 21% for 46% more performance? That's a 2:1 performance to price increase! Nvidia would have to drop the 1050ti down to $110 even for this to begin even break even.

Of note, the two lowest results (computerbase and TPU both used reference 470's, all other sides paired their custom 1050 ti's with a custom 470).

It appears that TPU is an extreme outlier and that the conservative average is something more like the 470 is about 45-50% faster.

I think it's important to note here that there are two different classes of 1050 Ti. There's the compact, ITX form factor, single fan version, which is the one being tested in most of the above reviews and then there's the normal length, dual fan versions, like the MSI 1050 Ti Gaming X

Small form factor 1050 Ti reviews:
Lanoc Reviews: 470 46% faster*
Hot Hardware: 470 47% faster
Computerbase: 470 34% faster*
Hexus: 470 69% faster*
Eurogamer: 470 50% faster
PCGames Hardware: 470 51% faster

*uses non-reference RX 470

Average: 49% faster

Full size 1050 Ti reviews:
Tweaktown: 470 42% faster*
TechPowerUp: 470 25% faster
PCGames Hardware: 470 31% faster (only tests the full size 1050 Ti in three games, where it is on average 15% faster than the reference 1050 Ti, so the results are extrapolated from those)

*uses non-reference RX 470

Average: 32% faster

So it looks like a reference RX 470 is about 40% faster than a small form factor 1050 Ti, and an aftermarket RX 470 adds about 5-10% to this for a gap of about 50%. Meanwhile the same reference RX 470 is only about 25-30% faster than the full size 1050 Ti, with aftermarket RX 470 being about 35-40% faster.

So all in all it is important to keep in mind that the two different kinds of 1050 Ti cards have a roughly 15% performance gap between them. So depending upon which RX 470 card and which 1050 Ti card we are discussing, both the 25-30% and the 40-50% number can be true.

Of course as you mention the full length MSI 1050 Ti Gaming X is currently at $170, however other full length versions are being sold for $150, whilst the small versions can be had for as little as $140 (compared to a $5-10 gap between reference and OC versions of the RX 470). So obviously price is a factor here as well.

 

[vissarix]

lol at the desperate amd red army trying to compare the gtx1050ti vs a rx470 on an gtx1050ti thread...might aswell compare the rx470 with the gtx1060 which is way faster at a little higher price tag...

 

[USER8000]

Ok in that case 40% faster.
But then rx470 is also 40% more expensive than basic 1050Ti models(in my country) so both cards are good in their respective budgets.

Well in the UK,the GTX1050TI cards are from £140 onwards. The ones in stock and actually tested are £150 to £160,so much so Hexus changed their conclusions once they saw how much the card they tested actually cost.

Overclockers UK had a decent RX470 for £165 on offer the day before the GTX1050TI launch and RX470 AIB cards can be had for £180 onwards(as are a few GTX1060 3GB cards).

So at this point,you are looking at 40% maybe 50% faster cards for £40 more.

I also have an AIB GTX960 4GB card which cost me under £150 last year and we had 2GB GTX960 and 2GB R9 380 cards for as low as £120 to £130 and even the odd R9 380X for just over £150 on offer.

So at this point,the GTX1050TI has had a rather tepid reception on UK forums. It has its niche mind you if you want a fast bus powered card,but its not really much of any improvement over what I had for the last year.

Your PSU is rated for 400W on the 12V rail. Typical gaming load for aftermarket RX470 + i5 is under 150W+50W = 200W which is exactly the sweet spot for your PSU in terms of average load, leaving safe room for other components and even additional overclocking.

My Core i7 3770 system with a few hard drives and an SSD is lucky to see 200W at the wall when gaming. I also had a GTX660TI in there before.

I use a 450W PSU and have done for years. That means a RX470 or GTX1060 will be fine.

 

[linkgoron]

lol at the desperate amd red army trying to compare the gtx1050ti vs a rx470 on an gtx1050ti thread...might aswell compare the rx470 with the gtx1060 which is way faster at a little higher price tag...

So it's ok to compare the 1050 and the rx 460 because the 1050 wins (almost easily) in that comparison, while comparing the 1050ti and the rx470 is wrong because the 1050ti loses soundly according to most benchmarks (some sites place it at 40%-50% faster, and even in the slower ones its usually at least 25%-30% better)?

Anyway, here are some reviewers who are also part of the red army:

guru3d:

Performance wise I had hoped the 1050 Ti would be much closer to the AMD Radeon RX 470, but it simply isn't. It is an affordable fun card, but is totally outperformed by that RX 470. Perhaps that is wishful thinking from my side as the 1050 series is to be positioned in the competitive Radeon RX 460 range. But then again a 1050 Ti will cost you 139 USD. For 169 USD you can pick up that RX 470 which is the better investment to make in the 1080P domain.

hot hardware:

. If you can afford the additional investly, however, the RX 470 offers significantly higher performance in most instanaces.

Hardware Canucks:

As a matter of fact, our YouTube editor Eber has a Gigabyte GTX 1050 Ti G1 Gaming on hand which will retail for a cool $169. $169! I can’t begin to tell you how poor those metrics look; a GTX 1050 Ti at $139 or even $149 makes perfectly good sense but paying upwards of that for moderately increased frequencies should be avoided at all costs, particularly with the RX 470 hovering at $169. I'll actually make this choice easy: if you want to spend $160 to $180 on a GPU, the RX 470 4GB is the hands-down winner provided it ends up hitting AMD's newly reduced MSRP.

lanoc:

The GTX 1050 Ti is just a good chance to get a little more performance for anyone who has a little extra room in the budget at the end of the day but keep in mind that the RX 470 is available for $30 more and it see's a significant bump in performance.

eurogamer:

However, in terms of price vs performance, it's hard to ignore the charms of the most expensive cards. An additional 35 per cent cash outlay gets you upwards of 60 per cent more performance in the form of the GTX 1060 3GB (albeit with a memory downgrade), while AMD is also aiming to cut prices on the four gig RX 470. It's not as fast as its Nvidia competitor, but provided you have a decent i5 processor or better, it will provide a noticeable higher level of performance.

hexus:

Bottom line: the EVGA GeForce GTX 1050 Ti SC Gaming is a competent choice for full-HD gaming, but if you're able to stretch your budget by another £20-£30, the GTX 1060 and RX 470 pack a far greater punch.

including the following:

Priced too close to GTX 1060/RX 470

I didn't go over other reviews, but I assume some of them also compare the rx 470 and the 1050ti.

 

[LTC8K6]

The 1050Ti is priced perfectly if you don't have, or don't want to use, a power connector.
The mini 1050Ti cards are going to be very popular.

Later on we could see other card versions such as a 4gb 1050 card.

However, a 4gb 1050 card makes little sense to me, because if I need 4gb of vram, I probably need more horsepower than that class of card can provide.

 

[USER8000]

Welcome back to the year 2015 - a demise of 2GB GPUs:

4GB 960 23% faster than 2GB version in 1080p. On average across whole indie crap filled 2015 year of testing. Give me a break!
2GB cards have no place in 2016 and beyond. They were dead in 2015 already. Why do we have to witness this zombie of a GPU apocalypse? Let the dead be dead.

Join me in lighting up the candles for 2GB cards!
[*] - RIP 2GB GPUs

Also, FYI I always adjusted settings on my 2GB 7870 to maximize visual quality to the limits of the card, which was basically trying to not spill from 2GB VRAM. Texture quality had close to no perf impact as long as I had enough memory. From low to high textures, the performance barely moved (5% if even that). Very high would often cause a large drop and introduce stutter.

But this is the other issue - many reviewers only had GTX960 2GB launch cards to compare the GTX1050TI 4GB cards to which probably were the blower type reference cards. So once you consider the fact that the GTX960 4GB cards are all AIB and running at higher clockspeeds,the only real thing the GTX1050TI 4GB has is a bit less power draw and perhaps improved DX12 performance.

This is the problem with most of the whole generation sadly. In some ways the GTX1070 is probably the best balance of a very meh set of launches.

 

[Gikaseixas]

If you compare it to the 460, it's a better alternative

If you compare it to the 470, it's a dud

Pick your poison

 

[Headfoot]

Also another small point that will be attacked by some of you but its still necessary to point it out.
Let's compare $140 1050ti vs $170 rx470.
Considering average 46% advantage for only 21% more money makes rx470 a better choice than 1050Ti as pointed out by swilli89 above..
Now add in the cost of power bill. According to Guru3D 1050Ti review,
Cost per Year 5 days week / 4 hrs day / would cost $24.31 on 1050Ti. So RX470 consuming double power would cost $50 so a difference of $25 which would mean actual cost of rx470 over a period of one year would be $170+25=$195 versus $140 for 1050Ti.
Now doing the math, you get 46% more performance for $39% more money. Not such big difference now is it?
Outright purchase cost makes rx470 a better buy at msrp prices but ownership over a period of one year shows that both are equally good value.
Of course keeping the card for 2-3 yrs would would shift the advantage towards the 1050Ti...
Also before RussianSensation comes in saying you can make up that cost and actually even make profit by mining with RX470, yeah i agree maybe you can but mining is a different thing altogether so its irrelevant to this discussion.

The power delta will never get even remotely close to that much actual power consumed in the real world

 

[Bacon1]

According to Guru3D 1050Ti review,
Cost per Year 5 days week / 4 hrs day / would cost $24.31 on 1050Ti. So RX470 consuming double power would cost $50 so a difference of $25 which would mean actual cost of rx470 over a period of one year would be $170+25=$195 versus $140 for 1050Ti.

Not only are your numbers probably wrong unless you have super expensive power (show your math?).

But if you have the time to game 20 hours per week, you should really do yourself a favor and buy a higher end card and monitor. I mean take a week off gaming and you could easily make $200 or more.

100 watt difference (way more than we'd see between 1050 ti and 470) at 28 hours per week (more than your 20) would only cost $18 per year @ 12c/kwh. 21 hr/w @ 12c is only $13 per year.