RX480 vs GTX1060 (need help with choosing)

Jakub1998

Commendable
Dec 18, 2016
25
0
1,530
Hi there. I was decided to buy gtx 1060 for my christmas pc build, but few days (maybe weeks) ago hardwarecanucks did new benchmarks which showed that they are almost the same in dx11. So my questions are:

1) RX480 or gtx1060?

2) MSI gaming X or Asus strix?
 
Solution
I'll do the easy one 1st. :)

1. MSI

2. 1060 versus 480 ... I'll give yu the info, you make the decision. I'll use the TPU site as they test over the widest number of games.

a) Original testing - As we can see here, the reference 1060 has a 10.2% (97/88) lead over the 480 on Jul 19th, 2016

perfrel_1920_1080.png


b) Overclocking - If you lok at the overclocking headroom, the MSI 1060 6GB OCs 18.8 % and the MSI 480 Oc's 12.2%

So if you are overclocking, the 1060 had a (97 / 88 ) x (1.188 / 1.122) = 16.7% advantage.

c) New Drivers - Several sites have retested with AMDs new drivers. Let's see what that does...
I'll do the easy one 1st. :)

1. MSI

2. 1060 versus 480 ... I'll give yu the info, you make the decision. I'll use the TPU site as they test over the widest number of games.

a) Original testing - As we can see here, the reference 1060 has a 10.2% (97/88) lead over the 480 on Jul 19th, 2016

perfrel_1920_1080.png


b) Overclocking - If you lok at the overclocking headroom, the MSI 1060 6GB OCs 18.8 % and the MSI 480 Oc's 12.2%

So if you are overclocking, the 1060 had a (97 / 88 ) x (1.188 / 1.122) = 16.7% advantage.

c) New Drivers - Several sites have retested with AMDs new drivers. Let's see what that does:

1920.jpg


The new AMD driver increased the average performance over the 21 games by 2.1% over lat version and they state:

The new drivers don't particularly offer significant performance gains over previous drivers; however, AMD's 4-6 % performance gain claims over the 16.7 drivers its Polaris 10 hardware launched with do appear to hold true.

If we do the math ... outta the box, then 10.2% advantage in July has been reduced by 4-6% (say 5% average) leaving the 1060 with a 5.2% average ... If we use the OC'd number of 16.7%, taking off the 5% leaves you with 11.7%. Now you can go web site shopping that **prove** either one to be superior by selecting the games you choose to test but, again, TPU used 21 games and i haven't found anyone else that has that broad a sampling as of yet.

The rest you might not care about but since i don't know what is important, I'll list them.

d) Power Draw - the MSI 480 draws about 200 watts to the 1060s 120 watts, A PSU that's adequate for the 1060, may need to be 100 watts bigger to handle the 480, however, your existing PSU is likely big enough for both so this may not matter, but if it does you will need to send more for the bigger PSU.

e) Heat - The GPU will produce about 80 watts more heat so to keep case temps about the same, you would need 1 extra 120mm fan (one 120m per 75 - 100 watts) to keep temps the same. Again, most folks don't care about this as they either won't care o consider the extra $15 trivial.

f) Electric Cost - The extra wattage required will mean about $50 extra in electric bills over 3 years at 35 hours per week and $0.10 per kw-hr. If in a US urban suburban area with no access to hydropwer, multiply that by 2.5 ... in Europe multiply by 5. Again, few peeps seem to take much interest in this issue.

g) New API's - In the past, AMD has jumped on the new APIs showing a lead early which later evaporated. With these things being new, driver development teams will spend the next few months tweaking... you'll notice that CF is currently scaling better than SLI )obviously this is not in context of the 1060 but in how deveopment teams spend their time) but as of now nVidia has a strong financial interest in not improving SLI performance as the only card which would get hurt by that would be the 1080.... So not quite sure how this will fall out in the end. You'll not that in the DX12 games in TPUs test suite, the 1060 does better in 2 outta 3....and the 3rd one couldn't be tested as they couldn't get it stable.

h) Essentially it comes down to what games you play rather than say a 21 game sampling. Look at the individual performance

 
Solution