Planning to upgrade from GTX 780 to GTX 1060, how much of a difference will I notice?

Iridar51

Estimable
Apr 14, 2014
14
0
4,560
Hello!

I own a ASUS GTX 780 DC2OC 3GD5 Graphics Card, and I'm more or less satisfied with it's gaming performance, but some of the VR titles are giving it trouble.

A friend of mine owns a EVGA SSC GTX 1060 6 GB, so we ran a 3D Graphics Passmark benchmark on both PCs, and there seem to be minimal difference in actual computational power between these two GPUs.

MpDpCkH.png

QAzOGoV.jpg


Does GTX 1060 offer some other boost for VR Gaming, like by using some sort of advanced technologies?

Because otherwise, I'm having a hard time justifying going from 9500 of something to 9900.

Should I just wait for consumer volta in a few months?

My goal is being able to comfortably play any VR game, but I don't really care about graphics settings (as long as they're not too potato) or super sampling.
 
Solution


It's never worth waiting for the next big thing when you can enjoy something right now and use it imo. I can't predict what consumer Volta will do in VR .

A good basic overview of Pascal improvements for VR https://wccftech.com/pc-vrps-vr-gap-bigger-nvidias-pascal/,

Brent at HARDOCP also does a lot of work benching VR and discussing the tech involved both on his YT channel and on his website.

Iridar51

Estimable
Apr 14, 2014
14
0
4,560


MSI Z270 Gaming M3 with Core i5 7600k and 4 x 4 GB RAM running at 3000 MHz.



Why, though? Clearly there's not so much difference in normal gaming.
 

Mark RM

Estimable
Jul 16, 2014
222
1
5,110
Because it (1060) literally has optimizations for VR built right into the hardware, something that wasn't considered when the 780 was launched.

If you can stretch to the 1070 you will be very pleased with VR. The 1060 definitely works better than the 780 for all VR tasks, that's a given.
 

Iridar51

Estimable
Apr 14, 2014
14
0
4,560

I could try stretching it, yes, but do you think it's worth waiting for consumer Volta? Maybe one of those would be more cost-efficient?

Could you please give more specifics on those "hardware optimizations"? Or where I could read more about them?
 

Mark RM

Estimable
Jul 16, 2014
222
1
5,110


It's never worth waiting for the next big thing when you can enjoy something right now and use it imo. I can't predict what consumer Volta will do in VR .

A good basic overview of Pascal improvements for VR https://wccftech.com/pc-vrps-vr-gap-bigger-nvidias-pascal/,

Brent at HARDOCP also does a lot of work benching VR and discussing the tech involved both on his YT channel and on his website.

 
Solution

Iridar51

Estimable
Apr 14, 2014
14
0
4,560


Thanks a lot, exactly what I needed. I'll still do some research regarding Volta, though.
 

Iridar51

Estimable
Apr 14, 2014
14
0
4,560


I'm using HTC VIVE, should've specified, sorry.

I think I'm gonna go with 1070 Ti in the end, so at least it's an upgrade in non-VR titles as well.
 

Iridar51

Estimable
Apr 14, 2014
14
0
4,560


Can you help me out a bit more, please? I wanna get a GTX 1070 Ti, but around here (Russia) prices jump all over the place like crazy.

I was going to get a decent deal for MSI GTX 1070 Ti ARMOR 8G, but that deal slipped away, and the next best thing is
Gigabyte GTX 1070 Ti GAMING (GV-N107TGAMING-8GD).

Now I know that I can get any GTX 1070 Ti and it'll work fine, but obviously I wanna be the most cost effective.

What has me worried is that MSI card had a 6 + 8 pin power connectors, and a lighter cooler, while Gigabyte card has only 8 pin connector, though with a beefier triple fan cooler.

Wouldn't that mean the Gigabyte card gets less power and has a worse overclock?

I'm not overly concerned with noise, but I would like to get a decent overclock if possible, to somewhat future proof the GPU. It's really hard to justify spending so much money otherwise, and can't say I had a good experience trying to overclock my GTX 780, so I'd prefer easy and reliable overclocking experience.

Can you give me any recommendations here? Here are the prices I'm dealing with, not sure if the site will display properly for you, though.

EDIT: Seems I've found another reasonable deal for MSI ARMOR 8G. Probably gonna go with it, unless you think there are better options?
 

Mark RM

Estimable
Jul 16, 2014
222
1
5,110
I haven't found OC's on any of the Ti's to be power limited, an eight pin connector provides a theoretical 150W +75 watts from PCIe for 225 watts.

http://www.tomshardware.com/reviews/nvidia-geforce-gtx-1070-ti-8gb,5311-16.html
Here they are showing max torture test consumption at 180W at stock clocks on the torture test and max use at 233Watts on the torture test at maximum OC.

Eight extra watts over a PCIe eight pin is nothing, I know for a fact I've pushed some of them to 300 , depends more on the PSU and the wiring Gauge. The nice thing about multiple PCIe connectors for board manufacturers is it takes such things are wire gauge out of the equation and they can just rely on standard specifications.

But for your personal use in this scenario with it only being a few watts out of spec, I really wouldn't worry about that particular problem as long as your PSU is a tier one type product - especially knowing that even at peak overclock you're not going to leave you PC on max torture test for weeks at a time with it's max OC.

During the gaming test with their max OC the card was only using 183 watts typically. It only boosted to 233 during the torture test.

So really, it's about which cards you can get good service for in the event you need an RMA down the road, something people need to consider. Also, there are NO factory overclocked cards in the 1070Ti, so you're choosing from good cooling and build quality.
 

Iridar51

Estimable
Apr 14, 2014
14
0
4,560


Brilliant, thank you!

I have a Corsair RM550x PSU, so everything should be fine on that front. I'm still going with MSI ARMOR 8G in the end, maybe having an MSI GPU with MSI Motherboard will help me squeeze extra something out of it. I still have an option of going with Gigabyte one. Out of the two, which do you think is better?
 

delta5

Honorable
Dec 29, 2012
48
0
10,610


Why would you go with the 8GB and not the 11GB? I have the MSI 1080ti armor 11GB and it is amazing.
 

Iridar51

Estimable
Apr 14, 2014
14
0
4,560


Because I'm not made of money and it's not like 8 GB of VRAM isn't a complete and utter overkill for 1080p gaming.

Here in Russia Graphics Card cost about 30% more just because, and that's on top of cryptocurrency mining craze that made prices skyrocket.
 

Mark RM

Estimable
Jul 16, 2014
222
1
5,110


I tend to like MSI in video cards more, but I just seem to like their shrouds and appearance. I have an MSI repair center nearby too, which makes that decision a lot easier if I need warranty service.
 

Iridar51

Estimable
Apr 14, 2014
14
0
4,560
I ended up getting an ASUS GTX 1060 3GB DUAL card. Current situation with GPU prices in Russia is just too obscene. Even for this GTX 1060 I had to pay $300, a 1070 Ti would run nearly $600.

I haven't done a lot of gaming since then, though VR games seem to be running a bit more smoothly.

I ran a few Passmark benchmarks though, and I have a hard time explaining them: https://imgur.com/a/lvZJ5

If you compare GTX 780 and GTX 1060 3GB at passmark.com, you'll get this:

https://www.videocardbenchmark.net/compare.php?cmp%5B%5D=2525&cmp%5B%5D=3566

3D Mark of 8000 for GTX 780 and 8700 for GTX 1060 3GB, a ~9% upgrade. Seeing those numbers is what initially turned me away from upgrading to GTX 1060, felt like I'd be paying $300 just to get nearly the same performance, except for VR titles, where 1060 is way more optimal.

Once I ran the benchmark on my own GTX 780, and saw a 3D Mark score of 9500, I felt like something is fishy, so I asked around, and found out that one of my friends recently upgraded to a GTX 1060. Per my advice, no less. So I asked him to run a passmark on his EVGA SSC 1060 6GB, and it scored a 3D Mark of 9900, which one could argue is more or less in line with numbers on passmark.com.

We could say that I have a really good GTX 780 and he has a really good GTX 1060, so naturally both would score higher than average score on passmark.com.

The point remained that GTX 1060 seemed only a bit more powerful than GTX 780.

However, as said, due to financial constraints and current market craze, I couldn't justify getting anything above GTX 1060, so I decided to get, with the thought of using it for VR games for now (wouldn't want to buy VIVE just so it collects dust), and saving the money now to get consumer Volta or w/e later.

So I got an ASUS GTX 1060 3GB DUAL - just because it was a good deal in a local shop (still $300 though!!!), and benchmarked it right away.

Imagine my surprise when it scored 11250 on 3D Mark, an 18% upgrade over previous GTX 780.

I'm having a hard time explaining all these inconsistencies. The most apparent conclusion seems to be that passmark is useless crap, both the benchmarking tool itself, and website statistics. They're as useful as average patient temperature in a hospital, with the inclusion of morgue and infectious deceases wing.

Note that this ASUS card comes with a factory overclock in so called "gaming" mode. However, switching it to "OC mode" had nearly no effect on 3D Mark score.