PDA

View Full Version : GTX 980 Ti SLI vs. GTX 1080 Ti SLI on AMD Ryzen 7 1800X



Comment Bot
04-06-2017, 08:05 AM
Please feel free to comment about our story entitled "GTX 980 Ti SLI vs. GTX 1080 Ti SLI on AMD Ryzen 7 1800X (http://www.tweaktown.com/articles/8116/gtx-980-ti-sli-vs-1080-amd-ryzen-7-1800x/index.html)"

Artas1984
06-03-2017, 06:00 AM
Greetings everyone! I have joined TT forums just because of this thread! I am a long time techpowerup forums member and i do custom PC hardware benchmarks in there. I could not help but notice this very interesting, yet very misleading thread! As an owner of SLI GTX 980 Ti Strix OC Gaming video cards i was shocked at the poor results of SLI scaling, and i was even more shocked at how poor the benchmarks was done. Let me indulge you!

1. First of all only 4 games were tested. That's laughable! Whether Anthony Garreffa had little time or he was just lazy including more games - i'd like to know that! In order to provide competitive SLI scaling results of any video cards, at the very least 10 games should be tested. Preferably 20. Have you noticed how many games are being tested at "hardware unboxed" or "gamer's nexus" - this is what i am talking about...

2. Include only SLI supportive games! if a game supports SLI, and the scaling is at least 10 %, but constant, it has to be tested. However, what's the point of testing widely known SLI non-supportive games? I am talking about Rainbow Six Siege. This is well documented fact that R6S does not support SLI, so why even bother?

3. Wrong testing methodology! Does not matter Ryzen 7 or Core i7, the problem is how do you test the game. Sometimes the build-in game benchmark tools are poorly optimized for testing SLI video card configs. I am talking about Metro Last Light benchmark tool - in this case custom single player mission autosaves should have been used.

I've tested SLI GTX970 before. If you want to see just how good SLI can work generally, take at look here:

SLI GTX970 G1 Gaming vs GTX980 Ti Strix OC vs GTX1070 Dual benchmarks | TechPowerUp Forums (https://www.techpowerup.com/forums/threads/sli-gtx970-g1-gaming-vs-gtx980-ti-strix-oc-vs-gtx1070-dual-benchmarks.232448/)

And now i will give my own SLI GTX980 Ti scaling results at 2560X1440. I've tested 20 games, maximum settings, with no AA. The FPS results i will present in Techpoweup forums, but here i will just give you the scaling results, so that you do not judge SLI scaling based upon this failed Tweak Town article, which gives no justice to SLI GTX980 Ti whatsoever, even more so to SLI GTX1080 Ti.

http://img.techpowerup.org/170602/sli-gtx980-ti-desktop.jpg

Intel Core i5 5775C 4.3 GHz OC
Gigabyte GA-Z97X-Gaming 3
2X8 GB DDR3 1600 MHz C8
SLI Asus GeForce GTX 980 Ti Strix OC Gaming 6 GB
Samsung 850 Evo 500 GB
Windows 7 Pro 64 bit
NVIDIA Forceware 382.33

The following 20 games were tested and SLI scaling performance is displayed in percents. Average FPS of SLI GTX980 Ti were divided by average FPS of single GTX980 Ti and multiplied by 100. What follows:

Alien Isolation: 76 %
Assassin's Creed Syndicate: 57 %
Battlefield 1: 52 %
Call of Duty Black Ops 3: 62 %
Crysis 3: 41 %
Deus Ex Mankind Divided: 71 %
Dragon Age Origins: 60 %
Dying Light: 88 %
Far Cry 4: 66 %
Mad Max Fury Road: 76
Mass Effect Andromeda: 67 %
Metro Last Light Redux: 58 %
Middle Earth Shadow of Mordor: 76 %
Mirrors Edge Catalyst: 72 %
Prey (2017): 60 %
Project Cars: 18 %
Rise of Tomb Raider: 17 %
Sniper Elite 3: 85 %
Watch Dogs 2: 48 %
Witcher 3 Wild Hunt: 79 %

As you can see, i've got Rise of Tomb Raider scaling at 17 % (TT has 0 %), Shadow of Mordor scaling at 76 % (TT has 13 %), Metro Last Light Redux scaling at 58 % (TT has 29 %).
Rise of Tomb Raider is one of the worst SLI scaling games out there, yet it is valid for benchmakrs, because there is constant SLI scaling of at least 10 %.
Shadow of Mordor scales great - i have no idea how TT got their scaling so wrong!
Metro Last Light Redux scaling is ok at 29 % by TT; i saw in some in checkpoints similar scaling, but you can get more than that dependent on the map.

Overall based on 20 games average FPS, i've got my SLI GTX980 Ti cards scaling at 61 % at 2560X1440.

That should put SLI GTX980 Ti Strix OC over stock GTX1080 Ti card, probably equal to a highly overclocked GTX1080 Ti. Had the resolution been 4K, the scaling would have been higher.

Cheers!

Artas1984
06-03-2017, 06:04 AM
Greetings everyone! I have joined TT forums just because of this thread! I am a long time techpowerup forums member and i do custom PC hardware benchmarks in there. I could not help but notice this very interesting, yet very misleading thread! As an owner of SLI GTX 980 Ti Strix OC Gaming video cards i was shocked at the poor results of SLI scaling, and i was even more shocked at how poor the benchmark was done. Let me indulge you!

1. First of all only 4 games were tested. That's laughable! Whether Anthony Garreffa had little time or he was just lazy including more games - i'd like to know that! In order to provide competitive SLI scaling results of any video cards, at the very least 10 games should be tested. Preferably 20. Have you noticed how many games are being tested at "hardware unboxed" or "gamer's nexus" - this is what i am talking about...

2. Include only SLI supportive games! if a game supports SLI, and the scaling is at least 10 %, but constant, it has to be tested. However, what's the point of testing widely known SLI non-supportive games? I am talking about Rainbow Six Siege. This is well documented fact that R6S does not support SLI, so why even bother?

3. Wrong testing methodology! Does not matter Ryzen 7 or Core i7, the problem is how do you test the game. Sometimes the build-in game benchmark tools are poorly optimized for testing SLI video card configs. I am talking about Metro Last Light benchmark tool - in this case custom single player mission autosaves should have been used.

I've tested SLI GTX970 before. If you want to see just how good SLI can work generally, take at look here:

SLI GTX970 G1 Gaming vs GTX980 Ti Strix OC vs GTX1070 Dual benchmarks | TechPowerUp Forums (https://www.techpowerup.com/forums/threads/sli-gtx970-g1-gaming-vs-gtx980-ti-strix-oc-vs-gtx1070-dual-benchmarks.232448/)

And now i will give my own SLI GTX980 Ti scaling results at 2560X1440. I've tested 20 games, maximum settings, with no AA. The FPS results i will present in Techpowerup forums, but here i will just give you the scaling results, so that you do not judge SLI scaling based upon this failed Tweak Town article, which gives no justice to SLI GTX980 Ti whatsoever, even more so to SLI GTX1080 Ti.

http://img.techpowerup.org/170602/sli-gtx980-ti-desktop.jpg

Intel Core i7 5775C 4.3 GHz OC
Gigabyte GA-Z97X-Gaming 3
2X8 GB DDR3 1600 MHz C8
SLI Asus GeForce GTX 980 Ti Strix OC Gaming 6 GB
Samsung 850 Evo 500 GB
Windows 7 Pro 64 bit
NVIDIA Forceware 382.33

The following 20 games were tested and SLI scaling performance is displayed in percents. Average FPS of SLI GTX980 Ti were divided by average FPS of single GTX980 Ti and multiplied by 100. What follows:

Alien Isolation: 76 %
Assassin's Creed Syndicate: 57 %
Battlefield 1: 52 %
Call of Duty Black Ops 3: 62 %
Crysis 3: 41 %
Deus Ex Mankind Divided: 71 %
Dragon Age Inquisition: 60 %
Dying Light: 88 %
Far Cry 4: 66 %
Mad Max Fury Road: 76 %
Mass Effect Andromeda: 67 %
Metro Last Light Redux: 58 %
Middle Earth Shadow of Mordor: 76 %
Mirrors Edge Catalyst: 72 %
Prey (2017): 60 %
Project Cars: 18 %
Rise of Tomb Raider: 17 %
Sniper Elite 3: 85 %
Watch Dogs 2: 48 %
Witcher 3 Wild Hunt: 79 %

As you can see, i've got Rise of Tomb Raider scaling at 17 % (TT has 0 %), Shadow of Mordor scaling at 76 % (TT has 13 %), Metro Last Light Redux scaling at 58 % (TT has 29 %).

Overall based on 20 games average FPS, i've got my SLI GTX980 Ti cards scaling at 61 % at 2560X1440.

That should put SLI GTX980 Ti Strix OC over stock GTX1080 Ti card, probably equal to a highly overclocked GTX1080 Ti. Had the resolution been 4K, the scaling would have been higher.

Cheers!