• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

1080ti Custom vs 2070S FE vs 5700XT 50th AE Bench Thread

Soldato
Joined
18 Oct 2002
Posts
4,548
I will be updating over next 2-3 weeks.

Test System
7700k @ 4.7GHZ
Windows 10 1903
16GB RAM
DDR4 3766 12-12-12-1T Dual Channel

AMD Clock adjustments done using Wattman
Nvidia Clock adjustments done using Afterburner.

All 3 cards work fine with a freesync monitor - in this case a Eizo 1440p 144hz display.

Framerates and frametimes captured using OCAT.
Powerdraw readings taken from a physical power-meter at socket.

GPU's

Card 1 1080ti Hall-Of-Fame (with bitspower liquid-cooling block)
http://www.galax.com/en/graphics-card/hof/galax-geforcer-gtx-1080-ti-hof.html

Card 2 2070 Super Founders Edition - air
https://www.nvidia.com/en-gb/geforce/graphics-cards/rtx-2070-super/

Card 3 5700XT 50th AE - air
https://www.amd.com/en/products/graphics/amd-radeon-rx-5700-xt-50th-anniversary

Game 1 - Forza 4 Horizon - In Game Benchmark

Power Draw approx (total system)

1080ti FE - 320w
2070S - 285w
5700XT - 237w
5700XT (matching 2070S In-game Boost Clock) - 325w

1440p - Custom Settings - all maxxed 4x MSAA

1080ti HOF
Founders Edition Boost Clock
100% PL - 1898mhz/5505 - 83.25fps

Stock Clock
100% PL - 1949mhz/5505 - 85.49fps

Stock Clock + 28% Power Limit
128% PL - 1949mhz/5505 - 84.77fps

2070S FE
Stock Clock

Note: actual boost clock at stock is approx 1935mhz.
100% PL - 1770mhz/7000 - 75.46fps - 13.25ms frame time - 99th percentile 15.28ms

120% PL - 1770mhz/7000 - 81fps - 12.72ms frame tme - 99th percentile 15.10ms

Reviewers note: 2070S very quiet however very hot to the touch.

5700XT
Stock Clock -1MHZ with undervolt to 955mv. Fans on auto approx 1500-2100rpm
Note: actual boost clock at near stock is approx 1852mhz.

100% PL - 1904mhz/875 - 78.79 fps - 12.69 frame time - 99th Percentile 15.06ms

Game 2 - Gears Of War 5 - 1440p - All Ultra - In Game Benchmark

5700XT
Stock Clock -1MHZ with undervolt to 955mv. Fans on auto approx 1500-2100rpm
Note: actual boost clock at near stock is approx 1852mhz.

100% PL - 1904mhz/875 - 74.1 fps - 13.5ms frame time 99th percentile 16.1ms


2070S FE
Note: actual boost clock at stock is approx 1935mhz.
100% PL - 1770mhz/7000 - 82.9fps - 12.06ms frame time - 99th percentile 20ms

World War Z - Vulkan - Epic Store Edition

1440p - Ultra - TAA+FXAA - 2x Resolution Scale

1080ti HOF
Founders Edition Boost Clock
100% PL - 1898mhz/5505 - 4229 - 71fps - 14.29ms average frame time

Stock
100% PL - 1949mhz/5505 - 4348 - 73fps - 13.89ms average frame time

Stock Clock + 28% Power Limit
128% PL - 1949mhz/5505 - 4380 - 73fps - 13.80ms average frame time

5700 XT
2070S FE Boost Clock Match
100% PL - 2004mhz/875/1100mv - possible driver bug

AE Stock Clock
100% PL - 1980mhz/875/1047mv - possible driver bug

Non-AE Stock
100% PL - 1904mhz/875/955mv - 3220 - 54 fps - xms frame time - 99th Percentile xms

Minimum Undervolt Clock
100% PL - 1429/875/682mv - 1886 - 31 fps - xms frame time - 99th Percentile xms

2070S
2070S FE
Note: actual boost clock at stock is approx 1935mhz.
100% PL - 1770mhz/7000 - 4251 - 71fps - 14.19ms frame time - 99th percentile 16.93ms

Watchdogs 2

1440p - Ultra

1080ti HOF Stock
100% PL - 1949mhz/5505 - 84fps - 11.9ms - 99th percentile 15.9ms

5700XT

2070S FE Boost Clock Match
100% PL - 2004mhz/875/1100mv - 68fps - 14.6ms - 99th percentile 17.5ms

AE Stock Clock
100% PL - 1980mhz/875/1047mv - 68 fps - 14.7ms frame time - 99th Percentile 17.8ms (big jumps gone)

Non-AE Stock
100% PL - 1904mhz/875/955mv - 61 fps - 16.5ms frame time - 99th Percentile 20.4ms (jumps to 55.4ms)

Minimum Undervolt Clock
100% PL - 1429/875/682mv - 48.77 fps - 20.50ms frame time - 99th Percentile 29.5ms (jumps to 100ms)

2070S
2070S FE
Note: actual boost clock at stock is approx 1935mhz.
100% PL - 1770mhz/7000 - 75fps - 13.4ms frame time - 99th percentile 18.6ms

Monster Hunter World

1440p - High-res texture pack, settings maxxed

1080ti HOF Stock
100% PL - 1949mhz/5505 - 65.94fps - 15.7ms - 99th percentile 16.66ms

5700XT

2070S FE Boost Clock Match
100% PL - 2004mhz/875/1100mv - 61.6fps - 16.2ms - 99th percentile 17.82ms

AE Stock Clock
100% PL - 1980mhz/875/1047mv - 60.39 fps - 16.56ms frame time - 99th Percentile 17.91ms

Non-AE Stock
100% PL - 1904mhz/875/955mv - 57.61 fps - 17.36ms frame time - 99th Percentile 18.82ms

Minimum Undervolt Clock
100% PL - 1429/875/682mv - 48.77 fps - 20.50ms frame time - 99th Percentile 22.30ms

2070S
2070S FE
Note: actual boost clock at stock is approx 1935mhz.
100% PL - 1770mhz/7000 - 4251 - 71fps - 15.3ms frame time - 99th percentile 19.1ms

The Division 2

1440p
1080ti HOF Stock
100% PL - 1949mhz/5505 - 79.95fps - 12.51ms - 99th percentile 15.27ms

2070S FE
100% PL - 1770mhz/7000 - 78.8fps - 12.7ms frame time - 99th percentile 15.9ms

5700 XT
Note: actual boost clock reported is 1865mhz approx 90mhz off the 2070S FE, card suffering from clock speed droop as heat-soak sits in.
100% PL - 1949mhz/875 - 67.3fps - 14.9ms - 99th percentile 19 ms

Note: actual boost clocks reported are approx 1865-1907mhz which better matches the 2070S FE,
card suffering from clock speed droop as heat-soak sits in.
100% PL - 1949mhz/875 - 62fps - 16.1ms - 99th percentile 22.8 ms

1080p
1080ti HOF Stock
100% PL - 1949mhz/5505 - 110.69fps - 9ms - 99th percentile 13.24ms

2070S FE
100% PL - 1770mhz/7000 - 110.9ps - 9ms frame time - 99th percentile 13.3ms

5700 XT
Note: actual boost clock reported is 1865mhz approx 90mhz off the 2070S FE, card suffering from clock speed droop as heat-soak sits in.
100% PL - 1949mhz/875 - 88.4fps - 11.3ms - 99th percentile 15.1 ms

Note: actual boost clocks reported are approx 1865-1907mhz which better matches the 2070S FE,
card suffering from clock speed droop as heat-soak sits in.
100% PL - 2004mhz/875 - 94.3fps - 10.6ms - 99th percentile 19.7ms
 
Last edited:
9900KF Testing

CPU Set at 6 cores 4.9ghz. RAM 3600 C14.

Gears 5 Retail Results - In Game.

Average FPS Average frame time (ms) 95th-percentile frame time (ms) 99th-percentile frame time (ms)

2070 FE (Stock Air)

4k: tbc
1440p: 73.8 13.5 15.1 15.6
1080p: 104.4 9.6 11 11.5

5700XT FE 50th (Stock water)

4K: 35.3 28.3 30.1 31.1
1440p: 66 15.2 16.7 17.1
1080p: 90.2 11.1 12.4 12.8

1080ti HOF (Stock water)

4K: 42.9 23.3 24.8 25.6 29.2
1440p: 79.8 12.5 13.7 14.3 17.4
1080p: 113.8 8.8 9.8 10.3 11.9
 
Last edited:
@Besty interesting.
On the 5700XT you used MSI AB for overclocking hence the 100% PL? Or soft tables?
 
Last edited:
Looking forward to more. These three cards are the main contenders for my next upgrade, which I'm keen to acquire sooner rather than later.
 
not entirely sure what is going on with the 2070S FE boost clocks, its going up to 1935mhz/1950mhz without me touching anything. Not complaining but I will need to re-adjust the Navi clocks to get a more like-for-like bench.
 
not entirely sure what is going on with the 2070S FE boost clocks, its going up to 1935mhz/1950mhz without me touching anything. Not complaining but I will need to re-adjust the Navi clocks to get a more like-for-like bench.

That's just Nvidia's boost isn't it?

Surely you just run both GPU's maxed out and see what their performance difference is. Who cares if one GPU can attain higher clocks that the other etc.
 
That's just Nvidia's boost isn't it?

Surely you just run both GPU's maxed out and see what their performance difference is. Who cares if one GPU can attain higher clocks that the other etc.

yeah i will do some more runs over the next week, it’s fairly clear that the 5700XT does
not boost as hard using defaults as the 2070S FE and this is hurting its performance
 
What other games are you going to bench?

waiting for the EK Special Edition block to arrive on Thursday/Friday.

will take requests, was going to do F1 2019 next.

could have done a few this weekend but needed to swap my 8GB ram kit over to 16GB.
 
waiting for the EK Special Edition block to arrive on Thursday/Friday.

will take requests, was going to do F1 2019 next.

could have done a few this weekend but needed to swap my 8GB ram kit over to 16GB.

Nice one, I will look forward to them. Good work!
 
I might get my EK block Saturday. For now I will continue with some amendments.

I unless the bench is real tight I will stick to using stock 1080ti HOF measurements there was only a couple of fps in it either way.

I have also had to swap out my 8GB Ram kit for a 16GB kit as it was causing some lag in very limited situations. It is roughly the same speed and is not expect to effect the results as I am running 1440p and higher.

Will proceed with World War Z, Watchdogs 2, Monster World Hunter and Division 2.
 
My experience of a 2070S at 2070Mhz and a 1080Ti at 2025 is that the 2070s takes the podium, no exact benchmarks this is just from reference to FPS when I've been playing and also smoothness.
 
I might get my EK block Saturday. For now I will continue with some amendments.

I unless the bench is real tight I will stick to using stock 1080ti HOF measurements there was only a couple of fps in it either way.

I have also had to swap out my 8GB Ram kit for a 16GB kit as it was causing some lag in very limited situations. It is roughly the same speed and is not expect to effect the results as I am running 1440p and higher.

Will proceed with World War Z, Watchdogs 2, Monster World Hunter and Division 2.

Would love to see benchmarks to show how these modern cards hold up in Rise of the Tomb Raider and of course Shadow of the Tomb Raider.
 
Oh on the subject or SOTTR, at 1080p settings as per SOTTR Bench thread I got 88fps, same as an overclocked 1080ti.
 
Have updated with results. Hit a driver snag (the first and only so far) with World War Z on the 5700XT, in Vulkan the card seems to be held at 89% GPU usage. I will see if its a config issue tomorrow.

In terms of results so far, I think the 5700XT cooler whilst serviceable is really holding the card back against the 2070S FE. On the flip side the FE is so good out of the box, I reckon Nvidias partners are pretty angry at them for producing a card which is so well made and has left nothing on the table for their partners to extract.

I am fairly confident that once I get the EK block onto the 5700XT it may eclipse the 2070S FE. Placing a block onto the 2070S FE will not yield as big an improvement as i think the chip is near its limits out of the box.

The 1080ti is going about its business in fine style but need to remember the card, the block and the vrms were pretty much state-of-the-art at the time (3x8pin) and it draws an awful lot of juice.
 
Have updated with results. Hit a driver snag (the first and only so far) with World War Z on the 5700XT, in Vulkan the card seems to be held at 89% GPU usage. I will see if its a config issue tomorrow.

In terms of results so far, I think the 5700XT cooler whilst serviceable is really holding the card back against the 2070S FE. On the flip side the FE is so good out of the box, I reckon Nvidias partners are pretty angry at them for producing a card which is so well made and has left nothing on the table for their partners to extract.

I am fairly confident that once I get the EK block onto the 5700XT it may eclipse the 2070S FE. Placing a block onto the 2070S FE will not yield as big an improvement as i think the chip is near its limits out of the box.

The 1080ti is going about its business in fine style but need to remember the card, the block and the vrms were pretty much state-of-the-art at the time (3x8pin) and it draws an awful lot of juice.

Doesn't the 2070s have about 150-200mhz OC headroom? Please do correct me if i'm wrong. I'm in the market for a mid-range GPU myself and so the 5700xt seems very alluring for its price/perf.; yet I hear about the drivers/compatibility issues and who knows how much more AIB cards will cost. The inherent 10% lead on the 2070s, stable drivers, g-sync and nvidia inspector (e.g lod bias editing...) begins to justify its price bump (for me).
 
Doesn't the 2070s have about 150-200mhz OC headroom? Please do correct me if i'm wrong. I'm in the market for a mid-range GPU myself and so the 5700xt seems very alluring for its price/perf.; yet I hear about the drivers/compatibility issues and who knows how much more AIB cards will cost. The inherent 10% lead on the 2070s, stable drivers, g-sync and nvidia inspector (e.g lod bias editing...) begins to justify its price bump (for me).

The overclock headroom doesn't mean the cards can do it. Ref model has boost clock to 1770, and regardless if AIB or not, the cards do not crash if you set the clock to 2050.
However, all 2070s settle at 1890-1950 real clock due to thermal throttling. And when was the last time you looked at 2070S AIB prices? Only the KFA is at sub £510. The good ones are >£550
Almost £200 more than the 5700XT that's not a small price bump.... That's >40% up.
 
Back
Top Bottom