Needed: Framerates of different graphic cards

[Wagon153]

Heyo, I'm the guy doing all them hardware threads. So time for another one.

Recently, I've been coming across some rumors that AMD cards get less FPS then their equivalent Nvidia cards, and the rumor does have some merit. So I need average framerates from users of cards from both brands. In the interest of keeping the bench as neutral as possible, I need the following.

Graphics settings:

Graphics: Fancy
Clouds: On
Smooth Lighting: Maximum
Mipmap Levels: 4
Anisoptric Filtering: 16
Vsync: Off
Particles: All
Render distance: 8
Brightness: Bright
Max Framerate: Unlimited
If you have Freesync or Gsync, please turn that off too, for obvious reasons.
Minecraft version: Vanilla 1.7.10
World type: Flat with redstone ready preset.

Specs I want:

Graphics card model(subvendor optional.)
CPU model(i3-4150, FX-6300, etc etc.)
OS(If Linux, please provide exact distribution and version. If Windows, just say number, such as Windows 10 or 8.)
Resolution you are running Minecraft in(720p, 900p, 1080p, 1440p, 4k, etc etc.)

Hardware requests:
Please run your CPU at stock clocks during this test if you have an unlocked multiplier/BCLCK. Results are already gonna be a bit challenging to sort through, don't need overclocks/underclocks making things even more confusing.
Please run your GPU at reference clocks if at all possible. You can alter clock and mem clock in MSI afterburner. Many GPU subvendors(Sapphire, EVGA, MSI, etc etc.) apply stock overclocks to their cards, thus why I ask.
Please try to use the latest drivers available for your card. If you can't, I understand.

Other requests:

NO OPTIFINE. Sorry guys, but optifine will just make the results murky.
No shaders, texture packs, or mods that affect graphical fidelity and/or FPS in any way.
Please run the test in Fullscreen.
Have MC running for at least 5 mins.

Thanks guys ily.

 

[Pyure]

You lose me the instant you say "equivalent" cards. Its hard to apples-to-apples on this one.

Also the test is a bit useless unless the players all open the same world and look in the same direction from the same place.

(disclaimer: you're brilliant and I'm not shitting on you)

 

[Wagon153]

Ah yes, let me do some edits.

An example of equivalent cards is GTX 960->r9 280/380 or r7 260x->GTX 750 TI. Cards that are on the same performance tier.

 

[SevenMass]

To elaborate what Pyure already told you; If you look straight up you'll get a different FPS than if you look straight down. This difference can be as much as a factor 2. This behavior exist even in void worlds.
I don't know if there is such a difference between looking east/west/north/south, but if there is, it is much smaller than up or down.

What this means is: You'll need to give any potential participants some extra instructions in order to get usable results.

 

[ChemE]

Would it not be best to specify vanilla MC, a seed, an exact position, and an exact eye direction? My understanding is some mods add blocks which render even though they are not visible. Surely all that will wash out the differences between graphics cards' performance unless it is accounted for.

EDIT: Or do what middleclaw and I did and stand in the same spot looking in the same direction on the same server but with different hardware.

 

[DZCreeper]

Processor: i7 4790K @ stock.
Graphics Card: GTX 980 Ti @ stock.
Resolution: 2560x1440
OS: Windows 7
Allocated RAM: 2GB
Java Version: 1.8.0_60 64 bit

Conclusion: "How many FPS do you get DZ?" "All of them."

Test result on suggested settings: 1400 frames per second. Variance of 30 caused by opening and closing of F3 screen. Visual confirmation: http://imgur.com/0wdR10v was taken using in-game F2 button.

If you would like some comparisons in different scenarios, let me know and I will update this post.

Edit:

Duplicate test environment with Forge 10.13.4.1492 resulted in 1310 frames per second. http://imgur.com/CYttqb9
Duplicate test environment with Forge + FastCraft 1.22-ctest14. Result is 1550 frames per second. http://imgur.com/bBGdfMf
P.S. Don't actually run Minecraft at this FPS. Very high FPS is known to cause coil whine, which is non-damaging but annoying.

 

[Wagon153]

Would it not be best to specify vanilla MC, a seed, an exact position, and an exact eye direction? My understanding is some mods add blocks which render even though they are not visible. Surely all that will wash out the differences between graphics cards' performance unless it is accounted for.

EDIT: Or do what middleclaw and I did and stand in the same spot looking in the same direction on the same server but with different hardware.

Yeah I edited the post to specify vanilla Minecraft 1.7.10, using a Flat world with the redstone ready preset. That should also eliminate the need to specify a seed.

Processor: i7 4790K @ stock.

Graphics Card: GTX 980 Ti @ stock.

Resolution: 2560x1440

OS: Windows 7

Allocated RAM: 2GB

Java Version: 1.8.0_60 64 bit

Conclusion: "How many FPS do you get DZ?" "All of them."

Test result on suggested settings: 1400 frames per second. Variance of 30 caused by opening and closing of F3 screen. Visual confirmation: http://imgur.com/0wdR10v was taken using in-game F2 button.

If you would like some comparisons in different scenarios, let me know and I will update this post.

Edit:

Duplicate test environment with Forge 10.13.4.1492 resulted in 1310 frames per second. http://imgur.com/CYttqb9

Duplicate test environment with Forge + FastCraft 1.22-ctest14. Result is 1550 frames per second. http://imgur.com/bBGdfMf

P.S. Don't actually run Minecraft at this FPS. Very high FPS is known to cause coil whine, which is non-damaging but annoying.

Thanks for the data!

And yeah, I know about coil whine. My r9 280 has it, and it is especially noticeable in menus.

 

[SevenMass]

ATI Radeon HD 5770 1024 MB
Intel Xeon 5400 series 3,2 GHz
MacOSX 10.9.5
Resolution: 1920 x 1200
allocated ram: 1 GB (6GB installed)
Java Version: 1.8.0_60 64 bit

FPS:
Looking at the horizon: ~285
Looking straight up: ~358
Looking Straight down: ~325

I normally play with render distance 16 or 12, and in windowed mode, and with a lot of mods. My FPS is about 50 in this case, and it can jump to 90 when I look up.

 

[ShneekeyTheLost]

This also doesn't take into consideration the computer's other stats, such as RAM and CPU. It also doesn't take into consideration any kind of custom garbage handling lines you use, which can SIGNIFICANTLY change your FPS, particularly after five minutes. You also forgot to ask what version of Java, which is also going to have an impact.

Benchmarking on something like this is just not going to be possible to be done crowdsourced. There's just too damn many variables. Even things like 'my antivirus was running in the background' can screw with your FPS.

 

[Wagon153]

This also doesn't take into consideration the computer's other stats, such as RAM and CPU. It also doesn't take into consideration any kind of custom garbage handling lines you use, which can SIGNIFICANTLY change your FPS, particularly after five minutes. You also forgot to ask what version of Java, which is also going to have an impact.

Benchmarking on something like this is just not going to be possible to be done crowdsourced. There's just too damn many variables. Even things like 'my antivirus was running in the background' can screw with your FPS.

Couldn't think of any other methods to test this, since the only graphics cards I own are AMD. And a lot of my friends, well, I'm lucky they know how to turn a computer on. RAM I don't think is too relevant here, since I asked for the test to be run in vanilla(which runs just fine even with 1 gig of dedicated RAM to it, especially since I specified a render distance of 8.)


Also, I did ask for the CPU model/number(even requested them to be ran at stock clocks.) When/if I get enough results, I plan on putting everything into a spreadsheet in Google Docs for everyone to look at at their leisure.

That said, for the very reasons you listed, I wasn't planning on making a definite conclusion off of these results. If there seems to be a definite difference between the two GPU vendors, I'll try to get a hold of a Nvidia card and run the tests all on my own hardware.

 

[DZCreeper]

This also doesn't take into consideration the computer's other stats, such as RAM and CPU. It also doesn't take into consideration any kind of custom garbage handling lines you use, which can SIGNIFICANTLY change your FPS, particularly after five minutes. You also forgot to ask what version of Java, which is also going to have an impact.

Benchmarking on something like this is just not going to be possible to be done crowdsourced. There's just too damn many variables. Even things like 'my antivirus was running in the background' can screw with your FPS.

For my testing I used no custom garbage collection, made sure my anti-virus wasn't running, and closed all my background apps. I even made sure to take the screenshot when the memory usage was between garbage collection runs to avoid that impact. It is possible to crowd-source this data but a standardized test is needed.

 

[Softpotatis]

Processor: i5 750@ stock

Graphics Card: R9 290 Tri-X OC @factory OC (sorry didn't want to mess with it)

Resolution: 1920x1080

OS: Windows 10

Allocated RAM: 1GB

Java: 1.8.0_45

FPS: About 325.

Spoiler

I've bought all the parts for my upgrade, which will be an i5 6600k, Z170 Gigabyte board, DDR4 and SSD and i'll do the test later on when I upgrade.

 

[Shevron]

Yeah I edited the post to specify vanilla Minecraft 1.7.10, using a Flat world with the redstone ready preset. That should also eliminate the need to specify a seed.

Thanks for the data!

And yeah, I know about coil whine. My r9 280 has it, and it is especially noticeable in menus.

Interestingly, I had coil whine on my Minecraft menus on both my old GTX285, and on my new GTX970.

The common denominator was my Coolermaster 700W PSU. When I replaced it, all coil whine disappeared and everything is as silent as a tomb now.

Turns out it was the PSU doing it on certain frame rates, not my GFX card.

 

[gold49]

While I understand this is targeted toward Nvidia vs AMD discrete gpus performance, RAM frequency would be an important factor in performance for people running integrated graphics so it could be needed to state the speed in that case (I agree it isn't that important for anyone running a discrete gpu however).

Also, the minecraft wiki has some benchmarks here. Although I take them with a grain of salt myself, they could have some use when comparing data or something.

 

[ScottulusMaximus]

Yeah this is just going to turn into an AMD vs Nvidia flame war with 0 usable data I'm afraid... Unless you use the same system and just swap out GFX cards you won't get any equivalent results.

 

[Softpotatis]

Processor: i5 6600k@ stock

Graphics Card: R9 290 Tri-X OC @factory OC (sorry didn't want to mess with it)

Resolution: 1920x1080

OS: Windows 10

Allocated RAM: 1GB

Java: 1.8.0_60

FPS: About 568.

Spoiler

Tbh I was expecting a bigger improvement in FPS, since i5 6600k has such a high IPC.

EDIT: Update, 64bit Java didn't change a thing.