Futuremark launches world’s first 4K ‘Ultra HD’ benchmark

[monstercameron]

It's not magic and there's nothing to debunk. Its because the memory bandwidth on AMD gpus is much larger. Nvidia uses an algorithm to recycle old textures to compensate for their smaller bandwidth, which at resolutions like 1080p works quite nicely, the problem is that the algorithm doesn't scale up to higher resolutions very well. The data still needs to be crammed through the same amount of space, and the amount of throughput required for 4K is too much for 'texture recycling' to handle.

I mean the amd cpus based on this thread http://linustechtips.com/main/topic/229893-amd-fx-8350-faster-than-i7-4930k-for-uhd/

[Me1z]

I mean the amd cpus based on this thread http://linustechtips.com/main/topic/229893-amd-fx-8350-faster-than-i7-4930k-for-uhd/

 

Oh, that one. My guess is they cherry-picked the test results in a way that favored AMD. I've seen up to 600 point differences between test runs on the same machine in 3Dmark, so I assume all GPU bound benchmarks are similar. There is a very long explanation for these cpu performance fluctuations that has to do with probability, the order that cores happen to receive instructions, and measurement accuracy but I'll save you the wall of text. Simple explanation: It's well within margin of error.

[deathjester]

maybe this will help debunk the magical amd lead at 4k?

 

On GPU? Prepare to be disappointed. A R9 290 beats a GTX 980 with 4k supersampling/downsampling in Ryse, and beats it in Shadow of Mordor with 4 k downsampling as well, and that is an Nvidia title. 

 

 

http://www.pcgameshardware.de/Ryse-PC-259308/Specials/Test-Technik-1138543/

 

Click 1.5x1.5 and 2x2 Supersampling on the benchmarks.  That is literally all that this 3DMark bench does on the two settings (1440p/4k) that aren't available in the free version. Memory bandwidth matters quite a bit at 4k and the AMD GPU's really shine at high resolutions. I can confirm this. I just came from a GTX 770 that SUCKED at downsampling/supersampling compared to my r9 290 that is around the same price atm. 

[monstercameron]

On GPU? Prepare to be disappointed. A R9 290 beats a GTX 980 with 4k supersampling/downsampling in Ryse, and beats it in Shadow of Mordor with 4 k downsampling as well, and that is an Nvidia title. 

 

 

http://www.pcgameshardware.de/Ryse-PC-259308/Specials/Test-Technik-1138543/

 

Click 1.5x1.5 and 2x2 Supersampling on the benchmarks.  That is literally all that this 3DMark bench does on the two settings (1440p/4k) that aren't available in the free version. Memory bandwidth matters quite a bit at 4k and the AMD GPU's really shine at high resolutions. I can confirm this. I just came from a GTX 770 that SUCKED at downsampling/supersampling compared to my r9 290 that is around the same price atm. 

follow the most recent comments.

[Osiris199]

Do you need a 4K Monitor (or in my case a 4 monitor eyefinity setup) or can you downsample it to 1080p?

[Cupitor]

I'm sure my voodoo fx thingy GPU will be making world records here....

[vm'N]

Do you need a 4K Monitor (or in my case a 4 monitor eyefinity setup) or can you downsample it to 1080p?

If you try to play a 4K video on a 1080p monitor, it will be downscaled.

[7850OC]

Do you need a 4K Monitor (or in my case a 4 monitor eyefinity setup) or can you downsample it to 1080p?

 

Fire Strike Ultra renders the test content at 3840 x 2160 (4K Ultra HD) before scaling the output to your PC's display resolution.

 

This means you don't need a 4K monitor to run Fire Strike Ultra, though you will need a GPU with at least 3 GB of dedicated memory

 

(this is what they say) 

[Sheldon_King]

Run the benchmark anyway just to see what happens. I'm curious to see how fermi handles 4k but I'm to lazy to swap in my GTX 580.

It might take a few hours to get the system out and replace cards, but I am willing to test my GTX480 on it for you, that is if you want me to.

 

As for AMD, here is a score for 2 7950s http://www.3dmark.com/3dm/4363660?

[vm'N]

It might take a few hours to get the system out and replace cards, but I am willing to test my GTX480 on it for you, that is if you want me to.

You want to burn down your house?

Don't let your 480 be alone. Couple it up with a 290x (reference design) and you should be fine.

[Sheldon_King]

You want to burn down your house?

Don't let your 480 be alone. Couple it up with a 290x (reference design) and you should be fine.

You must seriously not understand how hot it has to be to start a fire. Just a hint, plaster needs 780*C to flame iirc, and that is some 7+ times the maximum operating temp of Fermi GPUs (the driver will shut off the system completely at 110*C).  

 

Second, It would be retarded to leave AMD GPUs in with the Fermi. The AMD would just be sitting there being bored while the 480 is rendering the benchmark. Wasted energy is what the AMD would be (not that they are not wasted energy when they are rendering the scene, but that is a discussion for another thread). 

[vm'N]

You must seriously not understand how hot it has to be to start a fire. Just a hint, plaster needs 780*C to flame iirc, and that is some 7+ times the maximum operating temp of Fermi GPUs (the driver will shut off the system completely at 110*C).  

 

Second, It would be retarded to leave AMD GPUs in with the Fermi. The AMD would just be sitting there being bored while the 480 is rendering the benchmark. Wasted energy is what the AMD would be (not that they are not wasted energy when they are rendering the scene, but that is a discussion for another thread).

I'm sorry, I will never use sarcasm again without the sarcasm endtag.

[Osiris199]

My score 4354. Average of 25 fps For some reason the driver is not approved (yet). I'm using version 14.9.1

[Khvarrioiren]

My score 4354. Average of 25 fps For some reason the driver is not approved (yet). I'm using version 14.9.1

Beta drivers. Scored 2246 on my rig. Not even nearly there