Hardware Nvidia GeForce GTX 1060 Founders Edition Review

So I've noticed the my graphics cards outperform the test cards. Not by a ton but enough to call into question what I am doing wrong here. Was the same story with my R9 280 and now my GTX 970. Here are my Valley scores.



What am I missing here? I am triple checking to make sure I am using the same settings as the article. I am fairly sure my 970 isn't in par with a 980.

My settings:

 

What CPU frequency are you running?

Is it the same as in the bench rig that they're using? They have a 5960X at 4.2 GHz.

Furthermore, what clocks are your using on your card?

 

4Ghz stock 4790k (4.4ghz boost)
Gtx 970 (asus strix version)
16gb ddr3-1800 (4x 4gb 10-10-10-28)
using cryorig H7 for CPU cooler
Supernova G2 650W

what else you need to know?

Is the 5960X that much slower than a 4790k to bottleneck these cards in this manner or do I have some setting wrong in the benchmark that causes my scores to be artificially higher?

 

Personally I don't use SLI and have no intention of doing so but removing it completely does seem like as blatant a money grab as Intel locking all of its non-K CPUs & forcing you to spend more than you would want. Otherwise looks like a decent buy and price wise on a par with the 970 series for slightly improved performance. Wouldn't be surprised if it becomes the go to card for the mid range end of the market

 

Probably.
Your 4790k: 4GHz base 4.4GHz boost.
Stock 5960X: 3GHz base 3.5GHz boost.
For the same architecture, that's quite a gap. Few games would take advantage of the extra 4 cores (very few even take advantage of 4 cores as it is) so single-threaded performance is the important metric.

 

As I have said in other threads the 970 1060 and 480 trade blows across a range of games. A second hand 970 is still the Best Buy this year for 1080p gaming and 1440p gaming with reduced settings.

All reviews are usually done on the fastest CPU going so give a very best case scenario. The more likely scenario is the CPU performance of what anandtech recently tested the Athlon x4 845. Or older sandybridge chips.

As I Dout anyone buying a £200 gpu is spending a £1000 on a CPU.

 

1500 points difference is pretty large though. They have the CPU at 4.2ghz so there would be little if any difference single threaded.

 

Is a fresh install of Windows used between every GPU change? It makes a difference.

 

This is going into my media PC to replace my MSI GTX760, my MSI GTX970 will not fit in the MATX case as it's too tall above the PCI bracket...

 

Greetings!

Really should add Doom.
Regarding DX2 3DMark:
- http://www.overclock.net/t/1606224/...parallelism-than-doom-aots-also#post_25358335
- http://www.futuremark.com/pressreleases/a-closer-look-at-asynchronous-compute-in-3dmark-time-spy

 

I get a higher score with my Titan X and 5820k at stock speeds than I do with them both overclocked.

Junk benchmark is junk.

 

Why is a dx12 benchmark junk? It's probably the most relivent benchmark according to you in other threads.

Or is because this specific benchmark does not show good for AMD that it's junk?

 

Whatever do you mean? Are you suggesting Andy has a bias toward AMD? Surely not!

 

I am using Windows 10 Home x64.

If you do that though, what benchmarks will be left for us to do ourselves for comparison?

 

Interesting... from what I recall, the Unigine Valley benchmark favored the RX 480 in Linux. Nvidia's Linux performance is usually pretty close to their Windows performance. I didn't compare the numbers between OSes, but I wonder if that means there's a huge amount of potential in the 480 that isn't unleashed yet, or, if Nvidia's Linux driver are suffering in some way. I'm betting on the latter.

 

Outside of 3D Mark and unigine valley there's very few that we can do at home for games unless you own the specific game.

 

Unless you use the exact same hardware, with the exact same drivers and versions of the benchmarks themselves... comparisons between BT's results and your own home results won't give a true value anyway. As long as BT continue to use the same variables for specific benchmarks and scores then that is what should be used as a general reference and nothing more.

 

Wow. So now an x60 card beats out an (x-1)80 card? Either I've not been paying attention or that's quite an impressive new measure. How would this compare to 2x 670s in SLi? I'm sure it's better, but any rough idea by how much and is it worth upgrading for 1920x1200 gaming?

Also, given the removal of SLi (cheap trick) from this end of the market, can you Crossfire two RX480s?