![gtx 1070 very poor opencl benchmark score gtx 1070 very poor opencl benchmark score](https://static.techspot.com/images/products/2016/graphics-cards/org/2017-09-12-product-2.jpg)
Either get one RX 480 or get something more powerful. We're in the same place as we have been for multi-GPU configurations for years: Don't do it. See of this article for more info on all of that. If you had to operate with cards closer than ours were – we had a two-slot spacing – then you'd be looking at a lot more clock/fan fluctuation to account for the thermals. Noise levels increase a little bit when looking at raw dB values, but the “whine” and turbulence are more noticeable with CrossFire cards. In fact, the GTX 1080 showed some of the worst swings we've ever seen when overclocked with the stock cooler. This is something that happens with both nVidia and AMD as the clock fluctuates with thermals, so we're not just knocking AMD here. Games which are more frequency dependent will exhibit greater variability of 0.1% and 1% low values and show some stuttering. Frequency drops are regular, but do not always show consequence in all games. We see clock-rate swings greater than 100MHz during RX 480 CrossFire benchmarking, a bit more exaggerated than with a single card as a result of more frequent topping out on the ~80C cap.
#Gtx 1070 very poor opencl benchmark score software#
Thermals are maintained at about the same level as the single card, resultant of AMD's “auto” setting in Radeon Software adjusting the fan speed and clock-rate dynamically. Shadow of Mordor posts decent scaling (78.3FPS AVG to 127.7FPS AVG, or 63%), but also suffers from poor 0.1% low increases over a single card. We see OK scaling in Metro: Last Light, but nothing exciting. The cards don't “beat” the GTX 1080 in any single metric, but for the price, coming close is still respectable. Ashes sees the CrossFire RX 480s nearly achieving parity with the GTX 1080 when in Dx12 (though they post dismal performance with Dx11). We also see negative scaling in The Division – upwards of 34% in the 0.1% lows, moving the game from reasonably playable to unplayably spikey. Mirror's Edge Catalyst, what with its heavy reliance on post processing effects, sees significant negative scaling with CrossFire (upwards of 31.4% performance degradation, in some cases). You'd want to disable CrossFire in such cases, as having the second card actually diminishes the framerate performance. Some games post nearly 2x scalability and manage to maintain reasonable 1% low and 0.1% low frame metrics – Black Ops III is a good example – but about half of our test suite shows negative or near-0 scaling. CrossFire, just like SLI in our recent hacked-together 1070 SLI benchmark, is entirely hit or miss. This is the same conclusion we seem to reach with almost all of our multi-GPU benchmarks. Conclusion – Is CrossFire Worth It for the RX 480?