EVGA RTX 2060 KO review: is the entry-level RTX fast enough for ray traced gaming?
It’s rare that we review graphics card variants at Digital Foundry but in the case of the RTX 2060 KO from EVGA, we’re going to make an exception. Nvidia’s entry-level, feature-complete Turing card sat at a $349 price-point for quite some time – a touch pricey perhaps when the significantly superior RTX 2060 Super turned up costing just $50 more. However, with this new EVGA KO model clocking in at $299, it’s time to re-assess the product and specifically, just how capable the card is in handling games using hardware-accelerated ray tracing.
Primarily, it’s the RTX aspect of the 2060 that sets it apart from its nearest AMD competitor, the RX 5600 XT. Its ability to tap into the DXR API and by extension access the full range of visual options available in supported games is obviously a nice feature to have – and with ray tracing confirmed for the next-gen consoles, the broader adoption of RT is a case of not if but when. On top of that, the inclusion of Turing’s tensor cores allows for the 2060 to access hardware-accelerated machine learning features with the AI-powered DLSS upscaling pretty much the only application for this technology in the here and now. This is all in addition to standard graphics power that’s generally in excess of the RX 5600 XT – to the point that AMD had to deliver an 11th hour BIOS upgrade to bring its latest Navi release back into contention.
However, while generally well -received for its price vs performance level (especially with the recent haircut on pricing) there’s a lot of discussion surrounding the RTX 2060’s prowess in terms of delivering accomplished ray tracing support – and perhaps rightly so, when looking at the first wave of games with DXR functionality. In the past, I’ve managed to get an effectively locked 1080p60 performance level with the RTX 2060 when playing Battlefield 5 with ray tracing enabled – but it required some options tweaking and an overclock. On top of that, there’s perhaps a broader question to answer: is 1080p60 actually good enough to begin with bearing in mind that the RTX 2060 ordinarily performs rather well at 1440p?
These are all pertinent questions but perhaps just as important is the technical make-up of the KO version of the RTX 2060 itself. Clearly, corners are cut to deliver the more aggressive pricing. Nvidia’s reference model – the Founders Edition – is a more deluxe product with superior build quality, improved power delivery and higher quality materials. While possessed of a decent metal back plate, the KO’s shroud is plastic, the cooler is less substantial and its four-phase power delivery does the job but limits overclocking potential. Ultimately though, the main difference you’ll actually notice during gaming is that it’s somewhat louder than the Founders. That’s the only real grumble I can muster against it.
However, as initially discovered by Steve Burke at GamersNexus, the 2060 KO is somewhat fascinating in that it uses a TU104 Nvidia processor – the same processor that powers the RTX 2070 Super and RTX 2080. It’s a salvage part, with CUDA cores disabled to match the 1920 complement in the Founders Edition’s standard TU106 processor. In my tests, I found performance to be entirely identical, with one exception. As GamersNexus discovered, the CUDA path in the Blender rendering tool delivers much faster performance than existing RTX 2060 cards. The extent of the boost varies on the complexity of the workload, but one example demo I tested delivered a 19 per cent reduction in render time.
