The NVIDIA GeForce RTX 2060 6GB Founders Edition Review: Not Quite Mainstream
by Nate Oh on January 7, 2019 9:00 AM ESTCompute & Synthetics
Shifting gears, we'll look at the compute and synthetic aspects of the RTX 2060 (6GB). As a cut-down configuration of the TU106 GPU found in the RTX 2070, we should expect a similar progression in results.
Starting off with GEMM tests, the RTX 2070's tensor cores are pulled into action with half-precision matrix multiplication, though using binaries originally compiled for Volta. Because Turing is backwards compatible and in the same compute capability family as Volta (sm_75 compared to Volta's sm_70), the benchmark continues to work out-of-the-box, though without any particular Turing optimizations.
For Turing-based GeForce, FP32 accumulation on tensors is capped at half-speed, thus resulting in the observed halved performance. Aside from product segmentation, that higher-precision mode is primarily for deep learning training purposes, something that GeForce cards wouldn't be doing in games or consumer tasks.
Moving on, we have CompuBench 2.0, the latest iteration of Kishonti's GPU compute benchmark suite offers a wide array of different practical compute workloads, and we’ve decided to focus on level set segmentation, optical flow modeling, and N-Body physics simulations.
Moving on, we'll also look at single precision floating point performance with FAHBench, the official Folding @ Home benchmark. Folding @ Home is the popular Stanford-backed research and distributed computing initiative that has work distributed to millions of volunteer computers over the internet, each of which is responsible for a tiny slice of a protein folding simulation. FAHBench can test both single precision and double precision floating point performance, with single precision being the most useful metric for most consumer cards due to their low double precision performance.
Next is Geekbench 4's GPU compute suite. A multi-faceted test suite, Geekbench 4 runs seven different GPU sub-tests, ranging from face detection to FFTs, and then averages out their scores via their geometric mean. As a result Geekbench 4 isn't testing any one workload, but rather is an average of many different basic workloads.
We'll also take a quick look at tessellation performance.
Finally, for looking at texel and pixel fillrate, we have the Beyond3D Test Suite. This test offers a slew of additional tests – many of which use behind the scenes or in our earlier architectural analysis – but for now we’ll stick to simple pixel and texel fillrates.
Facebook
Twitter
RSS
134 Comments
View All Comments
zepi - Monday, January 7, 2019 - link
Watching with disgust due to high power usage and noise. I won't upgrade to something that sounds as bad as my current AMD card.eva02langley - Monday, January 7, 2019 - link
HBM2 and GDDR6 are fairly at the same price even without the actual numbers. GDDR6 is 70% more expensive than the GDDR5.But yeah, Koduri mistake was to push HBM2 on Vega, it should have been GDDR5 or GDDR5x.
CiccioB - Monday, January 7, 2019 - link
This legend that HBM costs like the other kind of memory is in place since people stated that HBM cost like GDDR5 (then it was against the lowly available GDDR5Xm produced only by Micron) and that using it against 12 chip of GDDR5/X/6 and the complexity needed for the PCB layout to handle them was almost the same.Unfortunatelty nothing of this is true.
HBM costs more per chip (and by GB) by itself for its construction that requires a high end process for stacking up all those layers.
Moreover, it requires a big silicon interposer that is expensive enough to cover the cost of any GDDRn memory type based PCB.
Third it requires a different path for mounting and aligning it on the interposer that is also an expensive procedure (see the problems AMD encountered for it) and that can't be done in the AIB fabs where GDDRn chips are usually mounted and soldered for 0,01$ each chip.
What AMD fanboys constantly states is their hope that AMD is not going to loose so much money for any Vega that they are selling. Unfortunately, they are, and this new video card by nvidia will make even more hard pressure on Vega as we already have announcements of further price cuts on such an expensive piece of crap that can't compete with anything in any market it has been presented and has required the constant price cut even before it was launched. In fact Vega FE cards started discounted by $200 at day one with respect to the former announced MSRP price... what a marvelous debut!
Ratman6161 - Monday, January 7, 2019 - link
Totally beaten in raw performance, yes. But I don't want it anywhere near badly enough to pay $349 for it. What you can actually buy for what I would be willing to spend is the Rx580 or GTX1060 at under $200.mapesdhs - Monday, January 28, 2019 - link
zepi, if all one wants is normal 1080p gaming then an RX 580 is a much better buy, especially used. Mine only cost 138 UKP. The real joke here is the price hiking of the entire midraange segment. The 2060 is what should have been the 2050. People are paying almost double for no real speed upgrade compared to two years ago at the next teir up (which should have been what the 2070 is now). Tech sites know this, some talked about it early on, but now they've all caved in to the new 2x higher pricing schema, the only exception being Paul at "not an apple fan" who continues to point out this nonense. If people go ahead and buy these products then the prices will keep going up. And AMD will follow suit, they'd be mad not to from a business standpoint.Opencg - Monday, January 7, 2019 - link
I would not expect the 2060 to be anywhere near msrp considering its the only turing card with a reasonable msrp/performance ratio the demand will be high. And we all know what happens when demand is high.Bluescreendeath - Monday, January 7, 2019 - link
@Benedict, Did you even read the article? The GTX2060 is more than 50% faster than the RX580 and the GTX1060. Furthermore, the GTX1060 6GBs cost around the same as the RX580 - the 1060 is not "much more expensive."mapesdhs - Monday, January 28, 2019 - link
Feel free to spend 100% more than what x60 cards used to cost, for a performance level that should be the tier below, but don't complain when the prices get hiked again because consumers keep buying this ripoff junk.kpb321 - Monday, January 7, 2019 - link
If you want the best price/performance you need to go a little bit lower than the 580. The 570's have been ~$130 AR pretty regularly with a couple dips below that. Personally, I picked up an 8gb 570 for $190 with three stacked rebates/GCs for a total of $90 off bringing the cost of the card down to $100. I also sold off my old video card and one of the games from the bundle for another $50 bringing my upgrade cost down to ~$50. I had been wanting to upgrade for a while and was hoping for a 580 or a 1060 3gb or 6gb but the 570 looked like such a good deal I couldn't resist. Yes it was quite a bit of rebates but at this point I've gotten all of them so that is my final AR cost. Granted this even further down the performance curve but a 8gb 570 is certainly going to be a lot better than %50 of the performance of an 8gb 580 but that's what I paid for mine.zaza - Monday, January 7, 2019 - link
the rx 570 can be found for 140$-150$ now and comes with your choice of 2 games out of three (unreleased games) (devil may cry, divison 2 and Resident evil 2 remake). for that price i think fir 570 is best GPU for the price, great for 1080p gaming.