2060 or wait for AMD?

Page 2 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

t99

Honorable
Jul 16, 2014
756
1
11,215
I wanted to wait for AMD, but with the 590 priced only a little less than a 2060 and performing significantly worse, the 2060 is a much better choice. What are the odds AMD will have something at 400 or less that beats out a 2060 by 20% or more? If this isn't coming then no point to wait further. The other option would be if they release a card at 300$ or less equal or better than a 2060.

I have a rx 480 now and the 2060 is impressive for the price. Some benchmarks on an OC'd 2060 are right there with a stock 2070. Think it's worth waiting further? I have until March until the new games are going to be out.
 
You do know the Vega should come with 3 games. I think division 2 is over of them. Since you mentioned that specific game.

I'm not seeing the bundle showing on the Newegg eBay listing, but I know I got an email from microcenter about it after I purchased the Vega.

Here's the link for the card on Newegg itself.

https://m.newegg.com/products/N82E16814131740

They do list the games. Maybe that's how they are selling cheaper if they aren't doing the game bundle on eBay listings?
 
Looking at what AMD did with the Vega56 /64. Were they faster than nVidia? NOPE. Were they cheaper than nVidia? Hell NO. They had higher MSRP at release compare to the 1070 and 1080. What did AMD annouce Radeon VII prices to be? Same as 2080? Without raytracing and less performance? So tell me what are you going to wait for AMD to do? Their prices won't be cheaper and it won't perform better.... So wait all you want, it'll be long wait before AMD's prices come down. Vega56/64 is finally coming to a more proper price... after how long?
 
I heard from YouTube where they said the Radeon 7 is so expensive because of it being HBM2 memory. It's a 16gb card, they are saying that is one reason they are expensive is because that vram costs so much. But the speculation is that whenever their new cards land, they will use gddr6 which should let them reduce costs.
 


crushing intel is not impossible the way they are right now but nvidia? to "crush" nvidia they need nvidia to make the same mistake as intel did and so far nvidia did none. to some the RTX might be underwhelming but in reality nvidia still consistently pushing 30% performance increase every year with or without new node which something AMD can't do. multi GPU in single package by sticking several small die together? don't be surprise if nvidia actually being the one that first coming out with such product. while AMD talking how it is possible to create GPU like that nvidia already have actual commercial device working upon that concept even if the GPU that being combined is still not on die level. do you remember when AMD said it is possible to combine all the available VRAM together instead each card needs it's own pool of VRAM with low level API back in 2015? there is not even a single demo how that can be done from AMD until now but such thing already a reality with nvidia SLI starting from Turing. i'm not saying AMD is bad or something like that but for AMD to beat nvidia it was a whole lot more difficult than to beat intel because nvidia is more aggressive in pushing the performance bar higher. also nvidia is not like intel when they look at AMD as a competitor. in intel view AMD probably will never really able to challenge them after bulldozer. not when AMD really struggle in their financial for years and literally have no R&D to put out something competitive on the market. but nvidia is different. yes they will said certain AMD product are underwhelming like recently announced Radeon 7 (they also said they expect more from AMD when 7970 was only around 15% faster than their GTX580 back in 2012) but they never disregard the potential of AMD future product that they have yet to see. this what makes nvidia seems unbeatable during maxwell and pascal era.
 


because not even competitor really willing to push the price lower. when AMD set the MSRP for RX590 at $275 a few weeks before 2060 launch that's like AMD telling nvidia to keep doing what they have been doing with those RTX price wise. and that actually is a big signal where AMD want the GPU market to go (when it comes to price structure) and yet there is still this crazy speculation about how AMD going to launch new GPU that is as fast as Vega 64 with the pricing of $250 at CES to kick nvidia butt. so AMD want to kill the recently launch RX590 that quickly? in the end Radeon 7 happen at CES. sometimes AMD already give the signal what they intend to do and what kind of product they will be coming out with but some people still choose to expect miracles from AMD. same with what happen to polaris back in 2016.
 


some people expect the 16GB HBM price to cost twice as the 8GB one but i think the 16GB could end up more than double the cost because of yield issue. 16GB have more stack (also resulting in higher total bandwidth when it comes to HBM) but more stack meaning the defect percentage will also increase. and this can quickly raise the cost. so the thinking 16GB HBM cost twice as much than the 8GB one might not be applicable here. Navi most likely use GDDR6 instead of HBM. and don't be surprise if AMD going to ditch HBM entirely from their gaming card going forward because cost wise HBM did not go the way AMD hope it to be due to manufacturing complexity.
 
And again, a lot of the rumors are speculation. One thing to keep in mind is that AMD is making the GPUs for the new xbox/ps5 so that will help them stay viable.

As much as I'm not a fan of Intel, them getting into the area of graphics cards might be good for everyone as well to give a 3rd choice.
 
And again, a lot of the rumors are speculation. One thing to keep in mind is that AMD is making the GPUs for the new xbox/ps5 so that will help them stay viable.

the console deal actually was sort of double edge sword to AMD. thanks to both xbox one and PS4 using AMD hardware that give a lot of advantage to AMD when their hardware become the center in which most modern game engine being develop for. and because of that they also able to push mantle so they can stir future 3D API direction to be more favorable towards their hardware. if you want to look at one mistake that nvidia did it was this: giving up all the console market to AMD. this what really hit kepler so hard down the road. it was lucky for nvidia that the effect is not really felt until they already completely move to maxwell. maxwell is not just a simple efficiency improvement from kepler but it mostly address kepler weakness from the new game engine that is build more for AMD hardware. and even with that AMD still have something under their sleeve that also hit maxwell quite hard later on: async compute. although it does not hit nvidia hard like it was with kepler because when it comes to DX12 it can be complicated even on AMD hardware resulting majority of game developer still put their focus on more stable and proven DX11.

but while it give certain advantage to AMD it also become AMD downfall. AMD strategy was to have both console and their pc graphic to have the same architecture. this is what will give them the direct advantage from having the hardware on console. but because of this AMD also cannot make drastic changes to their base architecture. too different and developer no longer able to re-use the optimization done on console on their PC GPU. probably why a few years back AMD said they expect next gen console life cycle will be shorten to 5 years instead 8 like how it was with 7th gen. because they probably aware there is no way they can stick with one architecture for a very long time and be competitive with nvidia when PC usually move to new architecture every two years. AMD probably expect PS5 and next gen Xbox with brand new architecture in 2018 time frame but instead console maker opt for mid cycle refresh while sticking with the same architecture. so in the end AMD still stuck with GCN until true next gen console comes out. during polaris vs pascal era it is really clear that AMD in dire needs of new architecture if they want to compete with nvidia.

As much as I'm not a fan of Intel, them getting into the area of graphics cards might be good for everyone as well to give a 3rd choice

they have the resource and the tech needed to be competitive. many think that there is no way intel able to compete AMD and nvidia because the lack of IP but that is not true at all. if they tackle the market with the right mindset this time even nvidia should be afraid with what intel can do with GPU.
 
nvidia have the interest before and their motherboard/chipset division (which later being absorb in by tegra division) was destroyed because of that ambition. in the beginning intel denied that the action they take against nvidia might have to do with nvidia interest in developing their own x86 CPU in secret (Jensen himself was ex AMD engineer) only that the cross licensing they did with nvidia back in 2004 did not cover what nvidia been doing in 2007/2008 time frame. but intel intention was completely clear in 2011 settlement. one of the agreement was nvidia will never allowed and completely bared from having the right to license X86.
 
Competition is always a good thing for the consumer. Even with AMD's competition, Intel had stagnated a bit. The release of the Ryzen processors finally gave Intel a kick in the ass. So they will have to compete again and release some better processors. And the same goes for Nvidia, but at least Nvidia hasn't stagnated like Intel has.
 
Since we don't see many rx590 models do you think these other companies know something we don't or perhaps think something is coming? Maybe they aren't making the 590 because they don't want to be stuck with inventory when this rumored cheaper card that kicks ass comes out.

I was kind of hoping we were going to see some new AMD cards like when the rx480 came out vs the 1060. 480 4g vs 1060 6gb is much closer in performance than a 590 vs a 2060, but the price difference is roughly the same.

One last option I have is pick up another rx 480 on eBay for about 100$ and run in crossfire. Gamer nexus did a great article taking a 480 crossfire with a 580, but the results were not exciting. It's inconsistent and hard to tell if a future game will make use of it. Some games run worse and others much better.
 
I actually briefly owned an RX 590. They aren't a terrible card, but they aren't earth shaking. They are essentially an overclocked RX 580, and the 580 is basically an overclocked RX 480.

My situation was I bought a refurbished GTX 1070 at Christmas time. Well the 1070 had an issue where the fan was very very loud. I returned the card (got it at microcenter). But for stock, and the money, the closest thing they had was the RX 590. So I got that.

A few days after, the RTX 2060 was announced. Since I'd paid 280 for the 590, and had store credit left, I decided I'd go get a 2060. A few days before I went to get the 2060, they discounted the Vega cards. I was still between the 2, but the day I went to get it, I settled on the Vega 56.

The 590 is a great card for 1080p no question, but if you are going to put down $280 for a card, spend the money and get the RTX 2060 or Vega 56, or a gtx 1070ti, something like those cards.

However, now I'm seeing the 590 coming down toward $260 in some cases. In that respect, they are a good value proposition I think because you can max out almost any game at 1080p and maybe dabble in 1440p with them, and get them for almost $100 less than a 2060. But if you have 300, I suggest saving a little more and getting a stronger card.

I still say one of the biggest reasons I did not get a 2060 is the 6gb of vram. The card is great now, but I like having that extra 2gb if vram on the Vega in in case in a year or so, as games become more demanding, I don't have to worry about that limitation. Plus from what I see online, you can get 1070ti/1080 performance with some tweaking on Vega 56. Which is right in line with a 2060.
 


intel really need that kick. they think they already safely put AMD in the track to become another VIA so they set their vision to destroy ARM in the mobile world next with their x86. they kick Paul Otellini out because they said he does not have clear vision on how to tackle the mobile market. so this is how you treat the guy that help you when your Pentium 4 sucks and coming out with new processor that put the pause to AMD advance back then? but in the mobile market they also have their own 800lbs gorilla with the name of Qualcomm. intel end up wasting a lot of money in their conquest to dominate the mobile world and yet they unable to make any solid foothold in the mobile market. and they are sour with qualcomm saying what Qualcomm been doing is anti competitive while at the same time they also did it before against AMD.

the situation with nvidia is a bit different. they were pure GPU company. unlike CPU where it can be everywhere GPU can only do graphic (in the past). what if integrated graphic become strong enough that you no longer need discrete GPU to play your games? discrete GPU could end being a niche product like how it was with dedicated sound card. for that reason nvidia need to create a reason where discrete GPU will always be needed. that's why there where big push towards increasing the resolution for the past ten years. the decision to support ray tracing at this point despite being very computation taxing even for the fastest GPU also have to do with this reason. and they also need to create where GPU will be useful not just in graphic related task. that way GPU can also be everywhere just like how it was with CPU for decades. intel see nvidia as a big threat in this regard.
 
Since we don't see many rx590 models do you think these other companies know something we don't or perhaps think something is coming? Maybe they aren't making the 590 because they don't want to be stuck with inventory when this rumored cheaper card that kicks ass comes out.

one of AMD usual problem is they tend to have more inventory that needed even when they were about to move to next generation batch of cards. this has been happening for years. this is one of the reason why we always see heavy discount on AMD GPU vs nvidia. AMD is expected to coming out with something new by H2 this year. and unlike how it was before AMD is more tight lip about their upcoming product (to prevent leak) that AIB probably will also be left in the dark until they were very close to release. so in AIB case they probably did not know what AMD have in stores either. but instead i think some board partner did not want to get caught in excess inventory issue so new product or not some board partner probably did not intend to offer RX590 but keep continue to deplete the existing RX580/570/480/470 in the channel. remember the mining bubble already burst (for now) so they might already have more cards than it should.

I was kind of hoping we were going to see some new AMD cards like when the rx480 came out vs the 1060. 480 4g vs 1060 6gb is much closer in performance than a 590 vs a 2060, but the price difference is roughly the same.

if we are lucky then we might see more competition in H2 2019.

One last option I have is pick up another rx 480 on eBay for about 100$ and run in crossfire. Gamer nexus did a great article taking a 480 crossfire with a 580, but the results were not exciting. It's inconsistent and hard to tell if a future game will make use of it. Some games run worse and others much better.One last option I have is pick up another rx 480 on eBay for about 100$ and run in crossfire. Gamer nexus did a great article taking a 480 crossfire with a 580, but the results were not exciting. It's inconsistent and hard to tell if a future game will make use of it. Some games run worse and others much better.

avoid crossfire. they were not that great even when multi GPU was in it's golden age. it is even worse right now.
 
I think crossfire and sli will kind of become more of a niche thing. Plus, not many games seem to support those. I think Nvidia now only supports sli on their 80 series. But single card solutions are becoming so powerful that I'm not sure most people need sli/crossfire unless you've got some special use case or just have money to burn.
 
well in reality there is no games with "native" support for multi GPU anyway. games are not build to take advantage of multi GPU the same way it did with CPU. it was more or less driver hack from AMD and nvidia to make it work in games. hence even if the said games will have CF/SLI support by game developer they still make it work from driver hack and not making the game engine to intelligently detect the presence of extra GPU in the system. this is the problem with multi GPU since the very beginning. DX12 finally integrate multi GPU function into 3D API (instead of needing separate API like CF or SLI) but the implementation still largely the same how it did with CF/SLI.