• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

Which graphics card would you buy?

What would you buy?

  • RX 5600 XT

    Votes: 3,943 22.8%
  • RX 5700

    Votes: 6,540 37.8%
  • RTX 2060

    Votes: 4,646 26.8%
  • GTX 1660 Super

    Votes: 2,186 12.6%

  • Total voters
    17,315
  • Poll closed .
There is something around that sort of simulate folding but I can't seem to find it. THIS is a database but since it based on reports from folders the worse WUs are left out. The newest "bad" WU are not shown and these are the ones that you might encounter.

Well, that's interesting. At the very least, the RX 580 average PPD seems more or less accurate, I used to get around 180k to 200k points for about 12 to 14 hours of folding, so it kinda adds up to the 330k average shown on that spreadsheet.
 
Waiting for "Big Navi" ;) some 5800 XT, 5900 XT, whatever ;)
 
LoL
Non of the ones in the list i am waiting for big navi
I refused to buy nvidia idiot priced gpu's as much as i could.
Sadly i had to buy a card because my old started artifacting ... asus crap ofcourse 1 month after the legal warranty ended.
So i settled for a 1070 and still use it, since the radeon VII did not perform as fast as i hoped it would.
If AMD is going to charge big prices on the big navi as well, i simply keep using this card till i see someone selling a second hand fast card for less than 500 dollar.
I admit i missed one already a 2080ti for 550 but saw it too late, it was sold before i could react.
Question is was it fake or not
 
Last edited:
If I had to produce a build today ... and assuming the grouping represents all cards fitting "within budget", the 2060 is obvious ... its the fastest among the bunch with all cards overclocked

Using the 2060 Reference card as baseline and TPU overclocking data and MSI test results except where unavailable

Reference RTX 2060 = 103 % / 103 %
MSI RX 5600 XT Gaming Z = 105% / 103% x 93.9 / 90.2 OC = 106.12%
Powercolor RX 5700 Red Dragon = 112% / 103% x 103% / 102 % OC = 109.80
MSI RTX 2060 Gaming Z = 103 % / 103 % x 121.1 / 107.9 = 115.60
MSI GTX 1660 Super Gaming X = 86 % / 103 % x 94.3 / 83.7 OC = 93.67

OK let's now factor in price of new system, again baseline = new system with MSI RX 5600 XT Gaming will be our baseline at $1200 ... ... and we've seen that from TPU and other reviews reviews that the Performance of the Gaming, Gaming X and Gaming Z are virtually identical ... the X even being insignificantly faster in 5600 XT testing. For the most part, they are identical cards with different overclocks ... The only thing we have to check is if they have the same memory. In the case of the 5600 XT, the Gaming Z has 14 GBps memory, the Gaming Z does not. While a price premium here would seem to be warranted for the Z, again, the Z is actually slowe in TPU testing. For the 2060, the Gaming and Gaming Z have the same 14 GBps memory.

MSI RX 5600 XT = $333.98
Powercolor RX 5700 Red Dragon = $359.99 adds (+$26.01) to our $1200 build
MSI RTX 2060 Gaming = $275.99 (subtracts (-57.99) from our $1200 build
MSI GTX 1660 Super Gaming X = $259.99 subtracts (-73.99) from our $1200 build

MSI RTX 2060 Gaming = 115.60 x 100 / $1,142.01 = 10.12
Powercolor RX 5700 Red Dragon = 109.80 x 100 / $1,173.99 = 9.35
MSI RX 5600 XT Gaming X = 106.12% x 100 / $1,200 = 8.84
MSI GTX 1660 Super Gaming X = 93.67x 100 / $,1126.01 = 8.31

So out of the cards listed, I'd be looking at a MSI Gaming 2060 or a comparable priced card from one of the other AIB vendor's top model series.

Unfortunately it is impossible to find tested cards for ever vendor and ever model. It should be obvious however that the same PCBs are used across many model lines with each AIB provider; the main differences are coolers, factory OCs and memory. As we was saw in the 5600 XT testing, the better memory actually produced lower overclocked performance. In addition, since the baseline for pricing was the 5600 XT, I tried to use MSI across the board, except when TPU did no testing. There is no reason to dismiss other AIB cards which you might find that are lower cost ... but keep the comparisons comparable ... Asus Strix line, Gigabyte Windforce, etc. Check TPU testing when available.

And yes, there is a bit of bias inherent to the source material

-The 5600 XT Gaming Z and Gaming with different memory have virtualy the same performance .... so used the cheaper X version in pricing
-The 2060 Gaming Z and Gaming use the exact same memory, with only **physical** difference being the factory OC



The power consumption is unimportant. If you worry about that you should not buy a PC at all.

1. All other things being equal, why not ? I mean if ya parents pay the bill ... or your electricity is included in the rent, I can agree. But the power savings over 4 years can almost pay for the purchase cost. Not as much of an issue at the low and middle range, but at the high end, it's certainly significant.

At the high end, differences in performance can be 100 watts or more ... over 4 years, I'll spend $178 more on electricity playing 30 hours a week

2. Using the (1) 120mm fan for every 50-75 watts of power draw rule of thumb (75 - 100 watts for 140mm), that means adding an extra case fan to compensate ... + $10 - $30, say $20

3. Adding 100 watts to the cost of a decent power supply, you're talking $10-$20, say $15

So... if $213 ($178 + $15 + $20) is insignificant, then giving and edge to a lower performing card because of price is counterintuitive defying any sense of logic.
 
Last edited:
None, since i'm tighten my wallet now
 
RTX 2060, because it has support for hardware accelerated Raytracing and machine learning. DLSS from Wolfenstein on is crazy good now and puts the performance of it way above a 5700. RTX also runs good at 1080p.

Furthermore it features a more futureproof architecture as it supports VRS and Mesh Shaders, I think that's more important for futureproofing than just 2 GB more VRAM.
 
RTX 2060, maybe RTX feature isn't good right now, but I definitely don't want to debug drivers for AMD ;)
 
I would get 2060 because I'm experimenting with ray tracing. I don't need billions of rays per second, just need to see how much faster they are comparing to previous compute pipelines.
 
Aslong AMD drivers are like this and , its totally random if you get a buggy pc, i would never buy amd, and just buy the nvidia i could afford at any time.
And if im to buy a new card, i would get a 2070 super minimum.
 
definitely would take the RX5700 if I had the money. It's not always the driver that screws up the GPU, it's always the software side of things. One major thing that no one ever mention (including AMD themselves) is to do a fresh driver install for every new driver release & disabling all the useless features like Anti Lag. Still will perform better than the 2060 & 2060 Super (assuming best case scenario that is).
 
definitely would take the RX5700 if I had the money. It's not always the driver that screws up the GPU, it's always the software side of things. One major thing that no one ever mention (including AMD themselves) is to do a fresh driver install for every new driver release & disabling all the useless features like Anti Lag. Still will perform better than the 2060 & 2060 Super (assuming best case scenario that is).

The 2020+ driver installer, uninstalls the old driver and there is an option to reset to factory before it installs.

Those features aren't enabled by default.
 
Tough call... a more performant card in the 5600xt with a higher potential of issues due to the driver, or a slightly slower and more expensive (depending on which model 5600xt is chosen) with DLSS and RT capability...
 
I voted for the RTX2060. I feel RTRT is the future and find the effects very appealing. AMD needs to get their RTRT game on and get the lead out.
 
As a Linux user I'm pretty much limited to AMD's offerings. Nvidia drivers are just a bunch of problems on rolling distros.
I have not experienced your problem. I run various versions of Linux, including Android, and mainly run NVidia GPU's without problems.
 
@lexluthermiester obviously that card is now the best value for an entry into Ray Tracing. Assuming that game studios know their way around in optimizing the game engines for it.
 
I have not experienced your problem. I run various versions of Linux, including Android, and mainly run NVidia GPU's without problems.
If your are a folder then it is a problem when new driver is installed without a re-start. At least I had the problem on Ubuntu 18.04 LTS. I learned to look for what it suggest for updates and act on it.
 
obviously that card is now the best value for an entry into Ray Tracing. Assuming that game studios know their way around in optimizing the game engines for it.
Exactly my thoughts.
If your are a folder then it is a problem when new driver is installed without a re-start. At least I had the problem on Ubuntu 18.04 LTS. I learned to look for what it suggest for updates and act on it.
Then disable automatic updates.
 
I learned....
 
I'm not buying any of these , next.
 
Tho limited opthions!
I want something better!
 
None, all are too slow.
 
I must have been insane to buy Radeon GPU.
Since i was a kid, i loved AMD, my first pc was an Athlon 64, with an x800 gto as a GPU.
I had my share of intel systems, but those were the times, and now, in late 2019, i.ve bought an new rig with r7 3700x(with aorus x470 gaming 7 wifi) + radeon rx5700xt sapphire nitro+.
Worst mistake of my life; why?
*Warcraft 3 shuttering
*Heroes 3 crash
*Nfs Payback crash
*Wicher 3 shutter
*Hell even MS excel was running slower than my last pc(i5 8400 + 16 gb ram + gtx1070ti)
And the general feel of the system is, even after all the driver updates is ok.ish, even with 2 nvme ssd.s.
And dont tell me that maybe i had bad luck, but i.ve builded another 3 other pc.s with rx3700x, with diferent motherboards, ram, ssd.s, gpu.s.
All of them are not as smooth as an intel system, no matter what. Plus, idle power consumption is with 45w more than an equivalent intel(even i9 9900k).
Ok, in full load it changes, but in normal tasks, aka gaming, using the same gpu, an intel i9 9900 still consumes less power, end in lite tasks there is no comparrison.
To exemplify:

I9 9900k all sys idle + gtx 1080 - 70w; 3dmark max pwr 550w; gaming nfs payback 350w, lite tasks 120-150w

R7 3700x + gtx1080 idle - 110w; 3dmark 450w, nfs payback 330w; lite tasks 180-200w

I9 9900k+ rx5700xt nitro idle 100w, 3dmark 590w, nfs payback 420w, lite tasks 150-180w

R7 3700x + 5700xt nitro idle 130w; 3dmark 500w, nfs payback 420w, lite tasks 220-250w.

The power supply is an seasonic 1000w titanium. Ram 32gb kingston predator, ssd.s A-data 8200 nvme.

Now i.m back on my old i5 8400 + grx 1070ti system as its infinitly more stable, and even if i loose some fps, the whole experience is much smoother.

All my experience with these compoments are in a large time frame( 4-5 months) and i am trully sorry to say all these things, but i cant change the way i feel. Now i.m selling the AMD rig at half the price, and keeping my last system, and waiting to find a good offer on an i9 9900k+motherboard+nvidia gpu.

I.m not a fanboy of intel+nvidia, i trully wanted to be happy with my new AMD rig, but i just could not endure all the small things that were happening, and waiting for AMD to fix them.
I.m done, i quit on AMD, sorry.
Hope my experiences helps others.
Take care
 
Back
Top