NVIDIA GeForce RTX 2070 SUPER and 2060 SUPER Video Card Review

David_Schroth said:
Given the amount of time allocated for launch reviewed, it's very difficult to cover every permutation out there, so we have to make judgement calls at some point.

The other challenge with this generation is that we do not have a stock of press samples sitting around to use as comparison hardware. In the cases where it absolutely makes sense for the review (and can be used for future articles), I shell out for the hardware, but if it's not crucial, I think everyone would prefer that money goes towards labor.

For the 2080 non-FE, that would make most sense to look at for a AIB 2070 Super review, but perhaps can be accomplished through overclocking our FE card. Though, with the 2080s leaving the channel, how relevant is that to someone looking to buy a new card today?
Click to expand...

Back when NV started releasing so many variations of Pascal I commented about this very dynamic of how most sites would eventually find it difficult or nearly impossible to compare even within the same generations of cards. I honestly don't remember how many iterations of the 1060 popped up back then not to mention 1050 or 1070's. We're seeing a similar effect now with Turing. From the original RTX's to 1660 to Super the pattern continues and I doubt it's far from over for the Turing fab.
 
David_Schroth said:
Given the amount of time allocated for launch reviewed, it's very difficult to cover every permutation out there, so we have to make judgement calls at some point.

The other challenge with this generation is that we do not have a stock of press samples sitting around to use as comparison hardware. In the cases where it absolutely makes sense for the review (and can be used for future articles), I shell out for the hardware, but if it's not crucial, I think everyone would prefer that money goes towards labor.

For the 2080 non-FE, that would make most sense to look at for a AIB 2070 Super review, but perhaps can be accomplished through overclocking our FE card. Though, with the 2080s leaving the channel, how relevant is that to someone looking to buy a new card today?
Click to expand...


More than fair enough and I am sure with all of the reviews coming we will have some valuable data points to compare to.
 
Grimlakin said:
More than fair enough and I am sure with all of the reviews coming we will have some valuable data points to compare to.
Click to expand...

Hopefully so. For each aib card we look at we will figure out the most relevant comparisons at that time and use those - can't promise our rationale will agree with yours though!
 
And let's not forget follow-ups, anything we don't cover in a launch review can be covered in a follow-up potentially. I like it that way, it allows us to focus on specific topics and tests in a much more thorough and focused way.

Trying to cram everything in one review can be messy, and you lose focus and thorough testing. By pulling things and comparisons out into separate reviews, we can be very thorough on specific comparisons. We aren't shy about doing that.
 
I agree with the core of what you are saying Brent. Where I find stumbling blocks, and this may just be me. Is when review 1 is done in one month. Then 3 months later review 2 is done with a completely different patch/driver set. While it might have parts I want to compare in my mental checklist it doesn't do anything for me because the data is then skewed.

Not to say it has no value. Far from it. And hell it's not like I'm going out and dropping my cash on cards to compare for others.
 
History has taught me that often large changes in performance through drivers over time doesn't really happen. The only case this happens is if a specific game receives direct optimizations because of some large performance bug or new optimization found. Just generally, testing drivers from the beginning of the year to the end of the year doesn't show much change in overall game performance. I've been doing driver over time reviews like that for years, and have never seen large changes. Therefore, the performance is rather similar still 2-3 months down the road.

On the other hand, it's good to continually test new drivers anyway just in case, and you can always look back at the launch review to make comparisons to see if performance has changed. You typically want to make a buying decision based on current data, not old data.
 
I had been a strong proponent of keeping a database or table of video card results. That makes it easier to compare any two given video cards.

Kyle thought it was a dumb idea because video card performance can change between reviews based on drivers, different testing, etc.

I still wish there were a more broad comparison tool, even if it's imperfect. Something like Anand's Benchmark Comparison tool.

I don't think it's an exclusive option with real world/real gaming comparisons; which I think are invaluable, just really hard to quantify when your trying to compare entire generations of older cards to a newer card)
 
Brian_B said:
I had been a strong proponent of keeping a database or table of video card results. That makes it easier to compare any two given video cards.
Click to expand...

We are planning to do something like that once we have enough data and have time to sift through it and make it meaningful. It will also take a bit of custom development I suspect thanks to WordPress not really doing that out of the box.
 
Brent_Justice said:
And let's not forget follow-ups, anything we don't cover in a launch review can be covered in a follow-up potentially. I like it that way, it allows us to focus on specific topics and tests in a much more thorough and focused way.

Trying to cram everything in one review can be messy, and you lose focus and thorough testing. By pulling things and comparisons out into separate reviews, we can be very thorough on specific comparisons. We aren't shy about doing that.
Click to expand...

Not to mention that a follow up allows to check in on those new but sometimes useful features that a manufacturer might release down the road in a driver such as HDR or RT or some other rendering related item.
 
Brian_B said:
I had been a strong proponent of keeping a database or table of video card results. That makes it easier to compare any two given video cards.

Kyle thought it was a dumb idea because video card performance can change between reviews based on drivers, different testing, etc.

I still wish there were a more broad comparison tool, even if it's imperfect. Something like Anand's Benchmark Comparison tool.

I don't think it's an exclusive option with real world/real gaming comparisons; which I think are invaluable, just really hard to quantify when your trying to compare entire generations of older cards to a newer card)
Click to expand...

It's been a few years so I don't remember the site but there use be a neat one that had a nice interface on it. You could pick from 10+ years of NV or AMD cards. I just did a quick search but couldn't find it. It had a side by side format with incredibly detailed specs. It even had a check box for multi gpu options. Pretty much every metric known for the cards and not just the basic stuff. You could pick from a drop down box on each side and it would give indicators next to each spec showing a increase or decrease. All that I remember is that it had a green/white/black kind of web page color design. For years it was a great reference tool but it seemed to stop being updated around the time of Pascal so I stopped going there. Really miss that site.
 
lostin3d said:
It's been a few years so I don't remember the site but there use be a neat one that had a nice interface on it. You could pick from 10+ years of NV or AMD cards. I just did a quick search but couldn't find it. It had a side by side format with incredibly detailed specs. It even had a check box for multi gpu options. Pretty much every metric known for the cards and not just the basic stuff. You could pick from a drop down box on each side and it would give indicators next to each spec showing a increase or decrease. All that I remember is that it had a green/white/black kind of web page color design. For years it was a great reference tool but it seemed to stop being updated around the time of Pascal so I stopped going there. Really miss that site.
Click to expand...

https://gpu.userbenchmark.com/ ?
 
SeymourGore said:
https://gpu.userbenchmark.com/ ?
Click to expand...

No unfortunately not. I saw that one and GPU boss and one or two others also. If I remember right it looked like another word press type site but with the sides being green and on the left direct links to various cards, then the middle where the content was white. I can't remember the top part but towards the middle, but below the top section were the drop downs to select the cards. I think the right was just ad's. I'll have to look more later.
 
Passmark has been what I typically turn to when looking at pretty old stuff. It's far from accurate, but at least you get some idea. If you were looking to upgrade, say a GTX780 to something released today - sure, almost everything will be an upgrade, but I couldn't tell you how much, or at what level you'd be looking at sidegrade over upgrade.

I liked Anand's tool because, provided they had some common bench data, you could do direct comparisons between two cards. Passmark is nice because it will show you a range of hardware similar to what your looking at. I am not familiar with userbenchmark, but it appears the same way.

Problem is, with Anand gone that site doesn't get the same attention. Passmark is a single number and is easily skewed. Not an easy task, I will admit.
 
You must log in or register to reply here.
Become a Patron!
Back
Top