[Question/Discussion] Tech Deals just Recommended Nvidia over AMD... Legit Review or Not?

[Katarn]

I was happy to see Tech Deals covering the RX 5700 XT yesterday. But in the end I saw that his results/inferences were the opposite of what I had expected or seen until then.

In a video comparing the Gigabyte Gaming OC RTX 2060 Super to the RX 5700 XT, he made the following points:

 

- performance difference in 1080 and 1440 is only marginal (and unnoticeable)

- Nvidia is better for 4k

- Nvidia has a "much better hardware encoder" for screen-capture

- Nvidia has better drivers

 

Basically, the AMD is only 6% faster in gaming and inferior in everything else.

 

The way he put it, it actually makes sense. I made a thread about the cards themselves in the GPUs forum. But there is something else that caught my attention, and that is that the video gives off a manipulative vibe:

- he downplayed the difference between the RX 5700 XT and the 2070 Super

- he used DLSS in the benchmarks

- he made no mention of sharpening filters

- he made a big deal about the effects of ray tracing, and showed only Minecraft (as opposed to other games that utilize the feature and make barely any difference in visuals)

 

On the other hand: Tech Deals is not an unreliable channel.

He has been impartial so far. He has recommended the RX 480 over the GTX 1060 in the past, and his recent video on CPUs concluded with AMD as the winner (dethroning the i7 8700K with a Ryzen 5)... so I don't think he would be making a video in favor of a brand.

 

Which means that the inferences in favor of the 2060 Super are most likely true...

But then how do we explain the fact that most other reviewers recommend AMD?

 

I am curious to hear your opinions on this -- is Tech Deals' review of the RX 5700 XT (vs the 2060 Super) valid or not? Would it sway your decision to buy? Have you decided between the two before (if so, what are your experiences with the cards you've chosen)?

 

[this thread is meant for discussion, and no negative connotation is meant for either the brand(s) or the reviewers(s)]

Peace

[Fasauceome]

This is why outlining testing methodology is so important. Everybody loves their userbenchmark and whatnot, but if you only read the conclusion without knowing the testing methodology, the data is useless. Tech Deals reached a conclusion that many on this forum haven't, but he also showed how he got there, so the data can be understood.

16 minutes ago, Katarn said:

performance difference in 1080 and 1440 is only marginal (and unnoticeable)

- Nvidia is better for 4k

- Nvidia has a "much better hardware encoder" for screen-capture

- Nvidia has better drivers

These are all legitimate reasons to favor a product over another, which should considered when buying.  Of course the issues noted

 

16 minutes ago, Katarn said:

he downplayed the difference between the RX 5700 XT and the 2070 Super

- he used DLSS in the benchmarks

- he made no mention of sharpening filters

- he made a big deal about the effects of ray tracing, and showed only Minecraft (as opposed to other games that utilize the feature and make barely any difference in visuals

are also significant, but the important thing is that we know this happened and can factor it into our own decisions instead of relying on the conclusion of the reviewer.

 

Overall not a bad review, even if you think 5700XT > 2060S there's plenty of meat here.

 

[MeatFeastMan]

The answer to your question is simple. Different benchmarks, different systems, different results and therefore.. different conclusions. He included a game where DLSS was used, therefore the gap closed between the 5700XT and 2060 Super to just 6%. He also only tested in 4 games.

 

He makes good points but as I say..people come to different conclusions and that's fine. He is in the minority but his opinion is valid.

 

For example..this Paul's Hardware video is on the other end, where the 5700XT is seen smashing the 2060 Super by nearly 15%, even at 4k. Again, this video is similarly in the minority of reviews, but on the other end of the spectrum.

 

Personally if you want the best picture for a graphics card as to how they perform against their competition, there is of course no better channel for that than Hardware Unboxed, as you can see here:

 

I would say Hardware Unboxed tends to be on the money every time, they are usually always in the middle ground and give you tons of benchmarks. I always go to them first when a new GPU or CPU arrives on the scene.

 

My point is..everybody is right because they interpret only what THEY saw with their own eyes using THEIR unique setup.

[Augustin]

36 minutes ago, Katarn said:

he made a big deal about the effects of ray tracing, and showed only Minecraft

When I clicked on the headline I was thinking, i bet this is 100% about ray tracing in minecraft.

[Jurrunio]

That's why one should always use multiple sources. Everyone has their bias in things and it could change over time.

[Katarn]

4 hours ago, Fasauceome said:

Overall not a bad review, even if you think 5700XT > 2060S there's plenty of meat here.

I don't think that anymore. I have taken quite an agnostic stance on the matter (which sucks because I need to make that decision sooner rather than later).

I am a bit more swayed on the Nvidia side at the moment, mainly for its hardware encoder and the fact that I've been exclusively using Nvidia until now. But I recognize that the latter reason is bias.

That is why I find it imperative to learn which of the cards performs better for rendering (Blender, video-editing software, Unreal Engine), whether the AMD driver/heat/sound horror stories are true, and whether the difference in in-game performance is important. But that discussion is in my other thread. This is about the review, and I think we are on the same wavelength on it: he may a bit biased, but not paid; his conclusions are valid.

I appreciate your input.

 

4 hours ago, MeatFeastMan said:

For example..this Paul's Hardware video is on the other end, where the 5700XT is seen smashing the 2060 Super by nearly 15%, even at 4k. Again, this video is similarly in the minority of reviews, but on the other end of the spectrum.

How is it possible to have different results when testing the same games, with the same hardware and on the same settings? How do we know how many of those settings are tweaked in favor of one card, as was the case with tessellation ruining performance for AMD cards in the past?

 

Also in that video, the table at the 1:00 minute mark shows the XT to have a base clock almost as high as the boost clock of the 2060 Super, and equal to the base clock of the 2070 Super.

But on every other site I've visited, the 2060 Super does not show a base clock, it rather includes a "core clock", and (in Gigabyte's Gaming OC case) it is higher than the base clock of the XT. I don't understand what that means.

4 hours ago, MeatFeastMan said:

Personally if you want the best picture for a graphics card as to how they perform against their competition, there is of course no better channel for that than Hardware Unboxed, as you can see here:

That is very true, yes.

I've been using Tech Deals because he is the "consumer-oriented 'this is the best deal' guy". That is why his unorthodox stance here surprised me.

How would you compare Hardware Unboxed to Gamer's Nexus?

5 hours ago, MeatFeastMan said:

My point is..everybody is right because they interpret only what THEY saw with their own eyes using THEIR unique setup.

But the problem is that that does not exclude manipulation, skewed results, and false inferences.

 

4 hours ago, Jurrunio said:

That's why one should always use multiple sources. Everyone has their bias in things and it could change over time.

Yes, but now I am more (way more) confused than I was when I first started building this system.

I have been especially baffled about the clock speeds, which I mentioned in the underlined paragraph of this reply.

[Jurrunio]

1 minute ago, Katarn said:

I have been especially baffled about the clock speeds,

Ignore this, simple. The two GPUs are of different architectures so you can't compare frequency directly even after considering number of compute unit differences

[Princess Luna]

Tech Deals is a great channel, so is HW and most others here mentioned I believe.

 

The only thing to have in mind is that Tech Deals didn't just do a performance comparison but tried to look at the whole in order to provide a solid argument in favor of nVidia.

 

Honestly both cards are fine, I'd simplify to if all you do and care is gaming than the 5700 XT is the way to go as you do get a couple more fps out of it in every instance compared to the 2060S and 2070 however if you gonna do anything else like streaming and hardware acceleration I'd pick a nVidia at the price point.

 

It sounds a bit shallow mind set but it really resumes down to it... have in mind that Navi as in 5700 and 5700XT are indeed a downgrade on what hardware acceleration workloads are concerned such as OpenCL based ones in comparison to the Radeon 7 and Vega64, so it is primarily by accounts a gaming card.

[Blue4130]

6 hours ago, Katarn said:

 

But then how do we explain the fact that most other reviewers recommend AMD?

 

It's explained by the nature of recommendations, they are simply opinions. You can choose to agree or disagree with their opinions based on your own knowledge and the strengths of their arguments. 

[Mihle]

12 hours ago, Katarn said:

- Nvidia has better drivers

That's very subjective or based on luck/unluckiness with the driver.

 

I personally prefer AMD drivers. And I dislike Getforce experience or whatever it's called a lot. (Telemetry and requiring a user to use it is just two of the reasons)

[Blue4130]

9 hours ago, Mihle said:

That's very subjective or based on luck/unluckiness with the driver.

 

I personally prefer AMD drivers. And I dislike Getforce experience or whatever it's called a lot. (Telemetry and requiring a user to use it is just two of the reasons)

You don't need to use GeForce experience, drivers are available separately if you don't want to install GeForce experience. 

[Anomnomnomaly]

I can't take any reviewer seriously if they show a clear bias towards a particular brand and adjust their benchmarking to favour said brand... So when it comes to reviews of products... try to eliminate those... or at least balance it out with fanboi reviewers of other brands.

 

Even some of the 'allegedly' respected tech channels on YT show bias... whether that's a fave brand or simply a lack of real world understanding of things because they get so much stuff for free and it distorts their ability to be able to objectively take into account value/performance and so forth.

 

Take Linus getting a $10,000 gaming monitor (kinda guessing on the price as the 27" is a cool 2k USD) for free and taking it home as per his video the other day... If I was given a freebie like that for my house... Pretty sure I could live with it too over an older screen I've had for a few years.  But when you have to save up your own money... making the right choice is far more important and I can't think of a single YT channel that is able to be truly objective on the subject.

[Mihle]

1 hour ago, Blue4130 said:

You don't need to use GeForce experience, drivers are available separately if you don't want to install GeForce experience. 

I know that. 

[Katarn]

20 hours ago, Princess Luna said:

if all you do and care is gaming than the 5700 XT is the way to go as you do get a couple more fps out of it in every instance compared to the 2060S and 2070 however if you gonna do anything else like streaming and hardware acceleration I'd pick a nVidia at the price point... [ ] ... have in mind that Navi as in 5700 and 5700XT are indeed a downgrade on what hardware acceleration workloads are concerned such as OpenCL based ones in comparison to the Radeon 7 and Vega64, so it is primarily by accounts a gaming card.

Gaming is not all that I do.
With that said, I don't think my work is of the type that demands a Radeon7 or Vega64-type performance.

In fact, I don't remember running into any trouble with my current 1060 work-wise. Won't the RX 5700 XT come as a straight upgrade, irrespective of it being AMD?

Also, since you mention the 2070... which the better deal between a 5700 XT and a 2070 with a price difference of ~$50?

 

13 hours ago, Mihle said:

I personally prefer AMD drivers.

What specifically makes you prefer AMD drivers? Is it just because GeForce Experience is annoying?

 

3 hours ago, Anomnomnomaly said:

But when you have to save up your own money... making the right choice is far more important and I can't think of a single YT channel that is able to be truly objective on the subject.

True.

But there are reviewers who try to avoid bias and see things from the consumer's point of view. That is one of the reasons I like Tech Deals so much. He tries to walk the " consumer" route; I have never known him to appear to be skewing results in favor of a brand until now. He even practices what he preaches: for instance, when building a PC for his kid, he built it with a 1060 (unlike someone we know who gave his 3-year-old a 1080 Ti).

 

[Princess Luna]

2 minutes ago, Katarn said:

which the better deal between a 5700 XT and a 2070 with a price difference of ~$50?

50 dollars in favor to which one? That's quite the decisive factor.

 

Also if your workloads focus on CUDA Cores like say Premiere Pro you will actually be seeing a side grade from the GTX 1060 on these types of workloads, so it's important to be sure what you're getting for what you'll be using.

 

Gaming wise it's a great upgrade regardless.

[Katarn]

11 minutes ago, Princess Luna said:

50 dollars in favor to which one? That's quite the decisive factor.

 

Also if your workloads focus on CUDA Cores like say Premiere Pro you will actually be seeing a side grade from the GTX 1060 on these types of workloads, so it's important to be sure what you're getting for what you'll be using.

 

Gaming wise it's a great upgrade regardless.

The Gigabyte Gaming OC 2070 is $50 more expensive than the Gigabyte Gaming OC 5700 XT.

 

I do use Premiere Pro. But isn't that rendering mostly down to the CPU?

By side-grade, do you mean little to no difference? In that case, a huge difference in gaming and a minimal difference in software in which I've had no trouble to begin with is not that bad. I think it is Blender and Unreal Engine where I strain the GPU the most; In Premiere, I mostly edit with minimal special effects.

But, just in case: is there video-editing software on-par with Adobe that doesn't discriminate against AMD?

 

Gaming-wise the only things I think I'll be missing are the NVidia-exclusive eye-candy things like the rain effects on the cape in Batman: Arkham Knight or the visual enhancers for GTA V.

[Mihle]

9 hours ago, Katarn said:

 

What specifically makes you prefer AMD drivers? Is it just because GeForce Experience is annoying?

Geforce experience is more than just annoying, but it's more than that AMD settings looks better

[Blue4130]

9 hours ago, Mihle said:

Geforce experience is more than just annoying, but it's more than that AMD settings looks better

But you don't have to use GeForce experience. That's completely optional. I run Nvidia drivers just fine without GeForce experience. 

 

You honestly can see a visual difference between Nvidia card output and amd output? (i haven't run an amd card in over 15 years, i truly don't know the answer) 

[Mihle]

6 hours ago, Blue4130 said:

But you don't have to use GeForce experience. That's completely optional. I run Nvidia drivers just fine without GeForce experience. 

 

You honestly can see a visual difference between Nvidia card output and amd output? (i haven't run an amd card in over 15 years, i truly don't know the answer) 

You ignore what I said earlier and read stuff wrong.

 

I know you don't need it, but then you don't have some things.

 

I mean the look of the driver program itself.

[williamcll]

Well you shouldn't base your views of a product solely on one sepcific website anyways.

[masethekiller]

Ati used to be the better value. On higher end 5700 is good if you want a higher end card and pay less than Nvidia.

 

  But the 1660 is much better than the RX 580 and only 20 dollars more for 200 dollar range cards   Also Nvidia cards are smaller and more power efficient.

 

But Ati could improve and be more competitive in the future. So you have to pay attention and get whatever is the best deal.