Review AMD Radeon RX 7800 XT Review: The Lateral Pass

[Order 66]

Yes I do like that video, thanks Neo! I've been subbed to IcebergTech for a long time. I like how there is no favouritism in his video whatsoever. To him, it's all about what he needs and who can deliver without passion or prejudice.

That's how it should be when something is reviewed.

I also am subbed to IcebergTech. I love his videos because he tests a lot of older hardware in newer games when no one else really does.

 

[NeoMorpheus]

Hmm, AMD/Nvidia/Intel won't like that. I'm sure they would prefer customers doing more frequent gpu upgrades. What's the average been, every other generation? IDK.
Some folks even saw/see DLSS/FSR/XeSS as a means to stretch out their next upgrade

I will be LOL'ing so hard when the next dlss requires a new GPU, just like they did with dlss 3.
And yes, AMD just pulled the same thing with FSR 3, but I think (could be wrong) they are planning in expanding that to the others gpus.

 

[Avro Arrow]

I also am subbed to IcebergTech. I love his videos because he tests a lot of older hardware in newer games when no one else really does.

Yep, something that gets overlooked a lot is the used market and some people will have a certain CPU or video card and want to know if it's worth spending the money on a new game or if they should upgrade first. He provides a pretty good service to a large group of people who benefit greatly from it.

 

[NeoMorpheus]

That's how it should be when something is reviewed.

Yet we consumers get crucified when asking for that.

I also am subbed to IcebergTech. I love his videos because he tests a lot of older hardware in newer games when no one else really does.

Thats what really boils my blood these days.

Every single video, watch how they always pull and use a 4090.

Be a case review, a build, games benchmarking, etc. all are the same.

If we were to believe them, the steam survey should be showing only one gpu, the 4090.

 

[Order 66]

Yep, something that gets overlooked a lot is the used market and some people will have a certain CPU or video card and want to know if it's worth spending the money on a new game or if they should upgrade first. He provides a pretty good service to a large group of people who benefit greatly from it.

I do love it when he revisits old hardware that is somehow surprisingly capable even in 2023.

 

[Deleted member 2950210]

Why are you even commenting on a mid tier graphics card article if you only care about 4090?

‘Cause i was trying to reply to a comment made by NeoMorpheus.

If you bother to go through this entire thread, i think you'll find that, bringing up 4090, is not so much off topic as you might believe.

And i don't even think i was the first to mention it.

In all honesty, and considering the way this whole discussion has evolved, i think we've all seen threads way more overstretched than this one.

In any case, if i went too far, i humbly apologise to you and everyone else here.-

 

[Avro Arrow]

Yet we consumers get crucified when asking for that.

Thats what really boils my blood these days.

Every single video, watch how they always pull and use a 4090.

Be a case review, a build, games benchmarking, etc. all are the same.

If we were to believe them, the steam survey should be showing only one gpu, the 4090.

I don't trust the Steam survey because I've never been able to get it to work on any of my PCs going all the way back to my old FX-8350 build. When I first got Steam, I had twin HD 7970s in crossfire and since then I've had twin R9 Furies, an RX 5700 XT, an RX 6800 XT and now my RX 7900 XTX. I don't know if Steam has any idea that I had these cards. Then of course there's the cyber-cafe question about several different Steam accounts using the same hardware. Steam knows about the issue but have never done anything to rectify it.

 

[Avro Arrow]

I do love it when he revisits old hardware that is somehow surprisingly capable even in 2023.

Yep, and that's why what he does is so important. People get to see if what they have is good enough for what they want to play.

 

[Order 66]

Yep, and that's why what he does is so important. People get to see if what they have is good enough for what they want to play.

I also love seeing older or less powerful gpus get "tortured" by running games at 4k ultra when the GPUs being tested have no business playing at 4k.

 

[Avro Arrow]

I also love seeing older or less powerful gpus get "tortured" by running games at 4k ultra when the GPUs being tested have no business playing at 4k.

Yeah, it just goes to show you how over-inflated some of the older GPUs were when released. The ridiculous RTX 2080 Ti comes to mind.

 

[Order 66]

Yeah, it just goes to show you how over-inflated some of the older GPUs were when released. The ridiculous RTX 2080 Ti comes to mind.

The 2080 ti may have been overpriced but at least it had more than 8GB of vram. You would have had to pay $699 for a 3080 to get a high end GPU with more than 8GB from the 30 series. (I know the 3060 12GB exists which is why I said high end.)

 

[NeoMorpheus]

I don't trust the Steam survey because I've never been able to get it to work on any of my PCs going all the way back to my old FX-8350 build. When I first got Steam, I had twin HD 7970s in crossfire and since then I've had twin R9 Furies, an RX 5700 XT, an RX 6800 XT and now my RX 7900 XTX. I don't know if Steam has any idea that I had these cards. Then of course there's the cyber-cafe question about several different Steam accounts using the same hardware. Steam knows about the issue but have never done anything to rectify it.

I mentioned these things before of this subject:

1- The only way that the Steam Survey could be considered accurate is if it ran every month on every single endpoint.

2- As you stated, many others have expressed the same experience, the survey doesnt show up for AMD gpus, but does for ngreedia and that has also happened to me. I'm not insinuating anything simply something curious that others have also experienced.

3- As much as the survey "sucks" accuracy wise, it is the closest thing we have to a reliable source of info for these things.

I personally wonder if my system is counted since its running ChimeraOS, which might also show as holoiso, which is what the SteamDeck reports as.

Also, I wonder, are these systems automatically counted? Is my system counted even though its not a SteamDeck neither running proper holoiso?

 

[Order 66]

I can use quotes from other threads in any other thread but not in conversations why is this? example:

Another benchmark that unfairly favours Intel. No surprise there. When they realised that AMD's CPUs were crushing Intel's, they were like "Oh we can't have THAT now can we?" and completely screwed changed their alogrithm.

As for people who use userbenchmark to help them choose a CPU...

 

[TJ Hooker]

1- The only way that the Steam Survey could be considered accurate is if it ran every month on every single endpoint.

If you want 100% accuracy, sure. If you're willing to accept a margin of error/confidence interval (like most surveys), you can have a much smaller sample size. E.g. you can get results within +-1%, 19 times out of 20, with 10k samples. If you're aiming for +-3%, you'd only need 1k samples, etc.

That assumes samples are being selected appropriately though, and sampling methodology is a big question mark for the Steam survey AFAIK.

Edit: Even if Steam attempted to survey 100% of client, it still wouldn't be 100% accurate. Not all clients would be online during the survey window, and some people would decline the survey.

 

[NeoMorpheus]

If you want 100% accuracy, sure. If you're willing to accept a margin of error/confidence interval (like most surveys), you can have a much smaller sample size. E.g. you can get results within +-1%, 19 times out of 20, with 10k samples. If you're aiming for +-3%, you'd only need 1k samples, etc.

I would be ok if the %100 userbase survey is pushed at least once a year, then the others as currently is.

That assumes samples are being selected appropriately though, and sampling methodology is a big question mark for the Steam survey AFAIK.

Thats the problem with this survey, seems to be unfair to certain devices and no clear rhyme or reason as to why some get them and others dont.

 

[Deleted member 2950210]

2- As you stated, many others have expressed the same experience, the survey doesnt show up for AMD gpus, but does for ngreedia and that has also happened to me. I'm not insinuating anything simply something curious that others have also experienced.

Of course you 're not insinuating anything. You just consistently mock one of the two great GPU companies, all the while implying that Steam is biased in favor of it, and you have absolutely no evidence to show for it.

You are the very paradigm of objectivity.

 

[TJ Hooker]

As you stated, many others have expressed the same experience, the survey doesnt show up for AMD gpus, but does for ngreedia and that has also happened to me. I'm not insinuating anything simply something curious that others have also experienced

I've had multiple Steam HW surveys with AMD cards.

Edit:

Thats the problem with this survey, seems to be unfair to certain devices and no clear rhyme or reason as to why some get them and others dont.

Well, the sampling should be random, so some machines getting selected and some not with no clear rhyme or reason is kinda what you'd expect. As far as being biased against certain devices, do you have any real evidence of that, or just a handful of anecdotes?

 

[Deleted member 2950210]

I've had multiple Steam HW surveys with AMD cards.

Exactly! Just as I've been running Nvidia GPUs for the past 9 months and i've yet to encounter a survey.

 

[Avro Arrow]

The 2080 ti may have been overpriced but at least it had more than 8GB of vram. You would have had to pay $699 for a 3080 to get a high end GPU with more than 8GB from the 30 series. (I know the 3060 12GB exists which is why I said high end.)

Sure, but back then, there was no question about only having 8GB of RAM. At the time, 8GB was more than enough, even at 4K. The only time that I had a second thought about only having 8GB of VRAM was when I was playing Far Cry 6 on my RX 5700 XT. I couldn't use the HD texture pack because it required a really odd 11GB of VRAM. One of the things that made me laugh about the Ampere cards was the fact that an RTX 3080 couldn't use the FC6 HD texture pack and it was a new enthusiast-class card!

You see, nVidia realised that a lot of people who are willing to pay the extra for GeForce cards weren't knowledgeable enough to realise that it was a bit of a scam on anything that wasn't high-end. Like, the RTX 2060 was a total scam because it had RT performance similar to that of an RX 6600. RT performance that poor made the RTX designation irrelevant but the noobs bought them in droves. Since noobs buy by brand and experts buy by spec, nVidia knew that most people who bought their cards wouldn't question the absurdly-low 10GB frame buffer on the RTX 3080 and would just accept it because it was more than 8GB.

This is why the vast majority of people with Radeon cards are enthusiasts and/or experts. We understand what we're reading when we look at video card specifications and Radeons tend to be spec monsters when compared to GeForce cards. Now, nVidia is a very smart company, far smarter than AMD ever was. What nVidia did was spec their cards with CUDA so that content creators (and influencers) would use them to make their videos. This is why so many YouTubers use them and people see that. They also make sure that they have the fastest single card so that TechTubers will use them for CPU tests. That gets the name out, basically free advertising. For the noobs, they just keep saying "RT" and "DLSS" over and over until it sticks.

When someone buys a high-end GeForce card because of DLSS, they're clearly a noob because by the time they actually need upscaling (maybe 4-5 years time), DLSS as we currently know it won't exist and neither will FSR. After all, neither of those technologies are comparable to what they were like only 3 years ago and our tech increases exponentially. In 4-5 years time, DLSS and FSR (and XeSS for that matter) will probably all be 100% perfected. The same is true for someone who buys a low-end GeForce for RT.

Now, don't get me wrong, I think that ray-tracing is a really cool thing but for the most part, our hardware isn't ready for it yet. Hell, the top-level of RT in CP2077 can bring a $1600 RTX 4090 to its knees! That tells me all that I really need to know about it, we're not even close to there yet so I don't factor it in to my GPU purchases yet.

 

[Avro Arrow]

I mentioned these things before of this subject:

1- The only way that the Steam Survey could be considered accurate is if it ran every month on every single endpoint.

Hell, they could be even more accurate than that. You could just register your system specs with the serial numbers of your CPU and GPU. Not only would it make the survey more accurate, it would also allow them to warn you if your system doesn't meet the minimum specs of a game that you want to buy. It would also make it impossible for anyone to hack your account because they won't have your serial numbers.

2- As you stated, many others have expressed the same experience, the survey doesnt show up for AMD gpus, but does for ngreedia and that has also happened to me. I'm not insinuating anything simply something curious that others have also experienced.

I did more than insinuate, I outright accused them of screwing up their own survey system.

3- As much as the survey "sucks" accuracy wise, it is the closest thing we have to a reliable source of info for these things.

There was a time that the closest that we could come to flying was to launch ourselves into the air using catapults. In this case, "close" doesn't necessarily mean "good".

I personally wonder if my system is counted since its running ChimeraOS, which might also show as holoiso, which is what the SteamDeck reports as.

Also, I wonder, are these systems automatically counted? Is my system counted even though its not a SteamDeck neither running proper holoiso?

I honestly couldn't tell you. You'd have to ask Valve themselves.

 

[Avro Arrow]

I can use quotes from other threads in any other thread but not in conversations why is this? example:

Yay! One of my all-time favourite posts!

 

[Avro Arrow]

If you want 100% accuracy, sure. If you're willing to accept a margin of error/confidence interval (like most surveys), you can have a much smaller sample size. E.g. you can get results within +-1%, 19 times out of 20, with 10k samples. If you're aiming for +-3%, you'd only need 1k samples, etc.

That assumes samples are being selected appropriately though, and sampling methodology is a big question mark for the Steam survey AFAIK.

Edit: Even if Steam attempted to survey 100% of client, it still wouldn't be 100% accurate. Not all clients would be online during the survey window, and some people would decline the survey.

The problem is that I have never been able to get the Steam survey to work on any of my all-AMD PCs and when I sent a ticket to Steam, they responded that they knew about the problem but didn't have a fix yet.

How can anyone have confidence in a company that says things like that?

 

[Order 66]

Yay! One of my all-time favourite posts!

and probably favorite movies as well.

 

[NeoMorpheus]

I've had multiple Steam HW surveys with AMD cards.

Well, that is expected, otherwise, AMDs numbers would be zero.

As far as being biased against certain devices, do you have any real evidence of that, or just a handful of anecdotes?

Just here, on this thread, you have 2 individuals that have never received one in over 10 years yet received them when not using an AMD gpu, yet you have.

Then, if you feel like gathering info, you can search the web and you will see many more stating the same for as many years.
Personally, i am curious about why this keeps happening.

And you and the others might want to read my posts again, because i never said that I was %100 sure of anything on this particular subject

But that would be me assuming that you are interested in the subject instead of posting in bad faith, which i think its the case here, by simply trying to discredit what i said because internet points i guess or at worse, losing sleep (like the ones that liked your post) just because i have dared saying that i like AMD’s gpus

 

[COLGeek]

This thread has run its course. Thank you to those who contributed.