Is It The 1080 TI The Best GPU Ever?

[dirtyferret]

It seems to favor new installations of Steam. I only had it once on my old PC, within 5 years. I installed Steam on Debian, got the survey. Then recently I got another PC and installed Steam, and then immediately got the survey before I could install a game.

I've done it a few times and have had STEAM installed on my PCs for years.

Not sure why people 1) fail to understand what the STEAM survey results show and 2) use it justify their opinion for or against something. The survey simply shows the most common hardware parts across computers at the time of the survey. Doesn't mean anything about popularity or what is better or worse.

 

[lilhasselhoffer]

I've done it a few times and have had STEAM installed on my PCs for years.

Not sure why people 1) fail to understand what the STEAM survey results show and 2) use it justify their opinion for or against something. The survey simply shows the most common hardware parts across computers at the time of the survey. Doesn't mean anything about popularity or what is better or worse.

While I can see where your second point comes from, I disagree with your first.

Let me take a step back...and state what I get from the steam survey. It's performed voluntarily, it's a snap-shot of what is running, and it's not representative of any particular group. As such, what it represents is a balance of the average affluence of its users and their ability to project said affluence to usable buying power decisions. The "quick" take-away is therefore not what the best thing is...but the best balance of relative performance to absolute cost and market availability.


That's a huge mouth-full....so let's boil down to layman. It's a survey that categorizes what users are willing to spend for relative performance. You take input hardware costs, output display sizes...and interpolate performance. What you get is something that's a softer conclusion...such as:
1) Minimum expected performance levels...like seeing someone run a 1060ti and 4k resolution isn't really a thing given it'd be at 12 FPS (a number pulled from my rear).
2) Best value builds. 1920x1080 with a 4070 makes little sense...based upon chosen builds
3) Relative trade-offs. Going from a 1060 to a 1660 was great...but a 1060 to a 3070 everything else even wasn't really a thing.
4) Forced generations. We're well into the 4xxx series...but the 1660 and 3060 are still huge hitters...seems like they either have staying power...or the "new features" aren't ready to drive sales yet


As such the survey absolutely shows popularity, can give a soft representation of value, and shows what is "best that people can afford." That last one assumes a bit...but it's how companies like Nvidia and AMD can put price points to relative performance based upon real data. While it is by no means a 1:1 representation, there's good reason to conclude that the 3060 is a great performer at whatever its price point is...just based on how many are active. It pains me to say this...but sometimes the data is clear. I think the 3060 is overpriced for its relative performance...but the market states its the best current option. This is from someone whose most popular GPU is about half as represented as the 3060...but I'd still say that the 3060 is a better value to performance ratio than most other options on the market.

 

[dirtyferret]

and it's not representative of any particular group.

statistics is difficult for most people who have no basic understanding of it but this opinion is 100% wrong making the rest of post...pointless

 

[freeagent]

That is a pretty tight split between Yes and No.. nice. I was expecting more from team yes and no though

 

[Zyll Goliat]

Latest video of Hardware Unboxed about 1080TI

 

[P4-630]

My MSI RTX 2070 Super Trio X is....

 

[AVATARAT]

So the question is simple Is It The 1080 TI really The Best Card Ever if we put following parameters as the most important:

Best GPU ever is Vega 56 at 300w with tweaked memory timings.

 

[xtreemchaos]

For muchness yes and staying power it still a great card today. i cannot think of any other card "yet" that will pinch its crown.

 

[lilhasselhoffer]

statistics is difficult for most people who have no basic understanding of it but this opinion is 100% wrong making the rest of post...pointless

It's funny when you have no response...then say something stupid.

Statistics...taken from a group that is defined by voluntary inclusion, and therefore has no reasonable data to work with (by design of an anonymous data collection) is kind of funny. That said...your immediate statement that this is useless is...stupid.



I do not need to know who sells the most pickup trucks to know that if I want to sell a pickup I want to start in the US. I don't have to know what the best setup for a computer is to know what most people have decided to spend their money on. I also laugh at the people who think not having 100% of the data makes it inconsequential...because there's an entire division of laborers out there that take these partial relationships and define our reality based on them....because things like the performance of the 4060 ti were entirely defined as a minimum price point for a minimum performance level.

I do appreciate the grade school logic of not having all of the data making conclusions less valid...but the fun bit of reality is the engineers who die on perfect knowledge usually die when a good solution has already presented itself multiple times and was resoundingly ignored because it wasn't a perfect solution. The good solution from the Steam survey is that if you are to build a system that most people will target for you need:
1) 16 GB of RAM
2) 6 cores
3) 2.3 GHz
4) An RTX 3060 or better
The thing is...this represents the best by virtue of concensus and actual purchasing...you are welcome to deny reality. That said, your denial is useless...because even things like the topic 1080 ti were never "the best." They were an option....that with the right set of requirements you could pretend was the best. Me...I define best by best I can afford and tolerate...but that doesn't invalidate the obvious patterns and trends of a survey devoid of strong bias by virtue of a truly huge sample size. Though I do see your whining about obvious bias being present because we don't have a perfectly balanced random sampling as...well...let's say the kid in the back of a statistics class will always hiss venom that truly unbiased data comes from an infinite pool of personnel....despite the impracticality of that.



Positive note though...you've read sampling size. You've read bias. You've seen that I know about hidden bias. It would take someone a grand total of 20 seconds to understand that practical statistics and classroom statistics is an argument I've had more than once...and that the classroom BS is something I rail against because an 80% solution today is always better than a 95% solution in ten months when you've spent thousands of dollars just getting a big enough sampling to matter. If this was about medical efficacy it'd matter...but buying a PC is not something that has an infinite rainbow of performance and the need for statistical exactitude...despite the part of me that grinds upon saying that. Of course...your entire point was steam survey says nothing about being the best...so you've already bastardized the point to get to this argument...so returned disrespect for disrespect seems fair.

 

[dirtyferret]

It's funny when you have no response...then say something stupid.

and then states

Statistics...taken from a group that is defined by voluntary inclusion, and therefore has no reasonable data to work with

now that is funny!

You don't under stand the parameters of the survey. You don't understand how basic data collection works for a survey. But you know the summery of the survey is incorrect? That is why your posts are pointless. You are no different than the guys with no medical degrees that are all the sudden experts on airborne disease control when a virus they don't like comes around. Two weeks later the same guys with no law degrees are experts on the judicial system in every state and municipality when a court case headline they don't like comes across their path.

It's a survey of hardware & software across PCs & Laptops that have the most popular digital video game distribution system on them. No more, no less. It doesn't mean anything is more popular or better. Just because you are butt hurt or feel inferior from a result doesn't make your hissy fits factual.

I freaking hate having to waste my time to post this and make you actually feel it's a debate when you are 100% wrong but

The good solution from the Steam survey is that if you are to build a system that most people will target for you need:

No! No! No! No its not a good solution, no it's not a system building survey. No it should not be used to target hardware by a consumer. Here is the survey verbatim from STEAM

Steam conducts a monthly survey to collect data about what kinds of computer hardware and software our customers are using. Participation in the survey is optional, and anonymous. The information gathered is incredibly helpful to us as we make decisions about what kinds of technology investments to make and products to offer.

Nowhere do they state the survey should be used for system building. They don't even state the survey should be helpful to the anyone but them. It's nice they post it but its for curiosity sake only for the consumer.

 

[mxthunder]

That was true after the 8800 GT launched, yes. But the 8800 GTX existed for a year before then, and it was a huge jump over the previous generation in both performance and technical aspects. It also remained a bit faster than the 8800 GT (if marginally), namely at higher resolutions, and had 50% more VRAM. I'd say the 8800 GTX still earns a spot in the conversation as possibly one of the best (I'd certainly say it does more so than the GTX 1080 Ti at least). It's just that the 8800 GT was, all things considered, so much better than anything else nVidia ever released so nothing else is even close. It's really unfathomable how unrealistic the 8800 GT was... yet it actually happened despite it.

Imagine a hypothetical RTX 4090 LE that has ~95% of the performance of the standard RTX 4090 at 1440p and below, and ~90% at 4K, and perhaps 18 GB VRAM instead of 24 GB, uses less power, and ultimately costs ~$550. The market would lose its collective sanity. That sounds unbelievable today. Yet that's the equivalent of what the 8800 GT was in late 2007. If something like that is not the greatest of all time, I don't know what is. The GTX 1080 Ti accomplishment, by comparison, was "accessible to a smaller portion of the buying market, has a higher asking price but a bit more performance that slightly exceeds that price difference, and it is still serviceable years later but at a resolution below what it was capable of at launch due to a stagnating market in recent years". And for doing that, it's called that the greatest of all time? Expectations have slipped. None of this is to say the GTX 1080 Ti was bad. To the contrary, it was great, and Pascal as a whole was. But the greatest of all time? I don't know. There's just too many obvious better contenders in my mind.

Man I totally forgot about how awesome the 8800GT was!
I remember the GTX460 being a runner up back in the day too. Great price/perf and many compared it to the 8800GT at the time.

Its hard to imagine your 4090LE scenario today... back then 8800GT was a totally different die than 8800GTX, and a cut down one too.
We went from 90nm, 128 cuda cores on a 384 bit bus to 65nm, 112 cuda core part with a 256 bit bus that was almost as fast.
That would be like 4080 being $550 and being 95% as fast as 4090. I doubt even if they put AD103 on a smaller node with faster clocks it could bridge that gap.

 

[LifeOnMars]

I remember my crossfire config of HD 4850s being pretty damn fast back in good old 2008. Especially with the pencil mod. They lasted well into the early 201X's with my golden sample Q6600

 

[overclockedamd]

I voted yes for the 1080ti my only second guess would have been the 8800GT but with the problems of having to literally bake that series of cards to reflow them every so often I bumped it down a notch.

 

[Carillon]

To me it is the 290, it can still play modern games, at very low settings, below 60 fps, but still playable, and is much older than the 1080ti.
And it, or the 390, had some chips that could be unlocked to the X SKU. Also very slow memory that a tinkerer could swap for much faster chips.

 

[oxrufiioxo]

Man I totally forgot about how awesome the 8800GT was!
I remember the GTX460 being a runner up back in the day too. Great price/perf and many compared it to the 8800GT at the time.

Its hard to imagine your 4090LE scenario today... back then 8800GT was a totally different die than 8800GTX, and a cut down one too.
We went from 90nm, 128 cuda cores on a 384 bit bus to 65nm, 112 cuda core part with a 256 bit bus that was almost as fast.
That would be like 4080 being $550 and being 95% as fast as 4090. I doubt even if they put AD103 on a smaller node with faster clocks it could bridge that gap.

Complete dreamland scenario these days it would be the same price as a 4070 non super and nearly twice as fast to repeat the 8800GT although the gpu market was drastically different in those days even 1080p was a struggle at that time in modern games with most gpus. I was rocking a 1600x1200 display and couldn't get 60fps in every game with an 8800GT now high end gpus are expected to game at 4k with reasonably high settings.

I still think the 1080ti is over rated but I will say they were extremely easy to sell even 4-5 years after release for a pretty decent sum so that's gotta count for something although my Titan Xp sold for twice what they sold for on fleabay.