confirmed: AMD's big Navi launch to disrupt 4K gaming

Derangel said:
That’s not how randomness works. Each year you have an x (amount of surveys they send) out of y (total number of Stream accounts) chance. The y increases every single year, giving you less of a chance as time goes on. Just because you don’t get it one year does not mean your chances to get it go up the next.
Click to expand...
I understand how it works, lol. If the roulette wheel lands on black 6 times in a row, it still has a 50/50 chance on the next go. My point was its odd that others have seen it relatively often and I have never seen it at all. If it's sufficiently random and the # polled is enough to be useful, you would think by now I'd have gotten one at some point. I dont actually think it's a conspiracy, just an oddity,a one of the reasons I personally don't put to much weight behind the survey, to much is unknown about the polling.
 
MangoSeed said:
And you think this means something because?
Click to expand...
I think it means I wasn't randomly selected.... in a large span that others where multiple times. It gives me reasonable doubt that their is sufficient randomness & sufficient #'s being collected to put much weight behind their results. Sure the #'s give a broad view of the landscape, but it's really hard to know with any sort of accuracy so I take those results with a big pinch of salt.
 
  • Like
Reactions: Meeho
like this
Ready4Dis said:
I think it means I wasn't randomly selected.... in a large span that others where multiple times. It gives me reasonable doubt that their is sufficient randomness & sufficient #'s being collected to put much weight behind their results. Sure the #'s give a broad view of the landscape, but it's really hard to know with any sort of accuracy so I take those results with a big pinch of salt.
Click to expand...

That isn't reasonable doubt. That is your mistaken assumptions about random sampling. People have assumptions about how random numbers distribute, and those assumptions are almost always wrong.

All kinds of apparently weird patterns can that can occur in random selection, at the individual sample level. I might be selected every year for 10 years in a row, while you are never selected during those same 10 years, and it could still be perfectly random.

It's the overall aggregate where the random sampled data smooths out, not on the individual samples.

Considering that Steam will largely be sampling different sets of people month-to-month, and that the data is relatively consistent month-to-month, indicates that Steam has a sufficiently large and sufficiently random sample, that it is indeed representative of their user base.

People complaining about the techniques here, are usually just unhappy with the result and trying to attack the source.

The Steam Survey data is by far the most accurate data we have on what HW is actually in the hands of people who play games.
 
Ready4Dis said:
Yeah, that's what I'm saying, 0 in over 7 years (can't remember exactly), and I have had 4 accounts (mine + 3 kids) for a few years now and 0 on all of them. Seems odd as others have seen it a few times before. Guess I have no reason to buy lotto tickets with my luck!
Click to expand...
-------

Steam Survey, it pops at least every couple of months, and sometimes monthly (which is fair given often systems I'm working with change) - I always say yes. My guess if someone asks Steam how they pull account names for Surveys they will tell.
 
singe_101 said:
Which games are straining a 1080 Ti or "old" cards?
Click to expand...

I mainly play big single player adventure games, but most of them can't stay anywhere near my 100hz 3440x1440 UW. Some examples like Jedi Fallen Order, AC: Origins, AC: Odyssey, mostly run in the mid 70's. I'll usually crank to Ultra everything and see where I end up with frame rate and adjust from there. I almost always use RTSS to cap at something just below my stable average. I'd love to be able to make a nice step up to something I can push that monitor fully with settings mostly cranked.

I have a 1080ti OC'ed as far as it will go and am definitely ready for an upgrade. Never came close with anything Turing offered, the 2080ti was the only thing that would have been an upgrade and prices were stupid for that miniscule of an upgrade. Hoping to be in the 3080 range with Ampere to be able to push that monitor and frame rate and not have to pay for the ti.
 
Smoked Brisket said:
Respectfully disagree. There is a reasonable price, however the masses arent going to know what that is and marketing can inflate value. It is up to people that are more knowledgeable than me to push back on these inflated prices. $1200 is not reasonable. $699 is reasonable, $799 is status quo, $1200 is a joke.
Click to expand...
Price per transistor only went down about 4% from Pascal to Turing, but the number of transistors went up about 58%. If you want a "reasonably priced" video card then you're going to be stuck with the same performance from 4 years ago and into the future unless the price per transistor scales down with the manufacturing process again.
 
Ready4Dis said:
I think it means I wasn't randomly selected.... in a large span that others where multiple times. It gives me reasonable doubt that their is sufficient randomness & sufficient #'s being collected to put much weight behind their results. Sure the #'s give a broad view of the landscape, but it's really hard to know with any sort of accuracy so I take those results with a big pinch of salt.
Click to expand...

It’s not really hard to know at all. The steam survey is a massive sample of the population so is likely extremely accurate. Basic statistics at work.

Any suspicion you have of steam numbers has nothing to do with math.
 
Armenius said:
Price per transistor only went down about 4% from Pascal to Turing, but the number of transistors went up about 58%. If you want a "reasonably priced" video card then you're going to be stuck with the same performance from 4 years ago and into the future unless the price per transistor scales down with the manufacturing process again.
Click to expand...

Not the only factor in cost. Nvidia and AMD are larger more efficient companies today than they were years ago. When products scale the cost generally goes down. Nvidia saw prices skyrocket during the mining craze and hoped(still hope) to keep prices at that level. With no real competition they thought enthusiasts would just have to eat it and smile until it seemed normal. Dont tell me that did not happen. I understand inflation not gouging. We see the truth every day with the likes of intel showing how much they were fleecing consumers when they slash prices on skews that are under siege by AMD. Before you whip out the charts, I respect that you know more than me on transistor cost. I contend that profits can be made even at reasonable prices, maybe just not 4th mansion and second beach house profits.
 
Snowdog said:
That isn't reasonable doubt. That is your mistaken assumptions about random sampling. People have assumptions about how random numbers distribute, and those assumptions are almost always wrong.

All kinds of apparently weird patterns can that can occur in random selection, at the individual sample level. I might be selected every year for 10 years in a row, while you are never selected during those same 10 years, and it could still be perfectly random.

It's the overall aggregate where the random sampled data smooths out, not on the individual samples.

Considering that Steam will largely be sampling different sets of people month-to-month, and that the data is relatively consistent month-to-month, indicates that Steam has a sufficiently large and sufficiently random sample, that it is indeed representative of their user base.

People complaining about the techniques here, are usually just unhappy with the result and trying to attack the source.

The Steam Survey data is by far the most accurate data we have on what HW is actually in the hands of people who play games.
Click to expand...
I have not once attacked the results, all I said is I take them with a pinch of salt not knowing about how they collect, who they ask, how many and other relevant information that is typically disclosed during polling. I even stated I'm sure there aggregate data is good enough to get a view of the landscape, but without more specifics nobody knows the actual accuracy. In understand that someone could be picked for every single survey at random, but if that happened would you question the randomness of the sampling? It's very unlikely and if your random sampling of getting the same person, maybe it's not as random as you thought. I am not saying their survey is useless and I'm sure their numbers are relatively close. If they are happy with the days they are receiving, then that's all that matters. But when you try to extrapolate the data and it's meaning without knowing it's accuracy or how it samples, it's ok in generalities, but you can't use it to be very specific (again, many unknowns).
My reasonable doubt statement that is if others are/were seeing it often, and it's not just one person who had said that, that means on average people where seeing it more than once. So if I haven't seen it at all, it leaves doubt in my mind that it's sufficiently random. I could be wrong, but without more information it's all I have to go from.
 
MangoSeed said:
It’s not really hard to know at all. The steam survey is a massive sample of the population so is likely extremely accurate. Basic statistics at work.

Any suspicion you have of steam numbers has nothing to do with math.
Click to expand...
Sure, it's massive... How many people get asked, how often, how are they chosen and what percentage accepts? Fill in those blanks and I'll agree with you. You state basic statistics, then you say any suspicion has nothing to do with math. Ummm...
You used the # "massive" in your statement to prove I'm not using math in my argument. Man, that's some complicated geotrigocalcometry you got going on, sorry I can't keep up with such number crunching. :)

PS. Don't be offended, I'm just saying that there #'s for the survey can be semi useful but not to put to much stock into them unless you know how they do their polling. I'm not suggesting that they sent close to reality, just that we don't know how accurate. Could be .1% could be 5%, we don't know for sure.
 
Ready4Dis said:
Sure, it's massive... How many people get asked, how often, how are they chosen and what percentage accepts? Fill in those blanks and I'll agree with you. You state basic statistics, then you say any suspicion has nothing to do with math. Ummm...
You used the # "massive" in your statement to prove I'm not using math in my argument. Man, that's some complicated geotrigocalcometry you got going on, sorry I can't keep up with such number crunching. :)

PS. Don't be offended, I'm just saying that there #'s for the survey can be semi useful but not to put to much stock into them unless you know how they do their polling. I'm not suggesting that they sent close to reality, just that we don't know how accurate. Could be .1% could be 5%, we don't know for sure.
Click to expand...

I think you’re missing the point. When you use a massive sample then all the other factors you mention become irrelevant. That’s what I mean by “basic statistics”.
 
Smoked Brisket said:
Not the only factor in cost. Nvidia and AMD are larger more efficient companies today than they were years ago. When products scale the cost generally goes down. Nvidia saw prices skyrocket during the mining craze and hoped(still hope) to keep prices at that level. With no real competition they thought enthusiasts would just have to eat it and smile until it seemed normal. Dont tell me that did not happen. I understand inflation not gouging. We see the truth every day with the likes of intel showing how much they were fleecing consumers when they slash prices on skews that are under siege by AMD. Before you whip out the charts, I respect that you know more than me on transistor cost. I contend that profits can be made even at reasonable prices, maybe just not 4th mansion and second beach house profits.
Click to expand...
I don't know about the economic strategies of these various tech companies, but at least for NVIDIA I would assume that they wanted to keep profit margins similar to keep up the level of R&D in all the sectors they've expanded to. You did make a good point that the masses are not going to understand pricing strategy and I agree that they generally don't care about what it looks like on the supply side. They just see the price go up on a product that has traditionally and reliably been within a certain pricing structure and get outraged without wanting or needing to know what cause the prices to increase. From my perspective I think the pricing of both NVIDIA's and AMD's products are reasonable when considering all the factors involved in bringing such a tech product to the market, especially one with limited competition.
 
MangoSeed said:
I think you’re missing the point. When you use a massive sample then all the other factors you mention become irrelevant. That’s what I mean by “basic statistics”.
Click to expand...

Wow, here we are again in another thread arguing about the validity of Steam stats? :D Well, Big Navi will put that to rest.
 
Armenius said:
I don't know about the economic strategies of these various tech companies, but at least for NVIDIA I would assume that they wanted to keep profit margins similar to keep up the level of R&D in all the sectors they've expanded to. You did make a good point that the masses are not going to understand pricing strategy and I agree that they generally don't care about what it looks like on the supply side. They just see the price go up on a product that has traditionally and reliably been within a certain pricing structure and get outraged without wanting or needing to know what cause the prices to increase. From my perspective I think the pricing of both NVIDIA's and AMD's products are reasonable when considering all the factors involved in bringing such a tech product to the market, especially one with limited competition.
Click to expand...

Except that their prices nearly doubled from one year to the next. And nope, mining did not to it. :)
 
ManofGod said:
Except that their prices nearly doubled from one year to the next. And nope, mining did not to it. :)
Click to expand...
GTX 1070 launch price = $379
RTX 2070 launch price = $499
Price difference = +31.66%

GTX 1080 launch price = $599
RTX 2080 launch price = $699
Price difference = +16.69%

GTX 1080 Ti launch price = $699
RTX 2080 Ti launch price = $999
Price difference = +42.92%
 
Armenius said:
GTX 1070 launch price = $379
RTX 2070 launch price = $499
Price difference = +31.66%

GTX 1080 launch price = $599
RTX 2080 launch price = $699
Price difference = +16.69%

GTX 1080 Ti launch price = $699
RTX 2080 Ti launch price = $999
Price difference = +42.92%
Click to expand...

Knowing what I know now, I would have easily opted for the RTX2070 at $500 over a "GTX2070" at $379 if such a thing would have been available.
 
ManofGod said:
2080Ti was at least $1199 at launch.
Click to expand...

Exactly. Sure there might have been one SKU that was $999, but it was rarely available and used poorly binned chip that didn't boost well at all. If you wanted a good boosting GPU, you had to pay in the $1199 range.

Auer said:
Knowing what I know now, I would have easily opted for the RTX2070 at $500 over a "GTX2070" at $379 if such a thing would have been available.
Click to expand...

I would say exactly the opposite. A $379 mid range card minus RTX would have been the go to card for many people.
 
Auer said:
Knowing what I know now, I would have easily opted for the RTX2070 at $500 over a "GTX2070" at $379 if such a thing would have been available.
Click to expand...
Huh.... you would rather pay over $100 extra for the same thing with a feature set that is basically unusable? I have exactly zero interest in ray tracing and I do not personally know a single person that does. If that feature set came with no performance hit then sure, I wouldn’t mind it, but I still wouldn’t pay extra for it.

I game at 1080p because I value a high frame rate over pristine visuals. I have a good 1440p monitor as well as a good 4K monitor, but I’m not willing to drop the cash to game at those resolutions comfortably (while still coming in at a lower frame rate then 1080p). Free sync / Gsync kind of levels this playing field but I wont have a solid opinion until my next system upgrade that’s coming soon.
 
Tengis said:
Huh.... you would rather pay over $100 extra for the same thing with a feature set that is basically unusable? I have exactly zero interest in ray tracing and I do not personally know a single person that does. If that feature set came with no performance hit then sure, I wouldn’t mind it, but I still wouldn’t pay extra for it.

I game at 1080p because I value a high frame rate over pristine visuals. I have a good 1440p monitor as well as a good 4K monitor, but I’m not willing to drop the cash to game at those resolutions comfortably (while still coming in at a lower frame rate then 1080p). Free sync / Gsync kind of levels this playing field but I wont have a solid opinion until my next system upgrade that’s coming soon.
Click to expand...

Different strokes :) Game on.
 
Ready4Dis said:
I understand how it works, lol. If the roulette wheel lands on black 6 times in a row, it still has a 50/50 chance on the next go. My point was its odd that others have seen it relatively often and I have never seen it at all. If it's sufficiently random and the # polled is enough to be useful, you would think by now I'd have gotten one at some point. I dont actually think it's a conspiracy, just an oddity,a one of the reasons I personally don't put to much weight behind the survey, to much is unknown about the polling.
Click to expand...
Actually a bet on Red or Black in Roulette only has a 47.37% chance, everybody forgetts abouts those pesky 0's.
 
Armenius said:
GTX 1070 launch price = $379
RTX 2070 launch price = $499
Price difference = +31.66%

GTX 1080 launch price = $599
RTX 2080 launch price = $699
Price difference = +16.69%

GTX 1080 Ti launch price = $699
RTX 2080 Ti launch price = $999
Price difference = +42.92%
Click to expand...
Mining didn't do it, Investors demands, and fabrication times did, nVidia and AMD both have access to limited fab time and there are only a limited number of facilities that are cabable of producing these GPU's on the new fab processes. As such they have to front load all their orders warehouse and distribute them over the course of their lifespan and as t hey don't want to be left holding a large number of obsolete chips they have to make sure they maximize everything they can get from the few they do produce. To do less would only enrage their investors which would cause them to get sued and their precisous CEO's fired. They have very complex formula's to determine that if they produce X number of chips to sell at Y price gets them to sell out by C date. so they maximize profits minimize costs and reduce the chances of being left holding unsellable stock in warehouses.
 
Ready4Dis said:
I have not once attacked the results, all I said is I take them with a pinch of salt not knowing about how they collect, who they ask, how many and other relevant information that is typically disclosed during polling. I even stated I'm sure there aggregate data is good enough to get a view of the landscape, but without more specifics nobody knows the actual accuracy. In understand that someone could be picked for every single survey at random, but if that happened would you question the randomness of the sampling? It's very unlikely and if your random sampling of getting the same person, maybe it's not as random as you thought. I am not saying their survey is useless and I'm sure their numbers are relatively close. If they are happy with the days they are receiving, then that's all that matters. But when you try to extrapolate the data and it's meaning without knowing it's accuracy or how it samples, it's ok in generalities, but you can't use it to be very specific (again, many unknowns).
My reasonable doubt statement that is if others are/were seeing it often, and it's not just one person who had said that, that means on average people where seeing it more than once. So if I haven't seen it at all, it leaves doubt in my mind that it's sufficiently random. I could be wrong, but without more information it's all I have to go from.
Click to expand...

Going on about "take them with a pinch of salt", and "reasonable doubt" is attacking the veracity of the results, based on nothing but your incorrect assumptions about statistics.

Just because you haven't been asked to participate, is NOT grounds for "reasonable doubt" in the survey.

You really should be applying more of that skepticism where it is really warranted. Like to WCCF Tech and similar rumor mongers, and in this case to companies hyping their own products. Now those are things which you need a pinch of salt and then some.
 
Derangel said:
That’s not how randomness works. Each year you have an x (amount of surveys they send) out of y (total number of Stream accounts) chance. The y increases every single year, giving you less of a chance as time goes on. Just because you don’t get it one year does not mean your chances to get it go up the next.
Click to expand...
It also means it should be highly improbable to be selected many times in a few years.

Snowdog said:
That isn't reasonable doubt. That is your mistaken assumptions about random sampling. People have assumptions about how random numbers distribute, and those assumptions are almost always wrong.

All kinds of apparently weird patterns can that can occur in random selection, at the individual sample level. I might be selected every year for 10 years in a row, while you are never selected during those same 10 years, and it could still be perfectly random.

It's the overall aggregate where the random sampled data smooths out, not on the individual samples.
Click to expand...
In a perfectly random sampling you should have massively more chance of not being selected ever than being selected 10 times. The fact that many people reported to have been selected many times indicates, but doesn't prove, that Steam's survey isn't sufficiently random.

Considering that Steam will largely be sampling different sets of people month-to-month, and that the data is relatively consistent month-to-month, indicates that Steam has a sufficiently large and sufficiently random sample, that it is indeed representative of their user base.
Click to expand...
The latter is just your assumption and the former would be true even with a biased selection.

People complaining about the techniques here, are usually just unhappy with the result and trying to attack the source.
Click to expand...
There's nothing to complain about since the techniques are unknown.

The Steam Survey data is by far the most accurate data we have on what HW is actually in the hands of people who play games.
Click to expand...
You have nothing to base that on. It is probably the largest data, but until it's transparent we know nothing about its accuracy.
 
Armenius said:
GTX 1070 launch price = $379
RTX 2070 launch price = $499
Price difference = +31.66%

GTX 1080 launch price = $599
RTX 2080 launch price = $699
Price difference = +16.69%

GTX 1080 Ti launch price = $699
RTX 2080 Ti launch price = $999
Price difference = +42.92%
Click to expand...

GTX 1070 was actually $350, but no cards were had at that price for a long time (closer to $380 as mentioned). RTX 2070 was $550.
 
ManofGod said:
2080Ti was at least $1199 at launch.
Click to expand...

Base model to base model, it is accurate. I bought a $769 (plus or minus $10 forget exactly) and the comparable 2080Ti was ~ $1299.

It's (lets round up even) about a 70% price increase when comparing the more expensive iterations of the cards. It's a lot no doubt, but it was not "double". Double = 100% increase.
 
Snowdog said:
Going on about "take them with a pinch of salt", and "reasonable doubt" is attacking the veracity of the results, based on nothing but your incorrect assumptions about statistics.

Just because you haven't been asked to participate, is NOT grounds for "reasonable doubt" in the survey.

You really should be applying more of that skepticism where it is really warranted. Like to WCCF Tech and similar rumor mongers, and in this case to companies hyping their own products. Now those are things which you need a pinch of salt and then some.
Click to expand...
Oh, I apply my skepticism everywhere. I was saying I wasn't attacking the results meaning I don't suggest they are biased or misleading, they are just results from a poll which I don't have the information on how it's conducted. Ergo, I question the results. The fact that I've seen 0 questions on 4 accounts is just an oddity, for better or for worse it makes me question the randomness. I even admit (and still agree) their results are probably good enough, but I'm not wrong saying that you don't know how accurate or what the margin of error is. I take anything wccf tech says with a bucket of salt and even stated I don't care what is said until I see real benchmarks, so not sure what you are even saying about this as I'm not discrediting steam and saying I believe rumors. I don't believe hyped and rumors, I believe real products and (multiple verified) 3rd party benchmarks. I also rely on my own experiences, if those differ to you then we can easily find ourselves in disagreement, but I'm not suggesting that they are misleading, just that without more information I don't just put my blind trust in someone or some company. If you want to believe their results are 100% accurate without knowing how they are collected, good on you for having that much trust in a company without any proof.
 
  • Like
Reactions: Mega6
like this
Meeho said:
It also means it should be highly improbable to be selected many times in a few years.


In a perfectly random sampling you should have massively more chance of not being selected ever than being selected 10 times. The fact that many people reported to have been selected many times indicates, but doesn't prove, that Steam's survey isn't sufficiently random.


The latter is just your assumption and the former would be true even with a biased selection.


There's nothing to complain about since the techniques are unknown.


You have nothing to base that on. It is probably the largest data, but until it's transparent we know nothing about its accuracy.
Click to expand...
Yay, someone gets it :). Yes, the fact that others got it a bunch and myself none doesn't prove anything, but it does leave some doubts on how random it truly is. This doesn't prove one way or the other, and the results probably (likely) are pretty close, but nobody here can say for sure. We can't prove it either way since we don't know enough about it.
 
Ready4Dis said:
Yay, someone gets it :). Yes, the fact that others got it a bunch and myself none doesn't prove anything, but it does leave some doubts on how random it truly is. This doesn't prove one way or the other, and the results probably (likely) are pretty close, but nobody here can say for sure. We can't prove it either way since we don't know enough about it.
Click to expand...

If you sample some accounts repeatedly over time you get cohort data which you don't get from only randomly sampling.
 
Lakados said:
Actually a bet on Red or Black in Roulette only has a 47.37% chance, everybody forgetts abouts those pesky 0's.
Click to expand...
Haven't ever played, not into gambling. Can't even win a steam survey, lol. But the point is still valid, the chances don't change based on past behaviour, but over time they should trend towards red/black being even.
 
Ready4Dis said:
Haven't ever played, not into gambling. Can't even win a steam survey, lol. But the point is still valid, the chances don't change based on past behaviour, but over time they should trend towards red/black being even.
Click to expand...

Not really. Each spin has a set percentage of landing on red, black, or 0 but that doesn’t mean it will always average out to that percentage. Assuming the exact same force and motion is used each time, 100 spins is not always going to result in the final numbers coming out to the normal percentage. Same with 1000 spins or a million. True random variables have zero logical pattern.
 
Auer said:
Knowing what I know now, I would have easily opted for the RTX2070 at $500 over a "GTX2070" at $379 if such a thing would have been available.
Click to expand...
What do you know that we don't? Keep in mind your talking about a RTX2070.
 
Thunderdolt said:
Funny title on the first link. AMD hasn't "disrupted" anything on the CPU side. They simply released a set of products which are somewhat faster than what was already on the market. If Big Navi is going to be "similarly disruptive," then that just means it's going to offer not quite enough performance for AAA 4K - just like the 2080 Ti that came out two years before it.

With AMD aiming to match 2 year old Nvidia performance, it sounds like they're still stuck in the rut of being the low-cost GPU maker rather than the high performance one.
Click to expand...

Lol hey its your opinion.

I can tell you that I have spent a lot of time in microcenter. Almost every single person in there ask about AMD. They dont sell Intel nearly as much and all thier shelves are stuffed with AMD chips and just a few Intel parts. But hey thats just one place but seems pretty disruptive to me.
 
Ready4Dis said:
Yeah, that's what I'm saying, 0 in over 7 years (can't remember exactly), and I have had 4 accounts (mine + 3 kids) for a few years now and 0 on all of them. Seems odd as others have seen it a few times before. Guess I have no reason to buy lotto tickets with my luck!
Click to expand...
Had about 4-5 of them since few months after Steam launched on one account..
 
You must log in or register to reply here.
Back
Top