• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

4070 reviews thread

Page 12 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Ok, another sample of last 8 cards sold:

RX 7900 XT
RX 7900 XT
RTX 3060 12G
RX 7900 XT
RTX 3060 8G Ti
RTX 3060 8G Ti
RTX 3060 8G
RTX 4070 - BTW nice white Gigabyte card.

last orders.png

So again, now one 4070 in 8 sold cards.

EDIT: These are orders, I just noticed that the first two cards were ordered by one person...
 
Last edited:
If this was a good launch,all the top slots would be RTX4070 cards. Retailers are already reducing the price of RRP models in the UK.
 
Without historical data of what that rotating list of 8 most recent sales looks like during a GPU launch, we have no basis for determining how successful (or unsuccessful) this actually is. I'd say that if you just use those two snapshots you couldn't conclude it's all that widely demanded. But trying to draw that kind of conclusion from two data points is similarly worthless.

Monitor it every two hours for an entire week and we'd at least have something to work with. There may be some issues with sampling that way if the sales move far slower at certain times and some of the samples partially (or even wholly) overlap.
 
Without historical data of what that rotating list of 8 most recent sales looks like during a GPU launch, we have no basis for determining how successful (or unsuccessful) this actually is. I'd say that if you just use those two snapshots you couldn't conclude it's all that widely demanded. But trying to draw that kind of conclusion from two data points is similarly worthless.

Monitor it every two hours for an entire week and we'd at least have something to work with. There may be some issues with sampling that way if the sales move far slower at certain times and some of the samples partially (or even wholly) overlap.

Better off just using best seller list from Amazon, since it has massive scale. Though I don't think they have disclosed what window that works on (rotating, last X days...).

Plus of course the ever popular Steam HW survey once they sell enough to show up. 😀

Though I expect they are selling well enough. It's much like 4070 Ti. People dumped on it, it never sold out, yet in the best seller list it's show up well now, and it's already in the Steam HW Survey. I expect 4070 will echo this.
 
I don't know how frequently Amazon updates, but it's probably better than something like this. It's still one data point and I don't know if you can generalize Amazon results much. How much hardware do they actually sell? I've never bought computer hardware from them and their search for GPUs is utterly atrocious based on the few times I've tried to price check something on their site.

Steam HW survey continues to be bad for making any kind of claims about anything for the same reason it always has been. No one who posts numbers from it is even going to try to do any kind of comparative analysis either, which still won't be all that great for a variety of reasons.
 
Steam HW survey continues to be bad for making any kind of claims about anything for the same reason it always has been. No one who posts numbers from it is even going to try to do any kind of comparative analysis either, which still won't be all that great for a variety of reasons.

I haven't seen a solid argument against Steam survey results. Mostly nit picks by people that don't like the results. Steam results also correlate with what the major analysts say, and since the same people don't like that info either they nitpick about the analysts.

Some will then will happily trot out one retailer in Germany. I wonder why...

I think the Steam HW survey is a fair indication of the proportion of cards out there in gamers hands. It's not perfect, but it's the best thing we have.

For new generation parts, it shows 4090 in the lead, followed by 4070 Ti, trailed by 4080, with RDNA 3 cards not showing up yet. Which seems quite reasonable.
 
with RDNA 3 cards not showing up yet. Which seems quite reasonable.
You think that by now, ZERO RDNA3 cards have found their way into the hands of Steam gamers?

You consider that viewpoint "reasonable"?

Really now?

There's a reason that there's a strong contingent of us that believes that Steam, through whatever mechanism, undercounts AMD gamers in the survey.

Edit: I have not personally verified that RDNA3 cards haven't shown up yet.
 
Last edited:
You think that by now, ZERO RDNA3 cards have found their way into the hands of Steam gamers?

You consider that viewpoint "reasonable"?

Really now?

There's a reason that there's a strong contingent of us that believes that Steam, through whatever mechanism, undercounts AMD gamers in the survey.

Edit: I have not personally verified that RDNA3 cards haven't shown up yet.
A card doesn't show up as its own entry on the survey unless it crosses the 0.15% threshold. And incidentally the 6800 XT is at 0.15%. So it is obvious that RDNA3 representation is lesser than that, which is expected.

Stop spinning conspiracy theories regarding the Steam HW survey somehow undercounting AMD GPUs.
 
I can't see Nvidia really dropping the price without a lot of pushing by AMD which I can't see either. Sure they will sell less, but you got to remember Nivdia can sell every AI chip they make for 10* the price of a 4090 - so why would they use their precious silicon allocation on gaming gpu's?
My hope is the AI boom stabilises or moves onto another manufacturing node, and we get a mid gen round of super's at slightly more sensible prices.
 
There's a reason that there's a strong contingent of us that believes that Steam, through whatever mechanism, undercounts AMD gamers in the survey.

Yes, and this makes my point. There's no evidence at all that Steam undercounts AMD users, there isn't even a reasonable speculation how this would happen.

Plus as I said. the Steam proportions are also correlated by the major analysts GPU sales reports. So you also have to come up with another theory to deny this additional bit of collaborating reality.

Or you know, just accept the reality that NVidia significantly outsells AMD, and move on from there. Then we can all use Steam Survey as a glimpse into what cards are getting into gamer hands. It's not perfect, but its the best tool we have, and MUCH better than looking at whether a cards sells out or not, or a retailer snapshot of the moment.

New generation Steam results make sense, because NVidia always outsells AMD and it's cards pass the survey cutoff first. Plus the results show 4080 Trailing both 4090, and the 4070 ti. 4080 is widely seen as the worse NVidia mistake of this generation.

Once cards are over $1000, "value" thinking has left the thought process for most people. Most $1000+ buyers are NOT price sensitive, and are just looking for the "Best" card of the generation, so these are inclined to just go for the 4090, and "value" buyers are looking under $1000, so the $1300 4080 is in Limbo.
 
It's not that Nvidia outsells AMD, it's the margin you see on steam that's odd. JPR's discrete GPU market share reports have had AMD pretty consistently around 20% from the launch of Ampere and RDNA3 up until the last couple quarters and the launch of Ada. There's only 3 RDNA3 cards on the Steam survey though, the 6700XT, 6600XT and 6600, for a total of 1% market. The list goes down to 0.15%, so even if all the others add up, total % must be under 1.5% or so. Compare that to the Ampere desktop GPU (ie exclude laptop one) and the total is 28.21%. So JPR shows Nvidia selling 4x more dGPUs during the most recent generation than AMD, but Steam shows 20x-30x more current gen Nvidia cards being used on Steam.
It's not impossible that both could be accurate and people buying AMD GPUs just tend not to game much, but it is strange and makes people question it.
 
Ok, I observed a few more "last 8 card ordered" and here are the results, including the previous:

Number of 4070, number of AMD cards/number of Nvidia cards
2, 2/6
2, 2/6
0, 3/5
2, 2/6
3, 2/6
2, 1/7
0, 5/3
2, 0/8
0, 3/5

13 out of 72 cards ordered were 4070. That is: every sixth card ordered here is a 4070. It is probably the second most ordered card here, behind the 3060.

21 out of 72 cards were AMD. That is nearly 30%.

BTW if I count the cards in the first 24 most sold cards, 8 of them are AMD, that is a third, but that does not have an information about the number of the cards sold in it.
 
It's not that Nvidia outsells AMD, it's the margin you see on steam that's odd. JPR's discrete GPU market share reports have had AMD pretty consistently around 20% from the launch of Ampere and RDNA3 up until the last couple quarters and the launch of Ada. There's only 3 RDNA3 cards on the Steam survey though, the 6700XT, 6600XT and 6600, for a total of 1% market. The list goes down to 0.15%, so even if all the others add up, total % must be under 1.5% or so. Compare that to the Ampere desktop GPU (ie exclude laptop one) and the total is 28.21%. So JPR shows Nvidia selling 4x more dGPUs during the most recent generation than AMD, but Steam shows 20x-30x more current gen Nvidia cards being used on Steam.
It's not impossible that both could be accurate and people buying AMD GPUs just tend not to game much, but it is strange and makes people question it.
Exactly, but.

This is easy. The Steam survey defenders will say, "obviously, few people buy AMD cards to game on."

Now why is this so contentious. /s
 
On the topic of 4070 reviews, check out the comments on this rather positive review on Ars:


That's crazy from even a year ago when the moaning was about the pricing (of course) and about how AMD wasn't worth buying because bad drivers, installed one in 2007 and it sucked, etc. There was full defense force out for any decisions nvidia made. It's changed - a lot.

There was a comment I really agreed with - this could have been the 4060ti. It could have been priced at $450 and been just a boomer of seller. But rather than incorporating the market feedback on the 4080 and near miss 4080 12GB, they went for max value and priced it at $600. This may actually make them more money in the long term, so it's not a terrible play. I concede this.

But wow, I think the perception bleeding and reputational harm is actually something to behold. They could have completely reloaded their brand equity, I think, with a solid 4060ti product but just decided not to.
 
I don't know if you can generalize Amazon results much. How much hardware do they actually sell? I've never bought computer hardware from them and their search for GPUs is utterly atrocious based on the few times I've tried to price check something on their site.
You really think Amazon doesn't sell a ton of computer components?

Just due to how far Newegg has fallen, and the lack of a MicroCenter in my area, most of the parts I've bought in the last 5+ years have come from Amazon. It's true their search isn't that good, but if you already know what you're looking for, it makes little difference.
 
It's not that Nvidia outsells AMD, it's the margin you see on steam that's odd. JPR's discrete GPU market share reports have had AMD pretty consistently around 20% from the launch of Ampere and RDNA3 up until the last couple quarters and the launch of Ada. There's only 3 RDNA3 cards on the Steam survey though, the 6700XT, 6600XT and 6600, for a total of 1% market. The list goes down to 0.15%, so even if all the others add up, total % must be under 1.5% or so. Compare that to the Ampere desktop GPU (ie exclude laptop one) and the total is 28.21%. So JPR shows Nvidia selling 4x more dGPUs during the most recent generation than AMD, but Steam shows 20x-30x more current gen Nvidia cards being used on Steam.
It's not impossible that both could be accurate and people buying AMD GPUs just tend not to game much, but it is strange and makes people question it.

One thing that will tilt the Steam numbers a bit more in NVidia's favor over JPR discrete, is that Steam includes not just discrete GPU cards, but laptop GPUs where I think NVidia dominates even more than GPU card sales.
 
You really think Amazon doesn't sell a ton of computer components?

Just due to how far Newegg has fallen, and the lack of a MicroCenter in my area, most of the parts I've bought in the last 5+ years have come from Amazon. It's true their search isn't that good, but if you already know what you're looking for, it makes little difference.

+1

Since moving away from a major center with computer stores, ALL of my computer parts have since come from Amazon.ca. Simply the trust factor, free shipping and easy return policy compared with horror stories for the other options. I built my current PC from parts I started purchasing from Amazon in 2020 and only got the build running recently.
 
You really think Amazon doesn't sell a ton of computer components?

Just due to how far Newegg has fallen, and the lack of a MicroCenter in my area, most of the parts I've bought in the last 5+ years have come from Amazon. It's true their search isn't that good, but if you already know what you're looking for, it makes little difference.
I generally avoid Amazon as they are a competitor for all online businesses, including mine, but for years I used NewEgg for research and Amazon for buying. Returns are so much easier with Amazon it's not even close.

Nowadays I will try to buy at Best Buy for in person returns but despite my misgivings about giving Amazon business I still purchase computer items their semi-regularly. It's likely equal to my NewEgg purchases.

But if someone said they were the No.1 retailer for damn near any consumer goods, including $600 4070's, I wouldn't be able to summon my shocked Pikachu face.
 
I generally avoid Amazon as they are a competitor for all online businesses, including mine, but for years I used NewEgg for research and Amazon for buying. Returns are so much easier with Amazon it's not even close.

Nowadays I will try to buy at Best Buy for in person returns but despite my misgivings about giving Amazon business I still purchase computer items their semi-regularly. It's likely equal to my NewEgg purchases.

But if someone said they were the No.1 retailer for damn near any consumer goods, including $600 4070's, I wouldn't be able to summon my shocked Pikachu face.

When I lived in a major center, practically ALL of my computer parts were from local shops. I really didn't see the point of mail order at all. Prices were not that different and returns were much easier.

EX: I bought a HDD, had it fail while transferring a couple of TB of media files. Drove to the shop, got a refund in under 10 minutes, and a new different brand HDD... It would have been so much hassle to deal with RMA, shipping, waiting if it was purchased online. Plus on top of that I was supporting local business.
 
One thing that will tilt the Steam numbers a bit more in NVidia's favor over JPR discrete, is that Steam includes not just discrete GPU cards, but laptop GPUs where I think NVidia dominates even more than GPU card sales.
That's why I didn't include them. There are separate entries for 3060, 3050 Ti, 3050, 3070 Ti and 3070 laptop GPUs. I'd assume there's one for 3080 and 3080 Ti mobile as well, but they are below the cutoff. These are just discrete GPU numbers.
 
Back
Top