RTX 4070 tops the sales chart at Mindfactory but it's not all good news for NVIDIA...

[kapone32]

The 7900 XT isn't a budget card (although it is the most value conscious of the current-generation batch, on both sides), but the 6700 XT is a last-gen midranger and the 6950 XT is on a fire sale to clear stock and cover for the 7800 XT's absence.

Last Gen Mid range is $600+ Canadian and that is not in anyway budget. The 6950XT is a funny card as the 50 cards do clock higher but were not really available. There is also the fact that As Rock is the newest GPU vendor and as such have since the start been selling their cards for a little but less but the Crypto bubble also created the crazy situation where I was able to buy a 6650XT for less than a 6600, only due to timing.

 

[Bomby569]

Mindfactory might not be representative industry-wide, but for the German market (guess even for the EU market).

sorry but no, maybe the Germans buy more AMD gpu's, but that's certainly not the case in all the EU, not even a close contest, if this was soccer they don't even play on the same league, just sort any store by most sold.

 

[kapone32]

sorry but no, maybe the Germans buy more AMD gpu's, but that's certainly not the case in all the EU, not even a close contest, if this was soccer they don't even play on the same league, just sort any store by most sold.

What Economy is bigger than Germany in the EU? I do agree that some Markets around the World do not list many AMD Cards.

 

[Bomby569]

What Economy is bigger than Germany in the EU? I do agree that some Markets around the World do not list many AMD Cards.

if you go to alternate or amazon.de i see the same thing, i think mindfactory must be a very weird store.
Just France and Italy together are biger then Germany.

And btw for the ones question steam surveys i keep seeing the 3060 as the most sold or amongst the most sold cards everywhere.

 

[Suspecto]

Mindfactory might not be representative industry-wide, but for the German market (guess even for the EU market). It's "extremely AMD-centric market" just shows that a lot of Germans think their money is better invested in a AMD card than in a Nvidia card, right now. Now why is that?

First, for raw rasterization performance AMD is right now just the better deal. Aslo AMD has right now pretty competitive deals while Nvidia thinks they can still milk the market with higher end cards (incl. Nvidia Premium tax). Second, German reviewers are more objective and give honest recommendations instead of turning their review into a sale advertisement. And Germans for the most part do their research before burning their money, esp. in the current economic conditions. The sale numbers show that people buy mostly in the sub €700 market, they buy what gives them the most rasterization performance, and from Nvidia there isn't much "good" to pick atm.

About the consistency of the Steam Hardware Surveys, they are pretty flawed and need to be taken with a big grain of salt. They don't reveal how they're conducted, it's like a "Black Box". The only reliable numbers on Steam are sale numbers & most played.

Steam Hardware Survey Reveals Bizarre Statistic for Most Popular Language In March

Valve's latest Steam Hardware Survey comes back with a strange statistic for what the most popular language was on the platform in March of 2023.

gamerant.com

The Steam Hardware Survey had some flaw causing cyber cafes to be over-counting users

Well this sure is interesting, Valve have announced some flaws in their Steam Hardware Survey that caused cyber cafes to over-count users.

www.gamingonlinux.com

Heck, they can't even include their current/peak player numbers on the store pages (you need to rely on 3rd party websites or use browser plugins like "Augmented Steam" to see them). They also don't include MetaCritic user rankings on the store pages. Guess both would hurt sales of dead/trash games.

Please, don't comment on issues you have no clue about.

Steam stats are pretty consistent, the fact you have found only 1 instance when they were wrong, well, in that one case, they weren't even wrong, just used wider pool than usual and included also cafes which should be a valid entry just like work, school pc, laptops are, just proves the point they are in fact valid and reliable, because they have been sharing data for over 10 years and showing CONSISTENT TRENDS confirming other independent data. Now compare the reliability vs politics polls and guess what, Steam is significantly more accurate than them.

I haven't seen AMD fanboys complaining when Steam showed the sudden changes of AMD market cpu share favouring AMD.

The fact is that Nvidia has been dominating the sales since the 9xx gen. It is not only represented by steam but also by the most popular benchmarks on the planet, when Nvidia releases highend gpus, the amount of entries for Nvidia is up to 10 or 20 times higher than for AMD, same with HW enthusiasts forums.

Deal with it.

And also, a 3060 was very popular card and people were often opting for it due to its "affordable price" and due to having more Vram than the TI, x70 and x80. It is simple as that and who watched the sale rankings on Amazon and other sites, absolutely no surprise. It was also heavily demanded by miners, again due to 12GB VRAM.

 

[kapone32]

Please, don't comment on issues you have no clue about.

Steam stats are pretty consistent, the fact you have found only 1 instance when they were wrong, well, in that one case, they weren't even wrong, just used wised pool than usual and included also cafes which should be a valid entry just like work, school pc, laptops are, just proves the point they are in fact valid and reliable, because they have been sharing data for over 10 years. Now compare the reliability vs politics polls and guess what, Steam is significantly more accurate than them.

I haven't seen AMD fanboys complaining when Steam showed the sudden changes of AMD market cpu share favouring AMD.

The fact is that Nvidia has been dominating the sales for since 9xx gen. It is represented by steam and also by the most popular benchmarks on the planet, when Nvidia releases highend gpus, the amount of entries for Nvidia is up to 10 or 20 times higher than for AMD, same with HW enthusiasts forums.

Deal with it.

I have 6 PCs in my house. Which one does Steam register?

 

[MarsM4N]

sorry but no, maybe the Germans buy more AMD gpu's, but that's certainly not the case in all the EU, not even a close contest, if this was soccer they don't even play on the same league, just sort any store by most sold.

Well, I am not aware of any other EU store putting out sale numbers. You got some sources? Btw. Germany is EU's biggest economy. And if AMD's/Nvidia's pricing is the same in other EU countries the market share could very well look the same. Btw. the sale numbers just show the market share right now (for 1 week), nothing more. When Nvidia brings out something more competitive in the sub €700 category, drop their prices or AMD sells out of their old gen stock it could quickly swing back into Nvidia's favour.

 

[Taisho]

Who is buying all those 8GB cards like the 3070, makes no sense.

I somewhat agree. Low-end 6600 XT 8GB models can be had for 260 euros or I can get a used Red Devil for 225 euros. Perfect for waiting out the still bad GPU market on a 24" 1080p monitor, before moving to 4k in a couple of years with a backlog of the most visually impressive, and hardware-demanding titles. 3070 is too expensive for a 1080p GPU and can't cover newer titles in 1440p.

 

[Vya Domus]

"can they make one that's worth it?"

Worth it from what perspective ? Navi31 is the top silicon that they've decided on manufacturing, the GCD is actually just 300 mm^2 so fairly moderate in size, they could have easily made a larger GCD to house more compute units to compete with AD102 while still having the cheaper product, they've decided not to, probably a good move. There is no point trying to compete with a halo product, Nvidia has the better marketing machine and would have still outsold them in that segment, so yes, not worth it.

As for the node advantage, I've heard a lot of speculation. Some say that N4 is just a minor node refresh and that they are functionally equal. Others say that AMD is at a disadvantage. I can't comment because I honestly don't know who's right in this debate.

4000 series chips appear to have a 10% transistor density advantage, obviously that does matter.

 

[Bomby569]

Well, I am not aware of any other EU store putting out sale numbers. You got some sources? Btw. Germany is EU's biggest economy. And if AMD's/Nvidia's pricing is the same in other EU countries the market share could very well look the same. Btw. the sale numbers just show the market share right now (for 1 week), nothing more. When Nvidia brings out something more competitive in the sub €700 category, drop their prices or AMD sells out of their old gen stock it could quickly swing back into Nvidia's favour.

Almost all sorts have sort by most sold. It's just as accurate as someone posting sale numbers.

 

[Dr. Dro]

Worth it from what perspective ? Navi31 is the top silicon that they've decided on manufacturing, the GCD is actually just 300 mm^2 so fairly moderate in size, they could have easily made a larger GCD to house more compute units to compete with AD102 while still having the cheaper product, they've decided not to, probably a good move. There is no point trying to compete with a halo product, Nvidia has the better marketing machine and would have still outsold them in that segment, so yes, not worth it.



4000 series chips appear to have a 10% transistor density advantage, obviously that does matter.

Performance/cost/power consumption balance. The 4070's scored the perfect balance between performance and power consumption, but the price is so nasty that it's just not worth it unless you don't have a high-end platform to lean it on to begin with (power supply, thermal constraints), that suddenly its premium outweighs the hassle and benefits of gutting your machine and upgrading it.

Navi 32 at $600, would it perform like the 4070 with equivalent sub-200W power consumption? Would it deliver equal RT performance to justify a similar cost? And to make it more interesting, would it consistently outperform the 6950 XT being sold at this price range? I personally doubt it. But as always, I will concede that this is my personal view and they may just surprise us all.

 

[Suspecto]

Worth it from what perspective ? Navi31 is the top silicon that they've decided on manufacturing, the GCD is actually just 300 mm^2 so fairly moderate in size, they could have easily made a larger GCD to house more compute units to compete with AD102 while still having the cheaper product, they've decided not to, probably a good move. There is no point trying to compete with a halo product, Nvidia has the better marketing machine and would have still outsold them in that segment, so yes, not worth it.



4000 series chips appear to have a 10% transistor density advantage, obviously that does matter.

You are talking about AMD, which decided to release Zen XT refresh for the premium with a huuuge 100mhz increase.

They could have not created anything remotely competitive without it consuming 600W+, melting under the full load and being a laughing stock. So they took the L. They even admitted it that in order to compete with Nvidia, the power consumption would be insane.

AMD is a company and like Nvidia, they also milk their customers the very moment they have a chance. They did it and tried it every damn time they got a chance in the CPU department. XT refresh, cutting the cpus support on older chipsets, 7xxx lineup and then 3D cache refresh months later, 7900x3D and 7950x3D with just one CCX having the cache, no stock coolers, premature end of the first gen thread ripper support and new socket, limited pcie 3.0 gpu support in laptops, misleading laptop naming scheme which should have introduced "clarity" etc..

If they could have, they would have done it. Easy as that. And like in the cpu department, having the halo product is IMPORTANT FOR THEM. Because it sells. And if they have the ability to introduce halo product to the market, they will do it. The 3D refreshes.

 

[evernessince]

Almost all sorts have sort by most sold. It's just as accurate as someone posting sale numbers.

You don't work in eCommerce eh.

Top Sellers or Best Selling sorts do not show you the straight up number of units sold. Typically they they look at a couple of factors with units sold being one of them. Other factors include frequency of sales (a bunch of grouped up sales for example will typically push an item up a Best Selling list), how new the item is, net sales, number of page views, how many people have wishlisted the item (assuming the website has that feature), date the item was added, or any other data the store might have handy.

Ultimately the purpose of the best selling list is to encourage sales when an item is at maximum popularity and not to show absolute number of products sold. I can tell you what a dumb "top sellers" list looks like because it's always what I recommend new shopify shop owners replace as that is the default for the platform. What ends up happening is that the same items get on the list and stay there the whole time. This is turn makes your website look stale to customers and has the opposite effect of what a best seller list is supposed to do. On top of that, directly reporting sales numbers is a bad idea because it gives your competitors data that they can use against you. Thanks to you they don't have to take a risk on buying certain products, they can just pick and choose your most popular products and add them to their own catalog which will in turn likely impact your own sales.

I can tell you for a fact that Amazon's top selling list isn't just a straight up list based on units sales based on two factors:

1) Lack of past top selling GPUs. If the list was just a dumb units sold list, 1080 Ti, 970, ect would 100% be on this list. At the very least this means Amazon has a date cutoff for that list wherein they no longer consider a sale for a product as qualifying towards being a top seller. Some websites separate all time best sellers and best sellers into separate pages as they can show a completely different picture. Depending on the industry either list can be useful or useless.

2) The fact that the list isn't entirely dominated by cheaper GPUs and accessories. Cheaper GPU and accessory unit sales vastly outstrip those of higher end GPUs. That the list isn't dominated by them means that at the very least the net revenue, ASP, or total sale value is significantly weighed when considering what qualifies to be on that list.

In short, Best selling lists are not a replacement for units sold figures and can be quite complicated.

 

[Vya Domus]

They could have not created anything remotely competitive without it consuming 600W+

Yes they could have, a GPU with let's say 25% more shaders would consume roughly 20% more power, still comparable to a 4090, it's not rocket science these things scale fairly linearly and it's easy to project what the numbers would have looked like.

melting under the full load

Let's not go there, things that are melting have been more of an Nvidia problem this generation.

 

[Wirko]

You don't work in eCommerce eh.

Top Sellers or Best Selling sorts do not show you the straight up number of units sold. Typically they they look at a couple of factors with units sold being one of them. Other factors include frequency of sales (a bunch of grouped up sales for example will typically push an item up a Best Selling list), how new the item is, net sales, number of page views, how many people have wishlisted the item (assuming the website has that features), date the item was added, or any other data the store might have handy.

Ultimately the purpose of the best selling list is to encourage sales when an item is at maximum popularity and not to show absolute number of products sold. I can tell you what a dumb "top sellers" list looks like because it's always what I recommend new shopify shop owners replace as that is the default for the platform. What ends up happening is that the same items get on the list and stay there the whole time. This is turn makes your website look stale to customers and has the opposite effect of what a best seller list is supposed to do. On top of that, directly reporting sales numbers is a bad idea because it gives your competitors data that they can use against you. Thanks to you they don't have to take a risk on buying certain products, they can just pick and choose your most popular products and add them to their own catalog which will in turn likely impact your own sales.

I can tell you for a fact that Amazon's top selling list isn't just a straight up list based on units sales based on two factors:

1) Lack of past top selling GPUs. If the list was just a dumb units sold list, 1080 Ti, 970, ect would 100% be on this list. At the very least this means Amazon has a date cutoff for that list wherein they no longer consider a sale for a product as qualifying towards being a top seller. Some websites separate all time best sellers and best sellers into separate pages as they can show a completely different picture. Depending on the industry either list can be useful or useless.

2) The fact that the list isn't entirely dominated by cheaper GPUs and accessories. Cheaper GPU and accessory unit sales vastly outstrip those of higher end GPUs. That the list isn't dominated by them means that at the very least the net revenue, ASP, or total sale value is significantly weighed when considering what qualifies to be on that list.

In short, Best selling lists are not a replacement for units sold figures and can be quite complicated.

Very well said. And we could go on. Do we actually have a guarantee from Amazon that their top sales lists are compiled automatically from hard sales data? They could as well be "we wish those would sell best" lists. Or possibly based on actual data plus "we wish this would sell more" factor added to each item. Every little pixel on their web pages is carefully thought out to promote what they want, so why wouldn't top sales lists serve the same goal?

On the other hand, Mindfactory's sales numbers are believable. They also reveal other important information that simple top-ten lists don't, like the fact that the first three cards were sold in almost equal quantities, while #4 is farther behind.

 

[DemonicRyzen666]

Steam is only one form of a PC game sale platform is not the god of all PC gamers many of are you make it out to be.

 

[Dr. Dro]

Steam is only one form of a PC game sale platform is not the god of all PC gamers many of are you make it out to be.

However, Steam is de facto the premier platform for PC, and an easy 90%+ of the market uses it, many exclusively so - and that's being generous to all other platforms combined.

Sure you have games on other launchers (mostly Epic, EA's and Ubisoft's nowadays, Bethesda discontinued theirs, Microsoft's... I guess it has some reach, but their distribution method is hostile to modding so most gamers opt for other platforms), the anti-DRM extremists who use nothing but GOG and physical media games which haven't shipped in a decade plus even if it means they have to live on crumbs (psst, don't tell them that account-based gating of software availability is also a form of DRM)... I'd argue Steam is very much the best way to come up with accurate metrics regarding hardware market share. Not that it is accurate by any means, you'd only have fully accurate data if each and every user was prompted to - and took the survey.

Let's not go there, things that are melting have been more of an Nvidia problem this generation.

PCI-SIG problem, really. The issue was mechanical in nature, people weren't plugging the connector all the way in until it clicked, in fear of damaging their expensive hardware.

AMD played for time in refusing to implement 12VHPWR this generation (remember: AMD is part of PCI-SIG and can probably be counted as one of its senior members so they definitely knew, and/or had a say in this connector's creation), but it's already water under the bridge, there will be improvements to the latch, cables and connectors going forward.

 

[DemonicRyzen666]

However, Steam is de facto the premier platform for PC, and an easy 90%+ of the market uses it, many exclusively so - and that's being generous to all other platforms combined.

No, it's not.
Steam is only 10% of the full worldwide PC market.
It's not 90% of the U.s.a market either.
Holding exclusivity doesn't mean "good", it means you're trying to monopolize the market to get customers.

 

[Bomby569]

You don't work in eCommerce eh.

Top Sellers or Best Selling sorts do not show you the straight up number of units sold. Typically they they look at a couple of factors with units sold being one of them. Other factors include frequency of sales (a bunch of grouped up sales for example will typically push an item up a Best Selling list), how new the item is, net sales, number of page views, how many people have wishlisted the item (assuming the website has that feature), date the item was added, or any other data the store might have handy.

Ultimately the purpose of the best selling list is to encourage sales when an item is at maximum popularity and not to show absolute number of products sold. I can tell you what a dumb "top sellers" list looks like because it's always what I recommend new shopify shop owners replace as that is the default for the platform. What ends up happening is that the same items get on the list and stay there the whole time. This is turn makes your website look stale to customers and has the opposite effect of what a best seller list is supposed to do. On top of that, directly reporting sales numbers is a bad idea because it gives your competitors data that they can use against you. Thanks to you they don't have to take a risk on buying certain products, they can just pick and choose your most popular products and add them to their own catalog which will in turn likely impact your own sales.

I can tell you for a fact that Amazon's top selling list isn't just a straight up list based on units sales based on two factors:

1) Lack of past top selling GPUs. If the list was just a dumb units sold list, 1080 Ti, 970, ect would 100% be on this list. At the very least this means Amazon has a date cutoff for that list wherein they no longer consider a sale for a product as qualifying towards being a top seller. Some websites separate all time best sellers and best sellers into separate pages as they can show a completely different picture. Depending on the industry either list can be useful or useless.

2) The fact that the list isn't entirely dominated by cheaper GPUs and accessories. Cheaper GPU and accessory unit sales vastly outstrip those of higher end GPUs. That the list isn't dominated by them means that at the very least the net revenue, ASP, or total sale value is significantly weighed when considering what qualifies to be on that list.

In short, Best selling lists are not a replacement for units sold figures and can be quite complicated.

i'm not talking about expert level data about sales, but as a general idea it works, and they are accurate for as much as i can tell, with hundreds of stores they pretty much all tell the same story, so does steam survey.
I think we don't need thin hats here.

 

[wolf]

Mindfactory has never been representative of anything industry-wide. It's market is for whatever reason, extremely AMD-centric.

The few times I've been there it's littered with AMD advertising on the sides, and more often than not had deals or bundles for AMD hardware running too. As retailers go, whomever is at the helm appears at face value to be pro AMD and pushes their products where possible. They may have been quite normal/neutral to start and just exacerbated by the vocal fans that have latched onto their shop for AMD, and as we know pretty much any advertising is good advertising, as long as people are actively talking about mind factory, it helps their business. As always, I'd rather hang my hat on statistics from a worldwide PC gaming platforms hardware surveys and glean them for trends and data points than on EU retailers sales charts.

 

[Bomby569]

Also why are the numbers all multiples of 5? that seems very weird and unbelievable. If they are so forthcoming with the data why not post the real numbers that can't be those rounded ones for sure.

 

[Dr. Dro]

No, it's not.
Steam is only 10% of the full worldwide PC market.
It's not 90% of the U.s.a market either.
Holding exclusivity doesn't mean "good", it means you're trying to monopolize the market to get customers.

LOL, if you account for all PCs including those that aren't at all used for gaming, yes, but it's more than safe to say that 9 in 10 gaming PCs have Steam installed

 

[Bomby569]

LOL, if you account for all PCs including those that aren't at all used for gaming, yes, but it's more than safe to say that 9 in 10 gaming PCs have Steam installed

if you're a PC gaming and don't have Steam installed (you may not use it much) you are a rare bird

 

[Redwoodz]

Please, don't comment on issues you have no clue about.

Steam stats are pretty consistent, the fact you have found only 1 instance when they were wrong, well, in that one case, they weren't even wrong, just used wider pool than usual and included also cafes which should be a valid entry just like work, school pc, laptops are, just proves the point they are in fact valid and reliable, because they have been sharing data for over 10 years and showing CONSISTENT TRENDS confirming other independent data. Now compare the reliability vs politics polls and guess what, Steam is significantly more accurate than them.

I haven't seen AMD fanboys complaining when Steam showed the sudden changes of AMD market cpu share favouring AMD.

The fact is that Nvidia has been dominating the sales since the 9xx gen. It is not only represented by steam but also by the most popular benchmarks on the planet, when Nvidia releases highend gpus, the amount of entries for Nvidia is up to 10 or 20 times higher than for AMD, same with HW enthusiasts forums.

Deal with it.

And also, a 3060 was very popular card and people were often opting for it due to its "affordable price" and due to having more Vram than the TI, x70 and x80. It is simple as that and who watched the sale rankings on Amazon and other sites, absolutely no surprise. It was also heavily demanded by miners, again due to 12GB VRAM.

Another case where you think you know but you don't.

90% of AMD gpu buyers can't even submit an UL(3DMark) benchmark because their Tesselation bias aginst AMD means users can't even submit a bench run on completely stock gpu's. You are supposed to be able to, but it never works right making your submission invalid. Now about those stats?

 

[oxrufiioxo]

Another case where you think you know but you don't.

90% of AMD gpu buyers can't even submit an UL(3DMark) benchmark because their Tesselation bias aginst AMD means users can't even submit a bench run on completely stock gpu's. You are supposed to be able to, but it never works right making your submission invalid. Now about those stats?

Never had that issue although it's been a while since an AMD gpu was even worth messing around with....

I scored 9 012 in Time Spy

AMD Ryzen 5 2600, AMD Radeon RX 5700 XT x 1, 16384 MB, 64-bit Windows 10}

www.3dmark.com

I scored 10 060 in Time Spy

AMD Ryzen 9 3900X, AMD Radeon RX 5700 XT x 1, 16384 MB, 64-bit Windows 10}

www.3dmark.com

No issues submitting scores with this Stock 5700XT on two different platforms. X470/X570