Latest Steam survey: RTX 3070 is the month's top performer, AMD edges closer to 30% CPU...

[HardReset]

And for that Radeon 6000 availability question. Using MaxSmarties' logic, TSMC's 7nm manufacturing tech is vaporvare because there is no capacity available for around 6 months. Why? Because they already have all capacity booked for future. But because there is no capacity, whole factory is essentially vaporware.

Now for AMD Radeon 6000 -series. There are still many undelivered RX6000-series orders that were made on launch day. What happens to availability? There is no visible availability until all orders on queue are shipped.

"Vaporware"...

 

[MaxSmarties]

I won't waste any time to answer to a fanboy.
It is useless.

Ryzen 5600X and 5800X are widely available at MSRP worldwide.
You are just spreading BS to defend AMD as usual.

 

[Geralt]

I find it depends on the game, if it’s a fast paced game then I typically would prefer RT off. But then again control is fast paced and that’s better with it on. But the magic in that game is DLSS2.0, death stranding has it too and it looks better than native. I’m not making that up, several tech outlets have reported the same thing. Alex Battaglia from Digital Foundry described it as objectively better with DLSS on.

You are however incorrect. RT is for now, I’ve played thousands of hours with RT on, mostly in minecraft but it’s awesome and it works and I don’t understand why people online (usually who don’t own an RTX card) go around saying it hurts performance too much and that it’s for the future. I’m getting like 80fps on minecaft, that is more than fine!

I do find it amusing reading all these comments from people saying things like “RTX hurts performance to much to be playable” or “you need a 3090 for RT to be worth it”. I’m sorry but these people are wrong. Good RT experiences are available right now. People who claim they can’t tell the difference need an eye test. The difference between RT on and off is bigger than the difference between low and ultra settings in most games that has it.

RT is absolutely a “now” tech. Even the consoles have it.

Sorry, but that's not true in 4K. The hit is massive. I have played in 4K only for the last years. I should have specified this first. My mistake.

 

[HardReset]

I run out of arguments as usual

As expected.

Ryzen 5600X and 5800X are widely available at MSRP worldwide.
You are just spreading BS to defend AMD as usual.

5600X is old out on many countries. Not still widely available worldwide but situation is improving. 5900X is still very short supply.

 

[m3tavision]

Steam Survey is not an up-to-date survey...!

I have a LAN room, with several systems and each has their own Steam account... I have not been offered a Survey on any of them, in over 3 years....

But over the Holidays, I log on my buddy's laptop to show him something.... and it asked me to take a survey... (when I am not even at my own computers).

 

[Bobbydpue]

I know quite a few people with 30xx cards at this point, many pre-ordered and received delays but most who wanted one have now got one. However I haven’t even laid eyes on a Radeon 6000 part. The drivers for the 6000 series are widely reported as being truly awful and make games like AC Valhalla unplayable etc and nothing appears to be happening to fix them.

That’s because it’s pretty much vapourware. Radeon exists purely for AMD to sell as branding for its console solutions to Sony and MS who pay far far more money than AMD would ever make with Radeon as a stand-alone solution.

I was able to buy a 6700XT, but returned it before I ever opened it after seeing the benchmarks and how much it should have cost. I bought it directly from a retailer, but it was still priced $400 more than it should have been.

 

[neeyik]

Steam Survey is not an up-to-date survey...!

It's up-to-date with regards to the fact that the survey is rolled out every month, but like all surveys, it's a sample of a population.

The latter, in this case, is the entire database of Steam user accounts; the sample size is harder to define, and for two reasons. Firstly, we have no idea how many sample points Valve uses to determine the statistics shown in the survey: it could be a thousand or ten thousand, or anything. Secondly, the survey requires a 'sample' to actively respond, so even if Valve attempts to garner ten thousand samples for the statistics (which would be more than enough, even for the size of the user database), there's no guarantee that every account selected will complete the survey.

But it's all we have. Despite numerous game and software developers/publishers sampling hardware and system specifications on a routine basis, the public never gets to see that data. The next best thing we have is the 3DMark database, but that has the same issues as Valve's. It's updated more frequently (it's a continuous process) but again, it requires active participants to download the software, run it, and all it to upload the result to the database.

Unfortunately UL doesn't offer enough temporal statistics, for my liking at least. For example, one can bring up a particular graphics card such as the RTX 3070 and you'll be shown a Popularity Rank graph:

NVIDIA GeForce RTX 3070 Review

NVIDIA GeForce RTX 3070 Graphics Card review with benchmark scores. See how it compares with other popular models.

benchmarks.ul.com

But what exactly does UL mean by Popularity Rank? It seems to be based on the % of that week's 3DMark entries coming from that particular GPU. One can also list of all of the GPUs recorded in the database and rank them by popularity:

Best Graphics Cards - April 2024

See the best Graphics Cards ranked by performance.

benchmarks.ul.com

But again, that seems to be based on that month's % of entries. An RTX 2060 is down in 15th place, but there's absolutely going to be more of those installed in active PCs around the world than, say, RTX 3090s (which is 2nd).

So for all its flaws, the Steam Hardware Survey is the best measure of the distribution of CPUs, GPUs, system configurations, etc that we have.

 

[Neatfeatguy]

I was able to buy a 6700XT, but returned it before I ever opened it after seeing the benchmarks and how much it should have cost. I bought it directly from a retailer, but it was still priced $400 more than it should have been.

Micro Center has 6700XT (if you can actually find any in stock if they ever ship in) priced up so high from MSRP.

1 card (no stock obviously) is priced at MSRP currently according to their website.
1 card is within MSRP, but still $90 over (not unreasonable)
All other cards are $200-350 over MSRP.

The 6800 they are priced starting at just shy of $1k.
The 6800XT cards are all around $1200.
6900XT are all around $1500.

GPU prices are a joke. I'll just keep gaming on my 980Ti until she dies.

 

[HardReset]

So for all its flaws, the Steam Hardware Survey is the best measure of the distribution of CPUs, GPUs, system configurations, etc that we have.

Agreed. But problem is: most people don't understand that best is relative term. Even if something is relatively best, it may still be crap in absolute terms.

And Steam survey is crap.

 

[Neatfeatguy]

Agreed. But problem is: most people don't understand that best is relative term. Even if something is relatively best, it may still be crap in absolute terms.

And Steam is crap.

Fixed it for you.

 

[m3tavision]

It's up-to-date with regards to the fact that the survey is rolled out every month, but like all surveys, it's a sample of a population.

The latter, in this case, is the entire database of Steam user accounts; the sample size is harder to define, and for two reasons. Firstly, we have no idea how many sample points Valve uses to determine the statistics shown in the survey: it could be a thousand or ten thousand, or anything. Secondly, the survey requires a 'sample' to actively respond, so even if Valve attempts to garner ten thousand samples for the statistics (which would be more than enough, even for the size of the user database), there's no guarantee that every account selected will complete the survey.

But it's all we have. Despite numerous game and software developers/publishers sampling hardware and system specifications on a routine basis, the public never gets to see that data. The next best thing we have is the 3DMark database, but that has the same issues as Valve's. It's updated more frequently (it's a continuous process) but again, it requires active participants to download the software, run it, and all it to upload the result to the database.

Unfortunately UL doesn't offer enough temporal statistics, for my liking at least. For example, one can bring up a particular graphics card such as the RTX 3070 and you'll be shown a Popularity Rank graph:

NVIDIA GeForce RTX 3070 Review

NVIDIA GeForce RTX 3070 Graphics Card review with benchmark scores. See how it compares with other popular models.

benchmarks.ul.com

But what exactly does UL mean by Popularity Rank? It seems to be based on the % of that week's 3DMark entries coming from that particular GPU. One can also list of all of the GPUs recorded in the database and rank them by popularity:

Best Graphics Cards - April 2024

See the best Graphics Cards ranked by performance.

benchmarks.ul.com

But again, that seems to be based on that month's % of entries. An RTX 2060 is down in 15th place, but there's absolutely going to be more of those installed in active PCs around the world than, say, RTX 3090s (which is 2nd).

So for all its flaws, the Steam Hardware Survey is the best measure of the distribution of CPUs, GPUs, system configurations, etc that we have.

Yeah, I remember an interview with Gabe years back. You do a go job of explaining it, but again new data's sample size, is the issue.

Many have suggested Valve dump the old data (from the survey results), and start new. Not weighing any of the old outdated hardware data, that has not ben in use in over 4 years from the survey... to weigh newer hardware at it's proper value.

 

[Geralt]

I'm surprised we're still on 16gb RAM. I bought my mine in 2014 for $90. 7 years later and they're the same price range and no gaming PC needs more than 16GB That has been the standard for 10 years now? The move from 4gb to 8gb and 2gb to 4gb was so much quicker.

It depends on the game. I am playing now RDR2 and it is sucking 18-20 GB of RAM and around 8 of VRAM. If the game sees that you have more, it will use more to its advantage. Obviously, I am playing in 4K with all maxed out.

 

[Geralt]

Micro Center has 6700XT (if you can actually find any in stock if they ever ship in) priced up so high from MSRP.

1 card (no stock obviously) is priced at MSRP currently according to their website.
1 card is within MSRP, but still $90 over (not unreasonable)
All other cards are $200-350 over MSRP.

The 6800 they are priced starting at just shy of $1k.
The 6800XT cards are all around $1200.
6900XT are all around $1500.

GPU prices are a joke. I'll just keep gaming on my 980Ti until she dies.

Here, in Russia, my 6900 XT cost me 2000 bucks