AMD RX 400 series (Polaris) MegaThread! FAQ & Resources

[jaymc]

Think this about sum's up what we know about Vega...

Except of coarse for the new information I recently got that were all living in a bubble belonging to a 5 year old Alien.

http://www.digitaltrends.com/computing/amd-vega-three-versions-navi-gpus-2019/

 

[jaymc]

Four Vega GPU's to be rolled out... Vega 10, Dual Vega 10, Vega 11 and Vega 20.
Dual Vega 10 eh... sounds serious.

The upcoming RX 490 is rumored to be the first AMD graphics card to feature the next generation Vega 10 architecture. (dash of salt with that please..well see)

http://www.scienceworldreport.com/articles/48905/20161003/amd-radeon-rx-490-release-date-specs-price-will-rx-490-outperform-nvidia-gpu-gtx-1080.htm

An what bout this, is this pure speculation or what, where are they getting this info..?
http://videocardz.com/amd/radeon-rx-400/radeon-rx-490

 

[Robert Cook]

Think this about sum's up what we know about Vega...

Except of coarse for the new information I recently got that were all living in a bubble belonging to a 5 year old Alien.

http://www.digitaltrends.com/computing/amd-vega-three-versions-navi-gpus-2019/

the fact that vega 10 is going after the 1080 makes a lot more sense then it going after the titan, as spec wise it did not seem possible, however, vega 11being larger than vega 10 makes little sense as that goes against the current Nvidia and AMD naming scheme. perhaps vega 11 is really Vega ten which will go after the titan? because 6144 stream processors should stack up much better than 4096. this rumor mill stuff seems more logical than previous sources.

 

[renz496]

An what bout this, is this pure speculation or what, where are they getting this info..?
http://videocardz.com/amd/radeon-rx-400/radeon-rx-490

pure speculation. they try to predict how the chip/card going to end up based on the rumor they heard.

 

[Warmacblu]

For some reason people hold things against AMD but let nVidia slide. i.e. If AMD had pulled the 970 3.5+0.5 VRAM thing or selling a cut down GPU and calling it the same name as the full one(1060 3gb vs 6gb), there would be picketing outside AMD's headquarters.

The 3.5+0.5 VRAM issue did not slide. A class action lawsuit was filed against nVIDIA and they settled by paying eligible consumers the difference.

 

[Robert Cook]

For some reason people hold things against AMD but let nVidia slide. i.e. If AMD had pulled the 970 3.5+0.5 VRAM thing or selling a cut down GPU and calling it the same name as the full one(1060 3gb vs 6gb), there would be picketing outside AMD's headquarters.

The 3.5+0.5 VRAM issue did not slide. A class action lawsuit was filed against nVIDIA and they settled by paying eligible consumers the difference.

The sad thing is, I know many people with 970s who are too lazy to bother getting their $20 ( or is it $30?) frankly if you look at how many they sold, and how many people are actually going to claim the compensation, they got off decently well.

on a lighter side, here is this

Spoiler

https://www.youtube.com/watch?v=BdwUsalwBJ8

 

[TechyInAZ]

Honestly I can see why people won't both getting $20 back. It's only $20, not that much money.

 

[Martell1977]

For some reason people hold things against AMD but let nVidia slide. i.e. If AMD had pulled the 970 3.5+0.5 VRAM thing or selling a cut down GPU and calling it the same name as the full one(1060 3gb vs 6gb), there would be picketing outside AMD's headquarters.

The 3.5+0.5 VRAM issue did not slide. A class action lawsuit was filed against nVIDIA and they settled by paying eligible consumers the difference.

It did in that people didn't really rip on them the same as they do AMD for lessor things. Plus, a small cash refund doesn't make it honest. nVidia doesn't get punished by the people as much as AMD does. i.e. there are still people talking about the power draw issue of the 480, even though it was fixed a couple of days later.

 

[James Mason]

It did in that people didn't really rip on them the same as they do AMD for lessor things. Plus, a small cash refund doesn't make it honest. nVidia doesn't get punished by the people as much as AMD does. i.e. there are still people talking about the power draw issue of the 480, even though it was fixed a couple of days later.

And there are still people talking about the 3.5 gb vram issue.

But AMD hasn't gotten sued (yet) because they fixed it, but depending on hardware damage it may have caused, we'll see.

 

[Rogue Leader]

It did in that people didn't really rip on them the same as they do AMD for lessor things. Plus, a small cash refund doesn't make it honest. nVidia doesn't get punished by the people as much as AMD does. i.e. there are still people talking about the power draw issue of the 480, even though it was fixed a couple of days later.

And there are still people talking about the 3.5 gb vram issue.

But AMD hasn't gotten sued (yet) because they fixed it, but depending on hardware damage it may have caused, we'll see.

All of those damage claims I took with a grain of salt (and there were not many). Tom's and others came out to say that it was unlikely it would cause damage. It all seemed too convenient, and very likely it was people with other system problems blaming it on the GPU, or just trolls. Meanwhile I ran one on day 1 on an OLD board that if anything was gonna get wrecked it would be that, and it had no problems or damage.

 

[TJ Hooker]

All of those damage claims I took with a grain of salt (and there were not many). Tom's and others came out to say that it was unlikely it would cause damage. It all seemed too convenient, and very likely it was people with other system problems blaming it on the GPU, or just trolls. Meanwhile I ran one on day 1 on an OLD board that if anything was gonna get wrecked it would be that, and it had no problems or damage.

Yeah, only one example springs to mind for me, and I think that was some guy running 3-4 card crossfire on some janky mining rig with riser cards on an old mobo.

 

[renz496]

It did in that people didn't really rip on them the same as they do AMD for lessor things. Plus, a small cash refund doesn't make it honest. nVidia doesn't get punished by the people as much as AMD does. i.e. there are still people talking about the power draw issue of the 480, even though it was fixed a couple of days later.

And there are still people talking about the 3.5 gb vram issue.

But AMD hasn't gotten sued (yet) because they fixed it, but depending on hardware damage it may have caused, we'll see.

and many people that did not own the card talk like they know how the card actually work in real games.

 

[BurgerandChips66]

I saw so many stories about the power draw issues 'damaging motherboards' but every time I did a bit of digging, ALL the stories were based on the same old dodgy Reddit thread. It was all boiling down to the same single event!
As far as I know it has never been proved that this particular incident was genuine. The guy in question claimed his PCIe slot was fried after a long gaming session. Many of us know there other reasons a PCIe slot can fail....
The last I saw he was using an alternate slot on the same motherboard with no other issues. Other than that, in terms of 'fried motherboards' I didn't see anything other than people recycling the same old dubious story time and time again.

That's the problem with the internet today.

People rehash and repackage the same old 'facts' until they become viral and part of a widespread mass epidemic based on apparently irrefutable evidence.

 

[-Fran-]

Personally, AMD and nVidia were on the wrong for both episodes.

When you market a Video card that should perform at X level (bear with the simplicity) and at some point users discovers that under certain and repeatable circumstances it performs at X/10, you have effectively deceived your customers by not fully disclosing it. And when you're supposed to be "within spec", and you're actually not, people is totally within their right to call them out.

Remember guys, there are no "nice" Companies out there; they are all demons trying to fight for your hard earned dollar. Do not let them pass any fault.

Cheers!

 

[cdrkf]

@yuka I think that's a bit unfair. I'm a design engineer working in the medical industry, and the companies I work with want to make a great product. I can assure you engineers at Intel, AMD and nvidia want the same thing.

Sure the companies want to make a profit- however they don't set out to try and cheat anyone. The out of specs power draw would have been a surprise to AMD I think, what I think happened was they tuned the card right up to the power limit in house, and didn't account for enough variability between components resulting in what happened. It could even be related to how the power draw was tested in house compared to the tech sites.

The 3.5gb vram issue on nvidia... that was possibly more dodgy- I think some manager in marketing decided it was better not to mention the unusual configuration. Again I doubt the technical teams set out to hoodwink anyone...

 

[jaymc]

@yuka The ol "caveat emptor" eh... always good advice.

@cdrkf I agree the engineers making the product want to make a good product, "they" don't set out to cheat anyone... Pretty sure that falls under the marketing departments jurisdiction 😛

 

[-Fran-]

Unfair or not, it's the reality of things. The only way consumers have to tell companies "don't do that anymore" is with their purchase decisions. Sure, Companies have a lot of internal colors in them, but what materializes into a product is what counts at the end of day. I can't just make a purchase decision based on "Jonny Doe at the Sells Department is a really good chap".

You can say, though, companies are better or worse than others on how they react to blunders like those. You have response times, the type of resolution to the issue and if there was any legal action required for them to acknowledge the blunder. Just keep in mind they are still fighting for your hard earned cash and you must protect your money. As creepy as it might sound, it's the first and most important line of defense you have against the bad apples in the corporate world: make sure they make things "right".

And yes, most of the "evilness" in Companies concentrate in their Marketing departments; I wholeheartedly agree with that 😛

Cheers!

EDIT: Typo.

 

[James Mason]

Also, all hard drive companies are little guilty of the same crime as Nvidia for the 3.5 ram issue, since every single hard drive and SSD has less space than it says it does on the label.

 

[-Fran-]

Hard drive manufacturers have always abused the "base 10 vs base 2" when displaying their hard disk sizes. That in turned moved onto SSDs (I think?) with the added "yeah, we're reserving some for cache and endurance".

The "base 10" logic is: we have X millions of bytes -> base 10 conversion to Kilo/Mega/Giga/Tera -> lost bytes compared to a base 2 conversion. That is to say: 10.000 bytes are 10K Bytes for HDD manufacturers, where in PC storage terms is 9.765625K Bytes. At least, that is how I remember it.

Cheers!

 

[Rogue Leader]

Also, all hard drive companies are little guilty of the same crime as Nvidia for the 3.5 ram issue, since every single hard drive and SSD has less space than it says it does on the label.

Hard drive manufacturers have always abused the "base 10 vs base 2" when displaying their hard disk sizes. That in turned moved onto SSDs (I think?) with the added "yeah, we're reserving some for cache and endurance".

The "base 10" logic is: we have X millions of bytes -> base 10 conversion to Kilo/Mega/Giga/Tera -> lost bytes compared to a base 2 conversion. That is to say: 10.000 bytes are 10K Bytes for HDD manufacturers, where in PC storage terms is 9.765625K Bytes. At least, that is how I remember it.

Cheers!

Correct, divide 10000 bytes by 1024 and you get your (correct) result above, while in HD manufacturer terms you have 10kb.

I agree its not the people building and designing these things that come up with these ways to deceive people into thinking they are getting more than they are. I also guarantee you there were many engineers, etc, at AMD that when the power issue broke were REALLY angry with themselves letting the product out like that.

Having worked for companies that sell various consumer and industrial products most of my career, it is completely true. Marketing takes what you have and spins it as high of a spin as it can go. Then sometimes they are brought down to a more reasonable level by management who says "ok we can't totally lie" but that check/balance doesn't always work as the pressure/desire to beat the "other guys" becomes overwhelming.

 

[Sakkura]

Also, all hard drive companies are little guilty of the same crime as Nvidia for the 3.5 ram issue, since every single hard drive and SSD has less space than it says it does on the label.

In decimal units, HDDs have exactly what they advertise, and SSDs have what they advertise plus a chunk more for overprovisioning. For example, a 250GB Samsung 850 Evo actually has 274.9GB of NAND flash.

The only problem is the transition from binary to decimal wasn't communicated clearly and was definitely influenced by the "let's get a bigger number" motive.

Nvidia's 3.5 issue was more egregious, but IMO not so much regarding the memory - the really icky thing was the number of ROPs. At least with the memory it was a slightly debatable thing because there technically was 4GB of memory there, and technically all the 256 bits of memory interface was there as well. The ROP count, on the other hand, was just plain wrong.

 

[TMTOWTSAC]

HD manufacturers also advertised the size of their drives unformatted. Which is fair, considering different file systems use differing amounts of drive space, but it was more complexity than the typical computer user was familiar with back in the day. I think the CRT monitor advertising with viewable/non-viewable area was worse, but again, a larger tube still made the viewable area flatter even if less total area was displayed so there's some justification.

Getting off topic though, we could discuss marketing deception forever and never catch up to new marketing schemes.

 

[jaymc]

I remember been on the phone with seagate must be ten or fifteen years ago, over that exact thing.... I was eventually told that Microsoft and the hard drive manufacturer's define the MB in different terms....
As far as I can recall it was over a 9 gb drive.... You would think that by now they would have started defining it in the same terms. An I taught I was stubborn.

 

[Turb0Yoda]

Yeah, 1024 v 1000 MB... For a 40 gig HDD, it should be 40960 MBs. Multiples of 1024

 

[TJ Hooker]

Yeah, 1024 v 1000 MB... For a 40 gig HDD, it should be 40960 MBs. Multiples of 1024

One megabyte, as measured by a computer, would 2^20=1,048,576 bytes, AKA one mebibyte. 40 gigabytes, as measured by a computer (AKA 40 gibibytes), would be 42,949,672,960 bytes, or ~42,950 megabytes (using the SI definition of mega as 10^6).