AMD R9 series artefacts

mildante

Limp Gawd
Joined
Sep 24, 2008
Messages
252
Hello, how many of these R9 cards do have the artefacts? From the reading on the internet it looks like that almost all of them. Does it depend on the model and manufacturer? I would like to get some R9 270X or 280 from MSI with the silent Twin Frozer cooler but am afraid that it will produce artefacts and I will have to return it multiple times which is not exactly how I would like to spend my time. So just asking how big is the probability I will get a faulty one.
 
Hello, how many of these R9 cards do have the artefacts? From the reading on the internet it looks like that almost all of them. Does it depend on the model and manufacturer? I would like to get some R9 270X or 280 from MSI with the silent Twin Frozer cooler but am afraid that it will produce artefacts and I will have to return it multiple times which is not exactly how I would like to spend my time. So just asking how big is the probability I will get a faulty one.

All of them? lol Is this the latest FUD campaign?
 
After mining with 4 280x's and currently using a 7990 I have 0 artifacts.

Also If you do get a card with artifacts, you send it back for a replacement. Doesn't matter if its AMD or Nvidia.
 
Ok, so which model should I get if I want the card to be as silent as possible in both idle and load. MSI R9 270X Hawk or MSI R9 280X Gaming 3G would be a good buy?
 
Ok, so which model should I get if I want the card to be as silent as possible in both idle and load. MSI R9 270X Hawk or MSI R9 280X Gaming 3G would be a good buy?

Depends on the performance you want. The 270x and 280x are 2 totally different chips. one has 2gb one has 3.

If money isnt a thing to you, snag a 280x (although I would stay away from the normal MSI ones and the Asus 280x since both have fan issues).

You also could snag a good used 7970 (which is a 280x) or 280x here at the hardforum.
 
Thank you. Hmm, by the 'normal MSI' models you mean for example MSI R9 280X GAMING 3G? I thought the fans on these are pretty good. Ok so if I decide for a 280X so which manufacturer and model exactly should I get? As I say - I prefer if the card is as silent as possible both in idle and load.
 
Hello, how many of these R9 cards do have the artefacts? From the reading on the internet it looks like that almost all of them. Does it depend on the model and manufacturer? I would like to get some R9 270X or 280 from MSI with the silent Twin Frozer cooler but am afraid that it will produce artefacts and I will have to return it multiple times which is not exactly how I would like to spend my time. So just asking how big is the probability I will get a faulty one.

What evidence do you have of this? Links please.
I read a lot and havent found this to be common.

Are the ones you found manually overclocked with too low voltage?
 
Was going to say I have 5x 280x and 2x 290s and I have 0 artifacts on my systems unless I am doing major OCing which I expect then when I OC them.
 
ditto, never had artifacts on my 7970, 290pro, or 290 vapor-x unless i overclocks the tits out of it.
 
With the refresh of the refresh of the refresh, AMD is pushing these cards too hard.
7870's weren't meant to run at 270X speeds. Same goes for the 280 and 280x. Then AIB's get their hands on them and start making "Uber" editions where they push even higher making them even more unstable (See: Sapphire Toxic cards).

As a result there's a lot of artifacting. You can see it plain as day in customer reviews (even though they usually tend to be unreliable).
But hey, you get what you pay for.
 
The silicon matures, and yields increase as they refresh, so there is no reason why they cant push them.

Personally, ive seen no credible evidence that this artifacting problem you are speaking of is "widesperead". Ive owned top-tier (uber cards as you call them) for the past couple generations and havent had a single artifacting problem.

The Toxic cards are widely concidered the absolute best of the best. This would not be the case if they did not perform.
 
The Toxic cards are widely concidered the absolute best of the best. This would not be the case if they did not perform.
https://www.sapphireforum.com/showt...-Sapphire-Toxic-Problem!-Artifacts!-New-VBios

Last I checked they gave up on vbioses and (to their credit) started replacing them with 290 Tri-X's.

"widesperead"
Even if it were a real problem, which it is, how would it ever be proven?
By its nature, it will always sound anecdotal. Unless some reviewer decides to buy a few hundred R9 cards to test it.

All I'm saying is, compare reviews of some Nvidia cards -- EVGA is a good example -- to reviews of various 270x/280x cards.
To make it easy, just CTRL+F the word "artifact" on their review pages and see how many individual results you can catch. At some point you have to admit it's not a coincidence.
 
Last edited:
The Toxic is a highly Binned Chip with a custom cooling solution and high-grade componants.

It has nothing to do with the vbios other than it had its own, as does the vapor-x, and the tri-x.

it goes like this:

Reference - Self explainatory

Tri-x: Reference PCB and componants with the Tri-X (dual-x on the last Gen) cooler and slightly OCed. custom bios.

Vapor-X: binned chip, plus tri-x cooler, plus vapor chamber. Custom PCB with additonal VRM cooling and high grade componants, mid range OC. custom bios.

Toxic (not out yet for 290/x): Heavily binned chip with same as vapor-x. Highly OCed. custom bios.
 
It just proves even the highest binned AMD GPU's aren't capable of the clocks that manufacturers are using.
That's evidenced by the widespread artifacting. If you're lucky, you get a golden GPU which is stable @ stock clocks.

Many of the 280X are running 1050~1100 MHz factory clocks, when stock 7970 is 950 MHz.
 
again, ive seen no evidence of "widespread" artifacting.

A few threads of people with bad cards does not constitude "widespread"

The silicon has matured after 2.5 years, i see no reason why it shouldnt be pushed to those levels. I would be surprised if they didnt after that long.

As i currently own a vapor-x, i can tell you its a beast of a card and have had zero issues with it.

Same with my Dual-X on last gen.
 
again, ive seen no evidence of "widespread" artifacting.
Average GPU failure rate is around 3% to 5% in the first 12 months.
Even if AMD doubled their failure rates, it's still less than 10% in 12 months.

Would you even notice if there was a widespread problem? 5% more people complaining about failures isn't noticable, but it still means the failure rate is twice as high.
I'm just saying, you can't prove whether it's real or not unless someone releases official failure numbers for each series.

As a person who has spent an inordinate amount of time looking into the issue, I can tell you without a doubt that the R9 series overall, Hawaii excluded, has stability problems.
Not a doubt in my mind.
 
Average GPU failure rate is around 3% to 5% in the first 12 months.
Even if AMD doubled their failure rates, it's still less than 10% in 12 months.

Would you even notice if there was a widespread problem? 5% more people complaining about failures isn't noticable, but it still means the failure rate is twice as high.
I'm just saying, you can't prove whether it's real or not unless someone releases official failure numbers for each series.

As a person who has spent an inordinate amount of time looking into the issue, I can tell you without a doubt that the R9 series overall, Hawaii excluded, has stability problems.
Not a doubt in my mind.

i suppose i probably wouldn't notice as 9 time out of 10 i would chalk it up to the vocal minority. People tend not to post when things are working correclty. Ive also notice a large number of people who are "doing it wrong", ie unstable systems, trying to OC the bejesus out of the memory, dirty installs, etc. I saw one thread where some guy was complaining about black screens, and found out several post later he had oced his gpu memory to 1700. well duh...

If i start to see it show up in the professional sites/publications, than i would tend to notice.

I can really only relay my own experience, and anecdotal information from others.
 
Last edited:
I'll ask this again...

All of them? lol Is this the latest FUD campaign?

It just proves even the highest binned AMD GPU's aren't capable of the clocks that manufacturers are using.
That's evidenced by the widespread artifacting. If you're lucky, you get a golden GPU which is stable @ stock clocks.

Many of the 280X are running 1050~1100 MHz factory clocks, when stock 7970 is 950 MHz.

FUD much? Where's your proof you claim. Even your anecdotal evidence holds no water. The "stock" 7970 was released years ago and O/C'd like a bear right out of the gate.

These are the kind of threads and posts we get when nVidia can't win on perf and still want to justify their higher prices. Anyone who wants to pay more for nVidia for whatever their reasons is up to them. I don't care. People passing FUD though to try and justify it is wrong and shouldn't go unchallenged.
 
FUD much? Where's your proof you claim.
This discussion pops up a few times a week on most hardware boards. I think the fact that we have to constantly ask "Does the R9 series have high failure rates?" is indicative of a problem. Keep in mind, it usually focuses on the 280X which is where I put all of my efforts.
It's incredibly easy to dismiss an issue as "Nvidia fanboy FUD". Just like the driver problems, too... Right? I suppose the black screen issue with the 290/290X is also FUD?

As for the 7970:
http://linustechtips.com/main/topic...ts-french-but-i-translated-nearly-everything/

When people start saying the sky is falling over Nvidia failure rates, then we'll be on a level playing field. AMD has higher failure rates than Nvidia, that's a fact.
If you need more info, do what I suggested previously... research. Nobody has compiled the data so you're going to have to dig. You'll figure it out eventually after reading through thousands of artifacting problems with the 7970 / GHz / 280X. Even then, it's still anecdotal I suppose.

For people who are capable of rational thought -- it's more than enough evidence. IF YOU BUY A 280X, specifically: You're taking a risk. I would also say the rest of the R9 series is a risk, to a lesser degree.
But then again that's always been true of AMD cards. Always a risk.
 
Last edited:
Wow, you're 2% more likely to get failures on cards that are used in heavy mining vs. cards that aren't used for mining at all? Based on data from an unnamed French online retailer? That translates to "AMD cards are always a risk"?
 
This discussion pops up a few times a week on most hardware boards. I think the fact that we have to constantly ask "Does the R9 series have high failure rates?" is indicative of a problem. Keep in mind, it usually focuses on the 280X which is where I put all of my efforts.
It's incredibly easy to dismiss an issue as "Nvidia fanboy FUD". Just like the driver problems, too... Right? I suppose the black screen issue with the 290/290X is also FUD?

As for the 7970:
http://linustechtips.com/main/topic...ts-french-but-i-translated-nearly-everything/

When people start saying the sky is falling over Nvidia failure rates, then we'll be on a level playing field. AMD has higher failure rates than Nvidia, that's a fact.
If you need more info, do what I suggested previously... research. Nobody has compiled the data so you're going to have to dig. You'll figure it out eventually after reading through thousands of artifacting problems with the 7970 / GHz / 280X. Even then, it's still anecdotal I suppose.

For people who are capable of rational thought -- it's more than enough evidence. IF YOU BUY A 280X, specifically: You're taking a risk. I would also say the rest of the R9 series is a risk, to a lesser degree.
But then again that's always been true of AMD cards. Always a risk.

I've seen this article and it was a red flag. It's never been verified or quantified. First, even if we take it at face value (which an anonymous report from a single unnamed etailer doesn't deserve automatic acceptance) , There is no info on what the failures were. Were they GPU failures or failures of other PCB components. The Sapphire 7870, for example, was caused by the use of caps that were from a non certified vendor which AMD had said would cause these issues. Sapphire admitted this and offered anyone who had the model (7870 dual X IIRC) a no questions asked exchange for anyone who wanted one. That one is documented. No others are, though.

You will occasionally get things like this with any product. We've saw a driver update from nVidia for the 590, for example. There were also issues with the 570. Both of these were VRM failures with reference cards. Not aftermarket models where the AIB was at fault. That's always been true of any product. There's always a risk.

Let's also add "Bumpgate" as a verified failure due to the manufacturer. Took a class action lawsuit to settle that one.
 
:D
Wow, you're 2% more likely to get failures on cards that are used in heavy mining vs. cards that aren't used for mining at all? Based on data from an unnamed French online retailer? That translates to "AMD cards are always a risk"?

LoL...had me laughing. You can always trust an unnamed French online retailer!
 
Newegg has 14 GTX 770's @ 5 star rating.
Not a single 280X has 5 stars.

280X:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127759
8 reports of artifacting out of 100 reviews.
Also suffers from fan failures (no surprise it's Twin Frozr).

www.newegg.com/Product/Product.aspx?Item=N82E16814121803
Over 100 reports of artifacting.
VBios update released in January.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814150690
9 reports of artifacting out of 100 reviews.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814202044
8 reports of artifacting out of 21 reviews.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814125475
Crashing/BSOD -- Has since been replaced with REV2.

GTX 770:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127741
4 reports of artifacting out of 148 reviews.
No reports of fan failures.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814121770
1 report of artifacting out of 161 reviews.

www.newegg.com/Product/Product.aspx?Item=N82E16814125463
1 report of artifacting out of 136 reviews.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814130921
1 report of artifacting out of 238 reviews.

More light reading:
http://www.tomshardware.com/answers/id-1973952/asus-280x-dc2t-severe-artifacting.html
http://www.techpowerup.com/forums/threads/asus-directcu-ii-top-r9-280x-artifacting.199964/
http://rog.asus.com/forum/showthread.php?43057-R9-280x-DirectCU-II-Top-Artifacts!!
http://www.reddit.com/r/buildapc/comments/1sx2xj/reliability_of_asus_r9_280x/
http://www.reddit.com/r/dogemining/comments/1z82se/asus_r9_280x_artifacts/
https://www.sapphireforum.com/showt...-Sapphire-Toxic-Problem!-Artifacts!-New-VBios
http://linustechtips.com/main/topic/84273-asus-r9-280x-artifacts-new/
http://www.mmo-champion.com/threads/1410299-Asus-R9-280x-artifacts
http://www.overclock.net/t/1442325/asus-280x-directcuii-top-problems-with-battlefield-4-artifacting
http://forum.overclock3d.net/showthread.php?t=65483
http://www.gpforums.co.nz/threads/487123-ATI-R9-280x-Screen-Artifacts
https://forum-en.msi.com/index.php?topic=181347.0
http://forum.giga-byte.co.uk/index.php?topic=14011.0
http://enbseries.enbdev.com/forum/viewtopic.php?f=23&t=3183
http://forums.hardwarezone.com.sg/hardware-clinic-2/sapphire-r9-280x-artifacting-4571929.html
 
Last edited:
280X:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127759
8 reports of artifacting out of 100 reviews.
Also suffers from fan failures (no surprise it's Twin Frozr).

www.newegg.com/Product/Product.aspx?Item=N82E16814121803
Over 100 reports of artifacting.
VBios update released in January.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814150690
9 reports of artifacting out of 100 reviews.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814202044
8 reports of artifacting out of 21 reviews.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814125475
Crashing/BSOD -- Has since been replaced with REV2.

GTX 770:
http://www.newegg.com/Product/Product.aspx?Item=N82E16814127741
4 reports of artifacting out of 148 reviews.
No reports of fan failures.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814121770
1 report of artifacting out of 161 reviews.

http://www.newegg.com/Product/Product.aspx?Item=N82E16814130921
1 report of artifacting out of 238 reviews.

Newegg reviews? Those are screened real well for accuracy and source (said no one ever).

Listen, you don't ever have to buy AMD if you don't want to. It's fine with me. I'm not here to start a smear campaign against nVidia. You obviously have an agenda against AMD though.
 
Newegg reviews? Those are screened real well for accuracy and source (said no one ever).
If you take all cards into account, it heavily favors Nvidia across the board. You're going to tell me this is due to some worldwide bias against AMD and not due to AMD making defective products?
If Newegg reviewers were the problem, then the poor reviews would be evenly balanced among Nvidia and AMD.

Where's my tinfoil hat...
 
If you take all cards into account, it heavily favors Nvidia across the board. You're going to tell me this is due to some worldwide bias against AMD and not due to AMD making defective products?
If Newegg reviewers were the problem, then the poor reviews would be evenly balanced among Nvidia and AMD.

Where's my tinfoil hat...


Youve been reasonable about it, so i wont call nvidia troll, but honestly, those sources are laughably suspect.

As i mentioned earlier, if there was a problem, the reputable sites/publications would have picked up on it long ago..

As far as newegg goes, ive ordered hundreds of product. You know how many reviews ive done? 2....yes, two.

So you need some kind of compensation factor to account for all the people that dont bother writing a review because their shit just works.

That lowers your percentage significantly.
 
Youve been reasonable about it, so i wont call nvidia troll, but honestly, those sources are laughably suspect.
There's a pile of "suspect" evidence against AMD's reliability, and not a single shred against Nvidia.
Normally I would say there's a plot against AMD since a lot of people go out of their way to discredit them. But in this case we're talking about 300-400 reviews spread across a dozen video cards, Sapphire's own product owners, and the failure rate stats I linked earlier.

At what point does it become credible? I'm sure there are more sources to be pulled out.
 
no one is saying anything about nvidia, other than some isolated incidences.

But evidence has to be credible, all the stuff you listed hasn't been qualified by anything remotely legitimate.

as i said, any newegg reviews need to be adjusted by some factor account for purchasers who dont review.
 
One of the largest e-tailors of GPU'S in UK and I sure If i look around more,it would be the same for most e-tailors.I am sure this could be spinned that Nvidia sells more cards so AMD has a higher return rate.Whatever floats any ones boat.
AMD has good products and I am convinced that alot of people are just brainwashed with so much crap on the internet.
http://forums.overclockers.co.uk/showpost.php?p=25749917&postcount=12

Quote from Gibbo on the site.
Our returns rate is less than 5%, same for a card like GTX 780, so I think that maybe the black screen issue is overly talked about but in reality can't be that bad when the statistics for returns seem pretty much around the same as other high-end cards from AMD and NVIDIA.

I don't think there is any issue with the reference board, I've got one in my home PC, I only get a black screen if I try to push the memory from its stock 5000Mhz to beyond 6200MHz, anything under 6200MHz and there is no issue. I've tested cards at OcUK upto 6600-6800MHz with no black screen issues. Several members of staff here have 290's in crossfire and tri-fire, again no black screen issues.

If there was a failure in the design we'd have a 100% return rate. So AMD's design seems perfectly fine and until more custom cards come out, there is nothing to say that cards not using AMD designs won't suffer the same black screens. What it could most likely be is the simple fact Hawaii is a brand new architecture and some of the game coders are still adapting their software to run properly on it, hence why we see both game patches and AMD software to address the issue. Which obviously has some truth to it as newer cards still using AMD reference design seem to be having less issues which is no doubt down too because new purchases are using new software, later AMD drivers, games already patched. But more importantly it was probably something to do with the BIOS on the cards.

My Asus 290X reference at home before would only benchmark upto 6200MHz and was only game stable and black screen free upto around 5800-6000MHz. Asus released a new BIOS and I can now game upto 6200MHz and benchmark upto 6500Mhz.

So I personally feel that the issue is a mix of BIOS not correctly optimised for the memory fitted to the cards and drivers/software patches not correctly optimised for Hawaii architecture.
 
One of the largest e-tailors of GPU'S in UK and I sure If i look around more,it would be the same for most e-tailors.I am sure this could be spinned that Nvidia sells more cards so AMD has a higher return rate.Whatever floats any ones boat.
AMD has good products and I am convinced that alot of people are just brainwashed with so much crap on the internet.
http://forums.overclockers.co.uk/showpost.php?p=25749917&postcount=12

Quote from Gibbo on the site.
Our returns rate is less than 5%, same for a card like GTX 780, so I think that maybe the black screen issue is overly talked about but in reality can't be that bad when the statistics for returns seem pretty much around the same as other high-end cards from AMD and NVIDIA.

I don't think there is any issue with the reference board, I've got one in my home PC, I only get a black screen if I try to push the memory from its stock 5000Mhz to beyond 6200MHz, anything under 6200MHz and there is no issue. I've tested cards at OcUK upto 6600-6800MHz with no black screen issues. Several members of staff here have 290's in crossfire and tri-fire, again no black screen issues.

If there was a failure in the design we'd have a 100% return rate. So AMD's design seems perfectly fine and until more custom cards come out, there is nothing to say that cards not using AMD designs won't suffer the same black screens. What it could most likely be is the simple fact Hawaii is a brand new architecture and some of the game coders are still adapting their software to run properly on it, hence why we see both game patches and AMD software to address the issue. Which obviously has some truth to it as newer cards still using AMD reference design seem to be having less issues which is no doubt down too because new purchases are using new software, later AMD drivers, games already patched. But more importantly it was probably something to do with the BIOS on the cards.

My Asus 290X reference at home before would only benchmark upto 6200MHz and was only game stable and black screen free upto around 5800-6000MHz. Asus released a new BIOS and I can now game upto 6200MHz and benchmark upto 6500Mhz.

So I personally feel that the issue is a mix of BIOS not correctly optimised for the memory fitted to the cards and drivers/software patches not correctly optimised for Hawaii architecture.

Better be careful or he won't be getting any Christmas cards from nVidia. :D

Makes it a bit more credible when someone actually puts their name and their company's name behind it.
 
no one is saying anything about nvidia, other than some isolated incidences.

But evidence has to be credible, all the stuff you listed hasn't been qualified by anything remotely legitimate.

as i said, any newegg reviews need to be adjusted by some factor account for purchasers who dont review.

Yep. I bought an Asus 280X a couple months ago and it's been fine. Maybe in another month it will die, maybe tomorrow, maybe it won't. From what I saw of reviews off newegg and amazon, there was a bad batch of cards from Asus at launch and that spread panic so Asus released a vbios update. Maybe it was all the asshats that mined with it and didn't have proper cooling. I haven't seen an issue with mine yet. What I haven't done is freak out nor post a positive review on newegg.
 
and not a single shred against Nvidia.

Older, classical fud campaigns that went against nvidia like the "their drivers kill your cards" and "omg blackscreens", are some that quickly come to mind.

I purchase from both camps often, so I'm not emotionally attached to either brand like some users on this forum.
 
hmmm,

i have a bizarre screen artifact whereby a strange pattern of curruption happens in the top-right menu bar of firefox when it is full-screened.

It only occurs in firefox, and parts of it will disappear if you mouse-over it.

This does not happen with any other application, i can leave all manner of benchmarks etc running at high-temps and it won't happen. The card is a R9-270.

Is this the kind of thing we are talking about?
 
I'm surprised that's the only problem you found with Firefox. I uninstalled it.
 
Yeah, firefox is a mess.

We are more referring to random blackscreens, texture corruption in games, and random desktop artifacts.
 
I've been having issues with corruption at the windows desktop that was in the form of 2 on screen squares of random pixels. I -think- that I accidentally ended up with 13.x drivers by way of letting the MSI utility update things during setup. This is on an MSI Gaming 280.

I also can't seem to stop the snap/crackle/pop of the AMD audio via HDMI when nothing is going on. Anyways, now all updated to 14.7 and hoping I've seen the last of the screen corruption.
 
Back
Top