Bugs in 7000 series and 600 series cards

[BrightCandle]

I think its time to get a confirmed list of bugs down for both sets of cards before the new generation comes out. This is to test that this format can work in preparation for the new cards and the bug reports that will inevitably appear here.

I will maintain the top post and below the list of bugs. I have put some recent bugs down with confirmations that came from threads we have had here previously but I want to expand this list, especially the 600 series list.

7000 series bugs
Crossfire microstutters in a lot of DX9 games
Stuttering in a lot of modern games
Eyefinity resolutions cause performance problems in Windows
Clock speeds sometimes drop in certain games (hitman, SWTOR) (fixed in 12.11 beta?)
Artefacts in some DX9 games with shadows
Arma 2 white speckles for antialiased foilage

600 series bugs
SLI microstutter in a variety of games
SLI severe microstutter in Heaven benchmark with vsync off

Rules
How do you get a bug added?

1) You list it in this thread
2) Two other people have to [confirm] that the bug exists independently in this thread.
OR
1) Provide a link to a thread that has a bug listed and you are sure has a confirmation from two other posters.

To remove a bug
1) Three confirmed reports of it being fixed along with some details in which driver.

 

[ShintaiDK]

PCI-E 2.0 is forced in drivers when the board and CPU do support PCI-E 3.0

This applies to desktop LGA2011 boards only. Might be worth noticing since the wast majority uses LGA1155.

 

[imaheadcase]

GTX 500 series. None.

 

[f1sherman]

PCI-E 2.0 is forced in drivers when the board and CPU do support PCI-E 3.0

Does that include the case of X79 + i7-3930k?

Because Intel seems to think those devices are only PCI-E 2.0 ready.
Both of them.

http://www.intel.com/content/www/us/en/chipsets/performance-chipsets/x79-express-chipset.html


http://ark.intel.com/products/63697/Intel-Core-i7-3930K-Processor-12M-Cache-up-to-3_80-GHz

 

[AdamK47]

Does that include the case of X79 + i7-3930k?

Because Intel seems to think those devices are only PCI-E 2.0 ready.
Both of them.

http://www.intel.com/content/www/us/en/chipsets/performance-chipsets/x79-express-chipset.html


http://ark.intel.com/products/63697/Intel-Core-i7-3930K-Processor-12M-Cache-up-to-3_80-GHz

Uh oh, not this again. It still surprises me just how many people are unaware of PCI-E 3.0 on SB-E.

 

[f1sherman]

Uh oh, not this again. It still surprises me just how many people are unaware of PCI-E 3.0 on SB-E.

Including Intel itself? LOL

When SB-E was launched it was advertised as PCI-E 3.0 kinda-ready, bit it never made it to 3.0

"The PCIe links can be arranged in any number of configurations, and although they’re listed as PCI Express 2.0 here, these integrated PCI Express lanes actually support PCI Express 3.0 speeds. We’re told the lack of PCI Express 3.0 GPUs and peripherals at this time have prevented Intel from definitively qualifying the Sandy Bridge-E’s integrated PCI Express lanes as PCI Express 3.0 capable, so they’re not being labeled as such. However, the functionality does exist, once testing and qualification is complete."



http://ark.intel.com/compare/64615,64622,63696,63698,63697

Straight from the horses mouth^^
Intel made a spelling error, did not update. Twice? Yeah, right...

Note PCI-E 3.0 for Xeons.

 

[Skurge]

For example, does really low performance in a game where it shoud be 2-3x faster count as a bug?

I'm guessing missing CFX or SLI profiles don't count.

 

[AdamK47]

Including Intel itself? LOL

When SB-E was launched it was advertised as PCI-E 3.0 kinda-ready, bit it never made it to 3.0

"The PCIe links can be arranged in any number of configurations, and although they’re listed as PCI Express 2.0 here, these integrated PCI Express lanes actually support PCI Express 3.0 speeds. We’re told the lack of PCI Express 3.0 GPUs and peripherals at this time have prevented Intel from definitively qualifying the Sandy Bridge-E’s integrated PCI Express lanes as PCI Express 3.0 capable, so they’re not being labeled as such. However, the functionality does exist, once testing and qualification is complete."



http://ark.intel.com/compare/64615,64622,63696,63698,63697

Straight from the horses mouth^^
Intel made a spelling error, did not update. Twice? Yeah, right...

Note PCI-E 3.0 for Xeons.

What PCI-E version are you running on your SB-E / X79 system?

 

[f1sherman]

Lets recap SB-E PCI 2.0/3.0:

SB-E had indeed been advertised as soon-to-become PCI-E 3.0, but it has never made it through rigorous testing and become fully 3.0 compliant.
In fact, from Intel's sheets it's clear that SB-E is only PCI-E 2.0 capable device, and SB-E Xeons are PCI-E 3.0 compliant.

You being able to run 3960k on PCI-E 3.0, does not in any way imply that everyone with SB-E should be able to do the same.

Running system out of specs is not that uncommon. Listing non-working "unofficial feature" as a bug is.

Now please stick to the subject, or if you think that my lack of platform in question disqualifies me for further discussion, make a thread "SB-E / X79 owners only" .

 

[notty22]

Sandy Bridge E & X79 PCIe 3.0: It Works



Anand has brought up the confusion in other x79 and the 3820 reviews.
Here is a google quote excerpt.
in SNB-E - AnandTech

www.anandtech.com/show/5091/intel-core-i7...e.../2


by Anand Lal Shimpi on 11/14/2011 3:01 AM EST ... the interface, Intel lists PCIe 3.0 support in the chip's datasheet but is publicly guaranteeing PCIe 2.0 speeds. ...

 

[AdamK47]

Lets recap SB-E PCI 2.0/3.0:

SB-E had indeed been advertised as soon-to-become PCI-E 3.0, but it has never made it through rigorous testing and become fully 3.0 compliant.
In fact, from Intel's sheets it's clear that SB-E is only PCI-E 2.0 capable device, and SB-E Xeons are PCI-E 3.0 compliant.

You being able to run 3960k on PCI-E 3.0, does not in any way imply that everyone with SB-E should be able to do the same.

Running system out of specs is not that uncommon. Listing non-working "unofficial feature" as a bug is.

Now please stick to the subject, or if you think that my lack of platform in question disqualifies me for further discussion, make a thread "SB-E / X79 owners only" .

You weren't aware of PCI-E 3.0 working on SB-E before. You and others reading this thread now know. That's all that matters.

 

[Elfear]

I think its time to get a confirmed list of bugs down for both sets of cards before the new generation comes out. This is to test that this format can work in preparation for the new cards and the bug reports that will inevitably appear here.

I will maintain the top post and below the list of bugs. I have put some recent bugs down with confirmations that came from threads we have had here previously but I want to expand this list, especially the 600 series list.

7000 series bugs
Crossfire microstutters in a lot of DX9 games
Stuttering in a lot of modern games
Eyefinity resolutions cause performance problems in Windows
Clock speeds sometimes drop in certain games (hitman, SWTOR) (fixed in 12.11 beta?)
Artefacts in some DX9 games with shadows
Arma 2 white speckles for antialiased foilage

600 series bugs
SLI microstutter in a variety of games
PCI-E 2.0 is forced in drivers when the board and CPU do support PCI-E 3.0 on LGA2011
SLI severe microstutter in Heaven benchmark with vsync off

Rules
How do you get a bug added?

1) You list it in this thread
2) Two other people have to [confirm] that the bug exists independently in this thread.
OR
1) Provide a link to a thread that has a bug listed and you are sure has a confirmation from two other posters.

To remove a bug
1) Three confirmed reports of it being fixed along with some details in which driver.

Are three people enough to confirm that a problem is a driver bug and not some other hardware/software issue?

 

[BrightCandle]

Fine I'll remove the PCI-E 3.0 bug for the 600 series.

 

[BrightCandle]

Are three people enough to confirm that a problem is a driver bug and not some other hardware/software issue?

Does it matter where the bug comes from? If we can have more qualification, like it only happens when you tap your heels together 3 times then I am fine with that. A lot of these bugs only happen in specific games, specific scenes or with certain combinations of hardware. If a piece of hardware has a known failure that happens frequently in hardware and the card needs RMA then I think we can put that in as well with a remedy (RMA it).

 

[Grooveriding]

Intel pulled a few bait and switches on X79/SB-E features. PCIE 3.0 was advertised and listed as available on Intel's site at launch, it was later removed, after already launching the product. Another was VT-D support, again listed at launch to have support pulled later and the ark site revised. That was an issue with the first stepping that C2 fixed.

PCIE 3.0 works fine on X79. AMD cards support it natively and officially, nvidia provides no official support/guarantee, but offer a program to enable it on X79. It works for most people with nvidia, for some it does not, it seems to be motherboard related.

Works just fine for me.

The basis for establishing a 'bug' in this thead is nonsensical to say the least btw.

 

[Granseth]

7000 series bugs
Crossfire microstutters in a lot of DX9 games
Stuttering in a lot of modern games
Eyefinity resolutions cause performance problems in Windows
Clock speeds sometimes drop in certain games (hitman, SWTOR) (fixed in 12.11 beta?)
Artefacts in some DX9 games with shadows
Arma 2 white speckles for antialiased foilage

I can't comment on the 600 series since I don't have one, but I don't have alot of the problems described here either.
I probably have the stuttering problem, but don't notice it. I don't have x-fire so can't comment on that.
Dont have problems with eyfinity (3840x1200)
Don't have or had clock speed drops
Don't have artifacts in DX9

So I don't disagree that it might be a problem, but it's not a universal problem. So maybe your list should indicate that.

I've got a bug with displayport though. Sometimes (totally random) when I watches a video (usually flash) my DP screen (the other is DVI) goes crazy and has a lot of horizontal stripes over a frozen screen image. I haven't researched the problem, but if other have something similar then it should get on the bug list.

 

[Plimogz]

This is a great idea -- Kudos OP!

If I may make a suggestion, and forgive me if this has already come up: It would be good if along with requiring driver version for reports of bug fixes, it was also required for bug reports.

While I suppose that this would come naturally enough (e.g. "I just installed driver X.XX.XXX and there's bug YYY with game Z"), I figure that we would aggregate better data right from the start if it was made explicit in the OP that we need the driver version.

Oh, and maybe include links in the top post for the bug report forms from both graphics vendors, just to try to increase the number of these reports which, hopefully, are immediately brought to the attention of the driver writers.

Just my $0.02.

And thanks for taking the time to do this, BrightCandle.

 

[Elfear]

Does it matter where the bug comes from? If we can have more qualification, like it only happens when you tap your heels together 3 times then I am fine with that. A lot of these bugs only happen in specific games, specific scenes or with certain combinations of hardware. If a piece of hardware has a known failure that happens frequently in hardware and the card needs RMA then I think we can put that in as well with a remedy (RMA it).

Is the point of the thread to send the list of bugs to AMD or Nvidia to fix? If so, then I think the source of the bug does matter. If Billy Bob and his two cousins frequently hit their greasy foreheads on their monitors and then complain about blurry games, I'm not sure how AMD/NVIDIA are going to fix it.

 

[Skurge]

I can't comment on the 600 series since I don't have one, but I don't have alot of the problems described here either.
I probably have the stuttering problem, but don't notice it. I don't have x-fire so can't comment on that.
Dont have problems with eyfinity (3840x1200)
Don't have or had clock speed drops
Don't have artifacts in DX9

So I don't disagree that it might be a problem, but it's not a universal problem. So maybe your list should indicate that.

I've got a bug with displayport though. Sometimes (totally random) when I watches a video (usually flash) my DP screen (the other is DVI) goes crazy and has a lot of horizontal stripes over a frozen screen image. I haven't researched the problem, but if other have something similar then it should get on the bug list.

While using a 7970, I didn't experience the Eyefinity issue in windows at 5760x1080. I also didn't have clock speed drops.

 

[BFG10K]

It’s not accurate to list micro-stutter as a driver bug for multi-GPU given it’s inherent to how AFR works. It’s like saying aliasing is a bug from rasterization.

But for single GPUs, excessive framerate swings is definitely a bug if the competition doesn’t have it.

Here are a couple of artifacting bugs in Black Ops and Stalker 3 on my GTX680:

Also Vietcong 1 consistently crashes to the desktop in the "crash in the jungle" mission when the game auto-saves the first time.

 

[Black Octagon]

Is the point of the thread to send the list of bugs to AMD or Nvidia to fix?

I too would like to know the specific purpose of this thread. Is it just for the general reference of the community?

The only problem I've experienced with my 7970(s) is microstutter in Crossfire (in all sorts of games, not just DX9 titles). Other than that, nada.

There's a thread on these boards dealing specifically with the topic of bugs unique to the 7950. No idea if there are any 7950-specific bugs, but if so would they be eligible for a thread like this, or are we talking about bugs shared by 'all' 7000 series cards?

 

[BrightCandle]

I have no doubt that AMD and NVidia are already aware of all the bugs we know about. They have been reported over and over. They just aren't getting fixed. We have numerous people asking the same question in these forums every week related to one of them and its hard to say that its a bug they are encountering because of the amount of ensuing partisan argument. These sort of its either a known bug or its not should be dead simple to answer. With so much argument about if bugs exist etc I would prefer us to channel that passion into a collaborative effort to confirm bugs that we can point people to. That is it.

Personally I would like a decent comparison as to the level of faults between the cards. I also want us to have thought about how we get these lists of bugs together for the coming new releases of hardware, because I don't want another year of arguing about a confirmed problem that has clearly been there since the beginning. It should not take an article from a major review site and like 5 confirmation sites to confirm a bug that is at least mostly accepted here, we don't need that level of proof.

There is passion here, there is knowledge and skill and I think we can channel that into building up something that maybe one day will have the manufacturers of these cards looking at such a thread to take bug reports. For the next generation we will want to report them alongside, and we will record on what date that was done. These bugs are for the most part as old as the cards, long ago reported by thousands of people.

 

[Black Octagon]

Ok sounds like a great idea. Unfortunately I have no specific bugs to report beyond what I already posted

 

[thilanliyan]

It’s not accurate to list micro-stutter as a driver bug for multi-GPU given it’s inherent to how AFR works. It’s like saying aliasing is a bug from rasterization.

But for single GPUs, excessive framerate swings is definitely a bug if the competition doesn’t have it.

Here are a couple of artifacting bugs in Black Ops and Stalker 3 on my GTX680:

Also Vietcong 1 consistently crashes to the desktop in the "crash in the jungle" mission when the game auto-saves the first time.

See, unless someone else is playing the same game with the same card and drivers (unlikely with Vietcong lol), this is not going to be listed as a bug even if it really is...

Assuming people are being honest, OP, you should list every bug mentioned here for completeness if anything (I know there are pitfalls to that). BFG10K has experience testing video cards and I would trust that bugs he is listing are actual bugs, and not something caused by his specific system, even if he is the only one listing them.

 

[thilanliyan]

nm