So 480 or 970?

[onlybuilt4cubanxlinx]

Is there any type of time line for when after market rx 480's will be released? Are the prices the same?

[Starelementpoke]

2 minutes ago, Robdarian said:

So buy a card that doesn't perform how they said and undervolt it?  Shouldn't have to do that, is what i'am getting at.  Do what you like, but I wouldn't want to have to dummy my card I just bought.

I'm not saying you should. But people have been reporting that the cards have been more stable after the fact, and that they've even been able to do a bit of OCing after the fact as well.

[App4that]

2 minutes ago, Starelementpoke said:

Clocks are still the same, if not better, with people undervolting it. Don't see where you're coming from.

Stability. I can show you a Firestrike score that I have no hope in hell of using in game. Because my reference PCB can't supply the power to keep it.

[Starelementpoke]

1 minute ago, onlybuilt4cubanxlinx said:

Is there any type of time line for when after market rx 480's will be released? Are the prices the same?

Like a week or 2 or so from what I know.

[Starelementpoke]

1 minute ago, App4that said:

Stability. I can show you a Firestrike score that I have no hope in hell of using in game. Because my reference PCB can't supply the power to keep it.

Decently sure that lowering the voltage has made the cards more stable as well, though I haven't had too much time to read the reddit.

[App4that]

1 minute ago, Starelementpoke said:

Decently sure that lowering the voltage has made the cards more stable as well, though I haven't had too much time to read the reddit.

Playing what game? And are we trusting anecdotal evidence? If AMD have given us a graphics card that performs better the less voltage you give it, I'll buy one.

[Starelementpoke]

2 minutes ago, App4that said:

Playing what game? And are we trusting anecdotal evidence? If AMD have given us a graphics card that performs better the less voltage you give it, I'll buy one.

Seriously? Hold up, asking if people have tested it yet.

[App4that]

2 minutes ago, Starelementpoke said:

Seriously? Hold up, asking if people have tested it yet.

Completely serious. If lowering the voltage increases stability and overclocking headroom. I'm throwing a 480 in the HTPC hooked up to a AA battery. 4k gaming here I come!

 

(last bit was a joke, I'm a bad person)

[Starelementpoke]

2 minutes ago, App4that said:

Completely serious. If lowering the voltage increases stability and overclocking headroom. I'm throwing a 480 in the HTPC hooked up to a AA battery. 4k gaming here I come!

 

(last bit was a joke, I'm a bad person)

Quick question, Firestrike is a synthetic benchmark right?

[App4that]

1 minute ago, Starelementpoke said:

Quick question, Firestrike is a synthetic benchmark right?

Yes, I get crazy clock speeds in Firestrike I can't use in game. Past 1530MHz. Why I only use it for quick testing.

[Starelementpoke]

3 minutes ago, App4that said:

Yes, I get crazy clock speeds in Firestrike I can't use in game. Past 1530MHz. Why I only use it for quick testing.

K, working on it. Gotta wait 3 minutes before reddit lets me save my post.

[Heesleemer]

vram 

[Starelementpoke]

9 minutes ago, App4that said:

Yes, I get crazy clock speeds in Firestrike I can't use in game. Past 1530MHz. Why I only use it for quick testing.

Eh, this might be enough to show it's stable. 

With the undervolt:

Without the undervolt:

Scores a bit higher too, but that's marginal and within margin of error I believe.

[Heesleemer]

so what your saying is undervolting made it perform better?

[App4that]

2 minutes ago, Starelementpoke said:

Eh, this might be enough to show it's stable. 

With the undervolt:

Without the undervolt:

Actually shows it's NOT stable. The timed measurement inconsistencies is a lack of stability. Once you fail in Firestrike you can't trust that clock speed at that voltage.

[Starelementpoke]

1 minute ago, Heesleemer said:

so what your saying is undervolting made it perform better?

Well not necessarily. It allowed people to maintain OCs, that in turn caused it to perform better.

[Starelementpoke]

3 minutes ago, App4that said:

Actually shows it's NOT stable. The timed measurement inconsistencies is a lack of stability. Once you fail in Firestrike you can't trust that clock speed at that voltage.

The 1st picture was with the undervolt, with the guy lowering the voltage. Valid score. The 2nd was without the undervolt, with the card at stock voltage.

[App4that]

Also, I never trust a Firestrike test without a link

[App4that]

1 minute ago, Starelementpoke said:

The 1st picture was with the undervolt, with the guy lowering the voltage. Was confirmed stable. The 2nd was without the undervolt, with the card at stock voltage.

No  

 

Unless you lock the voltage it's dynamic. You can pass a test in Firestrike and not have it stable. Why multiple tests are needed. And without a link all you have is what they're willing to share.

[Starelementpoke]

2 minutes ago, App4that said:

Also, I never trust a Firestrike test without a link

The guy's post:

[–]omair946600K, RX 480 6 points 10 hours ago* 

I kept clocks and power limit stock, only changed voltages: http://i.imgur.com/XY3gzxf.png

Firestrike (undervolt): http://i.imgur.com/hQH4Adf.png

Firestrike (stock volt): http://i.imgur.com/2DsOfpT.png

GPU-Z: http://i.imgur.com/depr1ui.png

should note that VDDC max was 1.1750, but this was a spike and while i watched it run, it was typically around 1.0500

same goes for GPU power draw, hits 131W, but typically was around 100 to 125.

Ill upload a log of GPU-z during Firestrike.

EDIT: GPU-Z Log: https://drive.google.com/file/d/0ByoDPh0-_YAgd3FDU09ub00xNUE/view?usp=sharing

[Starelementpoke]

3 minutes ago, App4that said:

No  

 

Unless you lock the voltage it's dynamic. You can pass a test in Firestrike and not have it stable. Why multiple tests are needed. And without a link all you have is what they're willing to share.

Eh, fair enough. Another guy's post, has Wattman up with the core voltage set to manual:

GPU-Z chart there as well.

[App4that]

2 minutes ago, Starelementpoke said:

The guy's post:

[–]omair946600K, RX 480 6 points 10 hours ago* 

I kept clocks and power limit stock, only changed voltages: http://i.imgur.com/XY3gzxf.png

Firestrike (undervolt): http://i.imgur.com/hQH4Adf.png

Firestrike (stock volt): http://i.imgur.com/2DsOfpT.png

GPU-Z: http://i.imgur.com/depr1ui.png

should note that VDDC max was 1.1750, but this was a spike and while i watched it run, it was typically around 1.0500

same goes for GPU power draw, hits 131W, but typically was around 100 to 125.

Ill upload a log of GPU-z during Firestrike.

EDIT: GPU-Z Log: https://drive.google.com/file/d/0ByoDPh0-_YAgd3FDU09ub00xNUE/view?usp=sharing

More pictures. Here is what you need to prove something.

 

http://www.3dmark.com/fs/9006144

[BrightCandle]

The only way AMD resolves the power issue with the RX 480 is in loosing some performance, they need their voltage margins to remain (they chose it for a reason) so its going to be a reduction in clockspeed and the power the card can draw will have to be -30 watts or so. That is a pretty dramatic reduction in power consumption and its going to have a significant knock on effect on performance. It reminds me of the Fury mess all over again where they dropped performance dramatically on retail cards and in the end just boosted the fans a tonne to compensate making them exceptionally loud.

 

But the 970 is long in the tooth now, its only got 4GB of VRAM and its been out a while, which IMO makes neither a good buy currently. I would either stretch a lot to the 1070 or await the 1060 as only AMD has released their mid tier currently and Nvidia is quite likely to compete well just like it did with the previous set of cards as they have quite an advantage around power consumption currently.

[App4that]

Again, and I say this as someone who has spent a lot of time in Firestrike. One test say absolutely NOTHING. It takes playing games for hours to see if a overclock is stable, and even then you can have issues. lowering the voltage will not add stability. That is counter intuitive.

 

 

[Starelementpoke]

9 minutes ago, App4that said:

Again, and I say this as someone who has spent a lot of time in Firestrike. One test say absolutely NOTHING. It takes playing games for hours to see if a overclock is stable, and even then you can have issues. lowering the voltage will not add stability. That is counter intuitive.

 

 

Would it's stress test count for something?