AMD: Speculation, Rumors, and Discussion (Archive)

Status
Not open for further replies.
Answering myself
AMD-Polaris-Tech-Day-2.jpg


Also the 380X looks pretty good in that Benchs.

Welll mostly on the first 1080p run ( i dont read polish but i can imagine it is minimum detail ).. I have only been able to compare Witcher, and without knowing much details on settings and system, cant be sure of the result, but they seems on paar with the review i have consullt.
 
I dont know how true they are, but for complete, Vz said the 970 is the Zotac Amp OC who run at 1304mhz.
970 Factory OC's, so not surprised, manually can OC higher. They tried to OC 480 but behavior so far is similar to Fiji OC capability. Since this is an official reviewer I assume the are using AMD's official drivers.

According to the review, 5% overclocking causes severe artifices in 3DMark, while 7% overclock causes system to freeze. Not a word about drivers uses though.
http://videocardz.com/61512/first-radeon-rx-480-gaming-benchmarks-hit-the-web
 
970 Factory OC's, so not surprised, manually can OC higher. They tried to OC 480 but behavior so far is similar to Fiji OC capability. Since this is an official reviewer I assume the are using AMD's official drivers.


http://videocardz.com/61512/first-radeon-rx-480-gaming-benchmarks-hit-the-web
I can say only: all this is strange. We did not get that kind of issues at all. Maybe unlucky sample?

EDIT: I mean we did not get artefacts or system freeze. If the cards is sane, an unbalanced over-clock should produce a driver recovery. Over-volt is of course needed.
 
Last edited:

So 380X is now somehow faster than both RX480 and superclocked 970 :D Albeit on med settings. And why are there three Firestrike Extreme bars for RX480?! This looks like the worse fake I've seen so far.
 
Last edited:
its out of a magazine so most likely not fake, just screwed up somewhere. Any case all the video leaks, showed the rx 480 being around the 970, so unless there is some Dx12 magic or magic drivers, I think that is where the reference rx 480 is going to land, which is actually a dampener again on current expectations of r390x level performance......
 
its out of a magazine so most likely not fake, just screwed up somewhere. Any case all the video leaks, showed the rx 480 being around the 970, so unless there is some Dx12 magic or magic drivers, I think that is where the reference rx 480 is going to land, which is actually a dampener again on current expectations of r390x level performance......

I'm not disputing the actual performance of RX480 which I expect to be at 390 level, I'm just pointing out some weird stuff which is even weirder considering this is suppose to be a printed magazine. I guess QC is in the toilet.
 
Keeping in mind that is the reference model, that's not bad at all if it will quite reach the 390 performance, 'cause this will open very interesting scenarios for custom AIB versions. Of course prices need to be lower.
 
When the attractiveness of GPU solely depends on the overclocked versions (and low price), you have a bit of a problem.
Overclocking is supposed to be a nice to have extra.
 
The tech media's overall sloppiness in making the distinction between reference and third-party overclocked models indicates there's utility in this. It has inflated Nvidia's reputation versus AMD, with initial impressions of cooler and quieter products becoming muddled with the performance of the more exciting third-party cards. People frequently miss when comparisons start using OC models, and reviewers might throw a "by the way, this is louder" caveat in the middle of one of the last pages.
I think that was effective in reducing the little goodwill AMD tried to get back in recent products.

It's a further step past the usual "wait for third-party coolers" since having a decent cooler or a vague semblance of quality control has also become a nice extra, although AMD in particular has suffered more and I think that could be argued as justified.
 

Wait what? It's showing the OC 380X as just 9-23% slower than a OC 970 depending on game/benchmark. Looking at Anandtech an OC 380X is about 25-35% slower than a STOCK 970 not even an OC version. Again, depending on game/benchmark.

I dunno. Somehow I don't trust that review.

Regards,
SB
 
I think it makes total sense to release overclocked versions for the reasons you mention. But rarely does that elevate a GPU from 'meh' to 'great'.


Actually I don't think OC versions ever have done that, not that I can remember at least. The best showing from performance and power consumption has always been the stock versions of the cards.
 
When the attractiveness of GPU solely depends on the overclocked versions (and low price), you have a bit of a problem.
Overclocking is supposed to be a nice to have extra.
Well in some apps cf 480 > 1080 > sli 980. That suggests it's reasonable to expect 980 level performance out of the box at least in some apps.

Now over at neogaf they were showing benchmarks with a custom OC 970 it was about 20% faster than the 980, the one[970] in this controversial print magazine has less than 10% difference in clockspeed with that OC extreme 970. So unless a less than 10% difference in clockspeed yields greater than 20% difference in performance, this card must be around 980 going by some benchmarks

Now, I could maybe buy the 480 reference having abysmal oc, maybe because of cheap cooler, but AIB must have decent OC. IF they had abysmal OC capability why make such a big deal of new OC software if you're only getting abysmal OCs? What of the 300$ cards, obviously that ain't selling if the OC is abysmal, like what would be the logic in even making such a product as it would be unmarketable?
 
Last edited:
A techpowerup forum member named gasolina seems to really have bought the card and shared some tests with the guys there. His firestrike graphics score is ~12800.

He ran Batman AK benchmark, and the performance he got is the same as a GTX980 at both 1080p and 1440p.


Still don't have 10 messages to post link, but i tell link is on Videocardz RX 480 rumors, part 7.

The strangest thing is that he is using a driver from May. Has AMD (with 16.6.1B and 16.6.2B) fooled all the leakers until now?
 
Actually I don't think OC versions ever have done that, not that I can remember at least. The best showing from performance and power consumption has always been the stock versions of the cards.

And yet, when we've seen in the past review sites comparing stock AMD cards to stock Nvidia cards, there's almost always an outcry about why they aren't using factory OC'd Nvidia cards since the benchmark for the Nvidia cards would be higher then.

It may not elevate it from meh to great as Silent_Guy said, but there certainly appears to be some fear from fans that it doesn't make their choice of vendor look as good versus the competition. I'm sure AMD fans would do that same given the opportunity, but there's not only a lot less of them, but the AMD factory OC cards have generally not been as impressive other than in reducing noise.

IE - the OC versions generate a disproportionate amount of publicity potentially driving sales disproportionate to the actual performance of the card (factory OC version versus stock version).

Regards,
SB
 
Ok yeah, performance wise the overclocked versions do look better on reviews, but the most important reviews are always launch reviews, those have the most coverage, and most views. And its important to hit those reviews out the ball park to have a successful product.
 
Status
Not open for further replies.
Back
Top