-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Radeon RX 480 "Polaris" Launched at $199
- Thread starter Kaapstad
- Start date
It's not like the new nvidia process OCes a lot, people hype about 200mhz boost on 1080s and 1070s but in reality it's barely 10% since the clocks are already high and the cards barely go further than the boost clocks on AIB cards.
I think we won't see more than 1400-1450mhz on those cards and the ones that hit more will be cherry picked chips that will come at extra cost.
Either way, at 1400mhz this is already a beast for the price.
Watch this video it highlights the card on an old motherboard which worked with 980ti but crashed with the RX480.
Also addresses the issue of HYPE that ran wild on the internet not only seen in this thread but like the power issue is being somewhat blown out of proportion.
Last edited:
Soldato
What perf is it at 1400Mhz ? (What card is it compared to at that speed?)
Soldato
Fury x by the firestrike score.
Soldato
Stop making me want to go out and grab one lol. 30 min drive.
Firestrike it may match FuryX but do the game benchmarks also go upto FuryX level?
Impossible to say before it's actually tested in real games.
I would guess closer to a well clocked 970.
Soldato
The problem is, now that it's out it doesn't matter actually how much of an issue it may or may not be. The nay sayers, the fanboys and the press have the news and therefore it won't die.
It worries me though how this is much more of an issue to a lot of people than the whole 970 3.5GB thing.
It worries me though how this is much more of an issue to a lot of people than the whole 970 3.5GB thing.
Soldato
Well going by the guys stream, he managed to get a watercooled reference rx480 to 1410mhz stable on core with fury x performance from the firestrike numbers. he couldn't get it past that even with modding it with more capacitors to smooth the voltage.
he tried to volt mod it, but it appears to be a pain to do, so he didn't manage in the end.
he tried to volt mod it, but it appears to be a pain to do, so he didn't manage in the end.
So the stream ended. Buildzoid's verdict "don't buy the ref card". 1410 was last stable-ish clock, some said it was artefacting a bit, but it passed the FS extreme at 1410, but crashed 5 sec in at 1420. He also said you need to run Power Limit at 50 out of the box, wasn't in the stream since the start so probably missed something.
Soldato
- Joined
- 19 Feb 2011
- Posts
- 5,849
Sapphire Edd said they had sorted the PCB for the Nitro, which hopefully means they sorted power delivery to actually a normal method which in turn I hope will help overclocking.
If it's 1350 and you can get a 10% OC that's going to put it at 1455, and if a 1410 oc puts the card at FuryX perf in 3DMark, will be interesting to see how it fares in gaming.
Then the Toxic is coming at a rumoured 1375, that could get close to 1500 potentially? Wonder what kinda performance that would yield?
If it's 1350 and you can get a 10% OC that's going to put it at 1455, and if a 1410 oc puts the card at FuryX perf in 3DMark, will be interesting to see how it fares in gaming.
Then the Toxic is coming at a rumoured 1375, that could get close to 1500 potentially? Wonder what kinda performance that would yield?
If people are getting 1350 overclocks on the reference then wouldn't that put it close to a 390X?. Not too bad for a budget card really.
The AIB cards will be the main cards to go for anyway since nearly every gpu has always sold more non-reference versions and with 1400MHz OC you are essentially getting a Fury for a budget price.
Soldato
If people are getting 1350 overclocks on the reference then wouldn't that put it close to a 390X?. Not too bad for a budget card really.
The AIB cards will be the main cards to go for anyway since nearly every gpu has always sold more non-reference versions and with 1400MHz OC you are essentially getting a Fury for a budget price.
so you can magic up the extra shaders, rops and compute units then
His equipment ran out of batteries, so didn't attempt to volt mod.
Also, I may have missed this, but did he say that the 6-pin only powers the core and not the memory? That's interesting if true, because that PCper video shows that it draws from the pcie a lot more than usual at the expense of drawing less from the 6-pin. (Around 75W both at stock).
If, at stock, the core alone is drawing around 75w, if they had used gddr5x that would have made the total much better. As Raja said in the PCper interview though, they could have gone for perf/w or perf/$, on the memory they chose the latter.
Last edited:
Soldato
https://www.twitch.tv/buildzoid/v/75850933
The stream in which a guy mods a 480, hear his closing thoughts at 1hr 42min mark. Thoughts? Also should this info be in its own thread as might get lost in here?
The stream in which a guy mods a 480, hear his closing thoughts at 1hr 42min mark. Thoughts? Also should this info be in its own thread as might get lost in here?
Calm down lads.
Simple really, a few the extra fps but with benefit of lower temps and avoid throttling for more consistent performance.
Its surprising sometimes when you look at cards like even the beastly Zotac 1080 Amp Extreme how few extra fps in many titles is only gained over the stock/boost.
Simple really, a few the extra fps but with benefit of lower temps and avoid throttling for more consistent performance.
Its surprising sometimes when you look at cards like even the beastly Zotac 1080 Amp Extreme how few extra fps in many titles is only gained over the stock/boost.
Last edited:
Soldato
if custom PCBs sort the issue out then no, it'll be nothing like the 970 ram thing. Rather people will just tell you to avoid the reference cards.
Last edited:
That is interesting and testable (change memory clock and see what effect there is on the sources) if the memory is powered mostly/entirely from the socket.
Wonder if AMD cheaped out and instead of the voltage controller distributing power the memory domain is hardwired to one source in which case only sacrificing performance would provide better safety or whether the system is granular enough they can update vbios or "bootstrap" some commands over I2C to the controller to change distribution.
Last edited: