AMD Radeon RX 480 Owners Club

TaintedSquirrel said:
Has anyone managed to flash a retail 4GB -> 8GB ?
Click to expand...
tommicn00bie




I was reading on the techpowerup site that people have peen reporting that the 4gig card was actually an 8gig with the bios locked to 4. So I tore my card apart and it was indeed 8gig with the 8 Samsung chips ending in FB as seen on the screenshot attached. I flashed my bios from the one provided on techpowerup and it is working perfectly. Happy days.

I bought the reference 4gig XFX card from Ebuyer.

Sorry in advance if I mentioned sites if I'm not supposed to.
Click to expand...
 
so few responses. i guess not very many [H]ers have been able to get their hands on a sample.
 
rennyf77 said:
so few responses. i guess not very many [H]ers have been able to get their hands on a sample.
Click to expand...

I think most are waiting on AIB cards. Makes sense as there should be much improved over clocks and no voltage issues.
 
  • Like
Reactions: Creig
like this
rennyf77 said:
so few responses. i guess not very many [H]ers have been able to get their hands on a sample.
Click to expand...

[H] members on average have higher end rigs than your average Best Buy customer. If the GTX 970 is 5% of the Steam Survey, 970 and higher is likely at LEAST 25% of the [H] population. As such, the 480 just isn't as appealing out our general population as it may be to a Best Buy "gamer."
 
there use to be a few guys on this forum who would buy new gear and share their experiences just because it was new gear. I use to be one of those guys back when work was banging and I could afford it. times sure have changed. no one seemed to want to take a chance on the rx480, voltage issues be damned, especially some of the more vocal amd guys. I had a 1070 incoming and by the time I managed to talk myself in to grabbing a 480, there were none to be had. the egg isn't even offering any model for back order. I guess everyone's waiting for aib units.
 
rennyf77 said:
so few responses. i guess not very many [H]ers have been able to get their hands on a sample.
Click to expand...
How do we when we can't even buy one :(? Sort of wished I had pulled trigger on launch day but waited a bit to see 1060, but I'm leaning towards there being nothing to see there and will stick with 4GB RX480, when I can get one :/.
 
OC results showing potential. I plan on picking up the aftermarket 480 as soon as it releases ....
Is that next week ?!
 
I have noticed sometimes 3dmark score will drop if you overclock to high, and your score just seems lower than stock clock 3dmark to me. You sure its not power throttling? Edit... Actually i see your Graphics Score 14,743...maybe its slightly cpu bottle necked.
 
Last edited:
I just built this rig last Friday.


My first run with a RX480 resulted in 14,305 graphics. All I did was set memory to 2,200 and turn power limiter all the way to 50%. I did not touch the core clocks.

Firestrike scores!

Graphics=14,305
Physics=10,505

I5 6600K 4.6Ghz 1.3v
Z170 mini itx
16gb DDR4 3000 can 14
Raid 0 ssd's
RX480 8gb


I love this video card. I only play at 1080P so, I can Max everything out! I have been building computers since 2002. And this is probably the best video card I've ever had.

Great job AMD!

I will post some pics tonight.
 
Got my 480 Nitro+ OC today, took some pics, not installed yet:
https://goo.gl/photos/UpRALNUJTvfCtxvW7

Feel free to add your own pictures to the album if you want. ;)
Edit: It's the 21260-01-20G SKU, if anyone's wondering--ordered monday from NewEgg.
Edit2: Measured boost clock with GPU-Z while running timespy (it had an error after the first GPU test, so I don't have valid results to show right now). Graphs are in the gallery linked above. Also added pics of the card installed in my PC.

First large block is the timespy demo (from about 112s to 240s), second block is GPU test 1.

Each minor line on the y-axis is 25MHz, each major line is 200MHz. Looks like the lowest it dropped during load was about 1280MHz during the Timespy demo, with everything else staying above 1300 or right on it. I haven't messed with power or clocks yet because Crimson isn't opening for some reason--I'll probably have to clean the drivers (again) and reinstall, maybe use the ones from Sapphire this time. :/
 
Last edited:
So far liking this 480. Reference design is super quiet on default, almost scarily so for me coming from a the old 7970. Also do they dynamically set the voltages for these things, my card has much lower voltages than the reference design for review samples. Seems the newer batches might be doing better?
 
gordon152 said:
So far liking this 480. Reference design is super quiet on default, almost scarily so for me coming from a the old 7970. Also do they dynamically set the voltages for these things, my card has much lower voltages than the reference design for review samples. Seems the newer batches might be doing better?
Click to expand...

yea they do. Are you looking at voltage during desktop usage? did you run GPUz and click sensors and click on the voltage section until it says max in green letters and run a 3dmark firestrike as its pretty intense. GPUz should record the max voltage once the benchmark is done.
 
NKD said:
yea they do. Are you looking at voltage during desktop usage? did you run GPUz and click sensors and click on the voltage section until it says max in green letters and run a 3dmark firestrike as its pretty intense. GPUz should record the max voltage once the benchmark is done.
Click to expand...

I am looking at Wattman settings and using GPUz sensor logs to track settings during gaming sessions. I'm starting to see why reference cooler kills this cards performance, in order to meet the temperature targets the card is keeping my average GPU speed at like 1150 or below. Ordered a Mono Plus, so will see how that works out.

edit: Lowered the temperature target to 75, increased power limit to 20% and rose target fan rate to 3000 and increased boost maximum to 1305. Seems to have stabilized things and the fan at 3000 (min 2200), even though loud is nothing to me (7970 BE OC isn't the quietest of cards).

Overall, though going to swap out cooler, numbers I'm seeing are so more consistent now. Middle-earth looks great with everything maxed and it uses a huge amount for frame buffer, almost 7GB.

BTW for GPU testing, Timespy seems to be better than actual gaming or Firestrike. Earlier comment about averages might have been flawed, with timespy my measured average using GPUz is more in the mid 1200s ok.
 
Last edited:
i'm hoping some of you guys with aib polaris in hand have held onto your hawaii cards. i'd really like to see how the higher boosting rx480s compared to hawaii oc'd to atleast 1100mhz. my last hawaii card was the 8gb 290x but it sold while i've been waiting to grab an aib polaris of my own.
 
Still have my 270x, but it's Pitcarin XT, and I could never get it to overclock very high.
 
rennyf77 Finally got around to playing some Battlefront. Made a quick vid showing the Speeder Chase in Ultra, and plotted the GPU-Z data (last image in album; finally figured out how to use the secondary y-axis in libreoffice).



Dunno what happened today--yesterday I was doing pretty good. lol
Edit: YouTube crapped all over the video when they transcoded it. >_<
 
Last edited:
Nobu said:
rennyf77 Finally got around to playing some Battlefront. Made a quick vid showing the Speeder Chase in Ultra, and plotted the GPU-Z data (last image in album; finally figured out how to use the secondary y-axis in libreoffice).



Dunno what happened today--yesterday I was doing pretty good. lol
Edit: YouTube crapped all over the video when they transcoded it. >_<
Click to expand...


I'm running the same card. I finally played through Doom this weekend. Minimum frames at 1080p with Ultra settings never went below 100fps in the worst spots (except locked 60fps choreographed plot bits) usually hovering around ~150 fps or more.

I think my next project will be seeing how far I can undervolt while maintaining the 1342mhz core.
 
I somehow managed to nab a Powercolor 8GB off Newegg about a week ago. Tried to buy 2 but they stopped me.
Anyway...
Haven't tried overclocking but I put it in a B150/6500 system I built for my daughter and fired up Tomb Raider from 2013 that I got for 4 bucks off GMG, and it is running ultra max with TressFX fully enabled FXAA on at 1080p 55-60 fps according to the built in benchmark. Pretty happy.
Have not heard any fan noise and that is with system is up on the desk with me with the side off while I finish the build even after several hours of gaming.
 
wrangler said:
Haven't tried overclocking but I put it in a B150/6500 system I built for my daughter and fired up Tomb Raider from 2013 that I got for 4 bucks off GMG, and it is running ultra max with TressFX fully enabled FXAA on at 1080p 55-60 fps according to the built in benchmark. Pretty happy.
Have not heard any fan noise and that is with system is up on the desk with me with the side off while I finish the build even after several hours of gaming.
Click to expand...

That disappoints me. I had higher hopes for this card. The 2013 TR generally does better (slightly) on AMD than NV. My GTX 970 did the same benchmark, fully maxed (Ultimate + maxing out that one setting that didn't on its own), FXAA, LOCKED 60fps on that benchmark.

My 1060 does the same, so I've upped the game to 4k (nowhere near max settings, of course).
 
Messing around a bit with voltages and freqs, managed a solid 1355 while running heaven (and almost 40fps 1440p max settings, 2xAA), then opened wattman again to take a screenshot and it crashed. lol



I had managed to finish the benchmark at 1360, but then I started getting greedy, dropped the voltage more and increased the freq, and it started crashing left and right. Managed to get it somewhat stable again at these settings.
 
Well. I don't have 4K but I could turn on super resolution and see how it does.
Also. Not sure how accurate the benchmark is in showing that it is dipping below 60. I got the same numbers with the ultra preset and then after I went in and turned up everything that ultra forgot plus TressFX. Exactly same numbers.
 
Nobu said:
Messing around a bit with voltages and freqs, managed a solid 1355 while running heaven (and almost 40fps 1440p max settings, 2xAA), then opened wattman again to take a screenshot and it crashed. lol



I had managed to finish the benchmark at 1360, but then I started getting greedy, dropped the voltage more and increased the freq, and it started crashing left and right. Managed to get it somewhat stable again at these settings.
Click to expand...

Under water with my reference XFX 1375/2250 seems to be my max. A couple FireStrike runs and Doom @ 4k on Ultra it never goes above 55'C though.
 
Seems like if you change it from % to "Dynamic" it actually stays at whichever frequency you set for the powerstate it's in, instead of the fluctuations you get with the offset OC. Rather than Dynamic, they should have called it Static, I think.

1375/2250 is pretty impressive, I've only squeezed out 2050 on the memory side, although I did leave that on automatic voltage. I think I'm content with what I got, at least until I put a block on it. 7W idle and 160W load is pretty good at 1355, IMO.
 
Nobu said:
Seems like if you change it from % to "Dynamic" it actually stays at whichever frequency you set for the powerstate it's in, instead of the fluctuations you get with the offset OC. Rather than Dynamic, they should have called it Static, I think.

1375/2250 is pretty impressive, I've only squeezed out 2050 on the memory side, although I did leave that on automatic voltage. I think I'm content with what I got, at least until I put a block on it. 7W idle and 160W load is pretty good at 1355, IMO.
Click to expand...
Speaking of which, does anyone know the safe-ish side for boosting the memory voltage? That seems to be my weak link as well but I've been afraid to tweak that voltage.
 
Magic Hate Ball said:
Speaking of which, does anyone know the safe-ish side for boosting the memory voltage? That seems to be my weak link as well but I've been afraid to tweak that voltage.
Click to expand...

So far I've had success hitting 2200-2250 on the memory with an undervolt of 980mv. That hasnt been tested in games too extensively just yet though. It's held up pretty well running some benchmarks.
 
unfortunately benchmarks are often unreliable when it comes to oc stability-unless you're testing with furmark. good luck with all of you testing the limits of your polaris cards. i will join you as soon as i can get my grubby mits on a gigabyte aib sample.
 
rennyf77 said:
unfortunately benchmarks are often unreliable when it comes to oc stability-unless you're testing with furmark. good luck with all of you testing the limits of your polaris cards. i will join you as soon as i can get my grubby mits on a gigabyte aib sample.
Click to expand...
Good luck--did you catch them earlier today? I saw one go in stock earlier, but it was out before I went to the page, don't remember if it was gigabyte or not.

Here's my Firestrike/Timespy scores (with OC, compared to futuremark's "VR Ready" machine). Please ignore the poor combined score--my A10-6800k is showing it's bad side. :p
Note that my monitor was set to a virtual resolution of 1440p for both tests, I assume the resolution was changed before the test began, but it's not made clear on the results page. Also, they aren't "valid" results because it fails a time variance test, but nothing looked out of place visually, so take that with a grain of salt.
 
Nobu said:
Good luck--did you catch them earlier today? I saw one go in stock earlier, but it was out before I went to the page, don't remember if it was gigabyte or not.

Here's my Firestrike/Timespy scores (with OC, compared to futuremark's "VR Ready" machine). Please ignore the poor combined score--my A10-6800k is showing it's bad side. :p
Note that my monitor was set to a virtual resolution of 1440p for both tests, I assume the resolution was changed before the test began, but it's not made clear on the results page. Also, they aren't "valid" results because it fails a time variance test, but nothing looked out of place visually, so take that with a grain of salt.
Click to expand...
lol that firestrike score is funny......maybe it was extreme or ultra test?:woot: lol my 280x scores WAY higher...j.k i know that cpu was really holding it back, but i never seen one so bottle necked before lol I think you need to redo that test
 
  • Like
Reactions: Nobu
like this
primetime said:
lol that firestrike score is funny......maybe it was extreme or ultra test?:woot: lol my 280x scores WAY higher...j.k i know that cpu was really holding it back, but i never seen one so bottle necked before lol I think you need to redo that test
Click to expand...
It's whatever the default is--I haven't paid for the full version. ;)
Reran the timespy test, results were about the same but this time it's a valid result, new run is on the left. In the physics test I was getting 14fps until 20s in, then 8fps until 50s, and 4-5fps for the remainder.
I also reran the firestrike test, but for whatever reason the system scan didn't happen, and it decided not to tell me until it finished running everything, didn't feel like running a third time. Scores looked the same as my old run, though.
 
primetime said:
lol that firestrike score is funny......maybe it was extreme or ultra test?:woot: lol my 280x scores WAY higher...j.k i know that cpu was really holding it back, but i never seen one so bottle necked before lol I think you need to redo that test
Click to expand...

you are looking at the overall 3dMark score which as you hinted you know it's badly affected by his CPU score, you have to look at the Graphics score which its 13.292. About ~46% Faster than my last FireStrike Result with a 280X which was 9096, so I think you are reading it wrong the score it's perfectly fine. I jumped early off that card so I really don't know how much optimizations it could have received but that score shouldn't be much more different than any other more recent and actual score from a 280X could you post one recent firestrike score please?.
 
just a FYI.....but overclocked rx480's should be in the 11K range in Firestrike under non-cpu bottle necked situations...So yea 6K range is REALLY held back.....its just a canned benchmark of course but i wonder if something is wrong. The timespy at least is in the correct ballpark but then again DX12 is supposed to benefit weaker cpus (so maybe thats why?) I bet your time spy score goes up over the coming months as AMD tweaks the drivers
 
Araxie said:
you are looking at the overall 3dMark score which as you hinted you know it's badly affected by his CPU score, you have to look at the Graphics score which its 13.292. About ~46% Faster than my last FireStrike Result with a 280X which was 9096, so I think you are reading it wrong the score it's perfectly fine. I jumped early off that card so I really don't know how much optimizations it could have received but that score shouldn't be much more different than any other more recent and actual score from a 280X could you post one recent firestrike score please?.
Click to expand...
I scored 8 598 in Fire Strike
 
see? everything is good in Nobu's Result, 9556 from your OC'd 280X to 13292 from nobu's stock 480 the big difference is the CPU but about 40% faster than OC'd 280X just more or less in line as how they compare in real world gaming.
 
I think it's just my cpu showing it's age, honestly. It's about 3 or 4 years old, been overclocked, moved between motherboards 5+ times, overvolted, undervolted, etc., and it wasn't a spectacular cpu when it was released. Searched "firestrike a10-6800k" in google, first result shows 17fps in the physics test, second shows 10fps. Just the way it is, I guess.
 
Does anyone have a Mono Plus? Should I be worried about fact I can't fit the ramsinks on :/?

edit: Outside of my worry about the ram temperature, I was able to get a stable 1340Mhz overclock and that gave me 4082 timespy (I scored 4 082 in Time Spy) and 10694 (I scored 10 694 in Fire Strike) Firestrike. Both scores are a nice contrast to 2311 and 7702 respectively that I got with 7970 BE OC.

The core speed was also a steady 1340 most of maps too.
 
Last edited:
You must log in or register to reply here.
Back
Top