Oh RX 480 might burn your PCIe slot even on stock, because its drawing 82W out of it... OCs RX 480 so it draws ~310W, 155W out of PCIe and it works. Really most reviewers have no Idea about specs, limits or electronics. Otherwise the nVidias Founders would have been a scandal as well. nVidia has better chips than AMD, but the Founders PCB is horrible and the crappy VRM can't feed the 1080 chip under temps, so it throttles... Another example is the higher energy conumption 165W on RX 480 instead of 150W TDP adverticed. All manufacturers haven't got the fans in their TDPs (not AMD, not nVidia and not Intel). The crappy Fan on the RX 480 is rated at 15W. So the card is probably running exactly on 150W with Powerlimit on +0%. Reference cards from AMD and nVidia are both terrible like always, but which reviewer got that? It's really sad to see such a lack of knowledge on TH-cam. The end user might have forgotten what he learned in physics, but the guys making their money with reviews should remember it, it's basically their job. And a bit of research with the web is so easy, at least if you're not searching fo answers in the gaming community, they seem to be immune to facts.
yo maybe you want to read this first, because it seems that you deluded by your fanboyism oc.jagatreview.com/2016/07/klarifikasi-dari-amd-mengenai-radeon-rx480-pcie-power-draw-issue/ run at 1.425 Ghz on the cheap motherboard for 7 hours stress test no problem, the reviewer is lucky n00b btw, he is world class overclocker
The first step anyone wanting to use a RX 480, is to check the motherboard's PCI-E specification to see if it's 3.0 or higher, the latest version if 4.0 slots, which is likely ONLY found in high end motherboards (mobo manufacturer uses 'spare' and surplus parts for mainstream boards because it's cheap).
PCI 2.0 should also be able to handle it. Also the AMD fix added one option to enable compatibility mode which throttles power so it doesn't exceed the 150 TDP in case you are too worried about it.
Excellent video, i would love to see an advanced overclocking video for the GTX 1080 without any voltage limitations. I have seen several videos that say that the voltage is limited on the 1080 and it would be very cool to see what the GTX 1080 could do without any limitations.
you have just won a subscriber, this video was the best i have seen in a while, and i watch frequently LTT and HWC and all the major channels, what are you doing is pretty amazing! keep it up!
Yeah! Someone saying that 480 isn't causing problems! There are probably 3-4 people around the world that breaked a motherboard. It was probably someone trying to use 480 on a 2006 mobo or a crypto currency miner using 5-6 cards/mobo. +1 sub
brilliant video its nice to hear from someone who actually has a good understanding of the technical aspects of what he is talking about and straight to the point.
This dude really nows what he is talking about. Damn nice video keep up the work. You should make more videos. all the information and concepts were very well explained.
Wow, die Fakten einfach 'mal in unter 15min auf den Tisch geknallt... sehr geil! I'm just gonna hope pcper, redgamingtech, gamersnexus, jayz2cents, techyescity&Joker, Paul and his Hardware, Bang4buckgamer, tek syndicate and even the guys @ Linus media group take and notice of this... techoftomorrow seems done after breaking the nda for a few hours, aren't they? Roman, keep the real stuff rockin' :-)
I did something similar for my RX480 (bad 70%ish asic quality) I used a bios editor (SRBPolaris) that set the voltage by messing with the I2C, unlike polaris bios editor who just change the pstate and voltage past 1.15 don't set. 1510Mhz@1.35V, TDC limit set to 170A, the other limit to something like 260W, stay at about 60ºC, thank to my custom loop :) Note that having the voltage set that way, prevent all userspace tool to work. Changing the clock with afterburner, even just 1mhz down or up, crash the system.
It's the first video I watch from you, but I must say Im really impressed with your knowledge about overclocking. I mean, compared to you the "popular" youtubers dont know a thing about overclocking. I really like these kind of videos!
You looked like you knew what you are doing. Then you gave us the firestrike score, instead of the Graphics score which is the actual GPU performance instead of the whole system. 15300 Graphics score is above a stock R9 Fury, and way above a stock GTX 980 which starts at ~13500.
Graphics score is not everything :) Combined test is really hard on the GPU and not limited by the CPU. Also with a slow CPU you will also limit the graphics score.
Yes. But not everyone has the same CPU and the total score is also influenced by the Physic score, which is purely a CPU test . Graphics score is the only score that varies based only on the power of the graphics card. Why don't you test using a i7 6950x then? Are you trying to make the card look bad? Obvious sarcasm.
I used the 6950X on my previous OC Video. The total score is the same. Physics score is weighted really low and has almost no impact on a card like this
/fs/3200194 GTX 970 + i3 /3dm/4515846 GTX 970 + i7 Both cards at comparable clock rates. One has 10k in Firestrike, with Graphics score of 11.5k The other one has 8k in Firestrike, with a Graphics score of 11.9k. Are you telling me that CPU has nothing to do with it? You are testing a graphics card, not review-ing your test bench. Stick to graphics scores.
Yeah Fire Strike GPU score please, and if possible a passmark GPU score. Graphics score is everything in this context. A combined system score is not relevant
Nice video der8auer. I've used your info on overclocking.guide on the shunt resistor mod on my gtx 770's and it works out great. They been running on the reference coolers at 1300 MHz for over a year now no issues. Only use solder on the back of the cards, the max TDP only gets to 65% leaves a lot of headroom for a water block. Keep up the good work you know you stuff man.
It's scary, how close you came in your prediction. My Sapphire RX480 Nitro+ 4GB, runs at CORE: 1400MHz, MEM: 2000MHz with an added plus to core voltage of +78mV(stable). As soon as I try to push it to 1410 or above, it fails... Nice video - cheers and thumbs up, from Denmark ;)
Great analysis - Really impressed with your efforts and the way you imparted your knowledge so showing the limitations for the RX 480 overclocking to 1500Mhz - Many thanks
Great video. I'm waiting on my Sapphire Nitro pre-order atm. I suppose keeping it a constant 1400mhz should be fine? Or should I only clock up b4 gaming?
Yea, but from what I can gather from screenshots you can OC by phases in Wattman so if you just OC on the higher phases you can actually keept it a low clock values in the lower ones (browsing, etc), ye?
you are doing this for 5 years and you dont have a ton of subscribers???? wow man you are strong! a lot of youtubers get 100k subscribers in not more than a year! You are awesome!
thanks mate, really appreciate someone who gives this much detail on how extrem oc works. i probably will never do anything like this, because i just dont have the money to risk a dead 1080 ( and doing something other then the top end cards is stupid for normal usage cause you can just drop a few € more to have the same result as a crazy oc midrange card. ) but knowing how this stuff works get my inner enthusiast going. only thing i notice it, i less enjoy playing games, then running them for seeing my performance these days :(
You guys are crazy! Thanks for pushing things to the limit and debunking some of the attention seeking people who publish misleading info! Keep clocking! Also noticed that you publish in two languages; should have assumed that :). Nice touch and greatly appreciate the English translation. My German is too rusty.
der8auer - thank you for your videos and explanations and for you unlocked air BIOS. I am running a 240mm AIO water loop on my card and I want to know what is a good and safe voktage for the GPU. Is 1.3v ok under this conditions? I am around 43C under load with the AIO cooler. Also, regarding VRM - 90C measured with laser thermometer - is this ok compared to the rated 125C for the VRMs? Thank you!
Very nice video. RX 480 might not scale so well, but it's not that important, most people don't overclock. Have you tried undervolting this card as far as possible? What are lowest stable voltages and power draw with such setup?
thanks for all the info m8! really interesting video about maximizing the rx 480 potential OC! hope we see some strong stuff from asus this time and hit that 1450 gold mhz you talked about!
can you test the FPS in various games with that 1.5ghz 480? it'd be interesting to see how this one performs against other cards in a real world scenario.
der8auer other tech youtubers said that these synthetic tests only strain one or 2 parts (like tessalation or rendering etc) of a gpu therefore not replicating a real world scenario. or is 3dmark diferent?
That simply makes no sense. 3DMark is using the GPU the same way a game does. Also for example GT1 is very GPU heavy while GT2 has a higher load on the memory. For a real analysis you have to take several applications into account. Some games work better with Nvidia, some games better with AMD. If I'd take one random game for comparison it has the same validity as 3DMark - not much but it's a start.
see, I used the Arctic Accelero Hybrid III 120mm and flashed tie vbios to your unlocked one and I use Asus gpu tweak II, and I set the voltage to 1.3v, but when I check in HWinfo64, it measures quite a bit lower for the gpu core. is this reading affected by the 80mm fan I have plugged into the pwm connection on the gpu PCB?
so is the voltage controller programmable in such a way where AMD can release a BIOS update to shift the load balancing around as I understand buildzoids video they ram and vcore and wired to physically separate planes I am not sure if program-able load balancing is even possible ?
The RAM is hardwired to one plane. The Vcore is split 3:3 between the PCIe slot and the external connector. So the latter one can be rebalanced within a broad range.
someone on OCN did confirm that the controller is programmable in such a way where it can be made to shift the load www.overclock.net/t/1604979/a-temporary-fix-for-the-excess-pci-e-slot-power-draw-for-the-reference-rx-480-cards
Not bad... Not bad.... I have the world record RX 580 /w 9900K. It's the fastest aircooled RX 580 in the world. I absolutely promise that. It does 1570mhz/2500mhz stable (UberTimings 3.4)1.430v GPU and 950mV IMC. It has copper shims installed between the memory and heating, liquid metal on the core, and thermalright TFX putty on the VRM. I'd put it up against anyone's watercooled RX 580 as well. It's so good, I'm probably going to watercool it just so I can make sure it lasts. 75c after gaming for a few hours. I'd like to get it down to 50c or less and maybe aim for 1600mhz @ 1.5v.
Wow! That's cool, I just got my GPU only water cooler work, I'd like to see how far I can get on RX480. Reference Card will arrive soon, it's for competition, and I'm waiting for Nitro+ launch in China, can't wait but the only thing I can do is waiting...
Thanks, you have access to some incredible knowledge here. As I understand it the people at PC Perspective were claiming the power coming through the PCIe (each individual pin) was exceeded the allowed / rating. They were indicating that there could be damage to the Board as a result of prolonged use and or worse (saying the protection for such PCIe slots might not kick in for one lane violation, only if all lanes exceeded the supposed limitation protection they were claiming. So in other words they were pretty much indicating melting / fire etc. Do you have any comments about what they say / did?
Usually there is no protection for PCIe slots. There is just a straight trace going from 24-Pin to the PCIe slot. In theory it can get hot and burn. If you keep the card at stock I don't see any problems tho. Overclocking might always exceed several limits so AMD can't be in charge here.
Disappointed , this means the card is not going to be a good overclocker, the mods and extreme voltage plus very expensive water cooling used here far exceed what any aib card will do plus this setup is pushing the card too hard for a retail card to even come close. I don't expect to see a 1400mhz aib card unless it is oc to an inch of its life..
I overclocked my MSI Gaming X 4GB from 1303MHz to 1400MHz on stock voltage, completely stable. My stock voltage never goes above 1.1250V. I was able to get 1450MHz stable with +30mV, but it can definitely go higher, I'm just not comfortable with such voltage for heavy use.Do I have a good overclocking card, because wherever I read people have trouble getting much above 1.4GHz, even with extra voltage?
Could you provide some benchmarks for performance with factory settings versus your custom overclocking? I'd love to see what kind of Fps increases this could supply.
i'd like to point out a technical error. when you said that SLI/Crossfire motherboards that are meant to handle more than one card won't have a problem force feeding more than the maximum 66W @ 12V the PCI-E specs allows is not true. PCI-E slots have five 12V leads that are rated at 1.1Amp each, providing a maximum of 66W (5 x 1.1A x 12V = 66W). we're ignore the 3.3V line. so while the motherboard may indeed be able to provide power for as many as 7 PCI-E slots, or 462W @ 12VDC, that doesn't mean that ANY ONE PCI-E slow can provide all that power. I'm sure a board like the Asus X99 deluxe can handle the RX480 even at 1500MHz, but people who are buying $200 GPUs are not in the market for $300-$400 motherboards. people with $200 GPUs will be rocking $100 motherboards that cost about $15 to build. there is no fiscal room on a product like that to go way up and above the mandatory specs.
Seems even cheapcomponents themselves are overspec'd by oem's (from who these components are sourced by board makers). Pci-e spec or platform spec to be more precise has little bearing on the actual capability of individual components like the pcie-slot connector.
Klar lässt sich diese Karte nicht so gut übertakten, wie die gtx 1070 bzw. 1080, aber dafür, dass sie mit 8gb nur 270€ kostet ist das schon mal ein großer Schritt in die richtige Richtung ,wenns ums übertakten geht, bei AMD. Danke dass du so ehrlich bist und eine Objektive Meinung uns mitteilst :) Denn es gibt fast nichts schlimmeres als Enttäuschung
wenn man etwas länger wartet und die yield rate besser wird kann man evtl. mit 1350-1400 stabil betreiben bei einer guten luftkühlung das währe echt super
Em...I checked from intel website, it explains that "All intel motherboard only supply 75W on PCIE slot". It means each PCIE slot can only supply 75W. Multi GPU get around 150W is no problem, only if each pcie slot draw lower than 75W.
Its like I guessed. The slit provided the r2c parameters to use with Afterburner in order to move the bigger ammount of the powerdraw to the Peg. www.overclock.net/t/1604979/a...e-rx-480-cards Now it should be totally clear that amd shouldn't have a Problem to Balance the powerdraw in a different way. Either by a driver which performs the i2c commands at startup or with a new bios. I'd prefer the last. But not eb is willing to risk his card flashing a new bios even if the driver install routine does it on its own You could even use i2c commands for your Overclocking and surpassing AMDs boundaries. Would live to see you trying to get the 480 to the limits without any hard mod.
Ist es denn in der Regel so, das sich PCI-E-Slots eine große Powerlane teilen, die dann dementsprechend für 75Watt*Anzahl der PCI-E Slots ausreicht? Dann wäre ja ne einzelne RX wirklich kein Problem. Aber grundsätzlich finde ich die Vorstellung schon bedenklich, weil ich mir denke das die stromführenden Kontakte auf nem Mainboard ja schon deutlich dünner sein müssen als z.B. ein 6-Pin-Powerkabel - ob sich da also nicht an ungünstigen Stellen im Mobo zusätzliche Wärme entwickeln könnte? Das könnte ja schon irgendwo Ausfälle begünstigen. Naja, würde gerne deine Meinung dazu hören. Gutes Video, weiter so!
It's a very silly excuse to say that PCIe power draw isn't an issue just because a motherboard has multiple PCIe slots. Just because a board might be able to output >150W across /all/ of the slots at once, doesn't mean it's okay to draw above-spec on any individual slot. That's why the specification talks about power draw for one slot, not total! - And at any rate, what if you had multiple 480's? Then all of the 16x slots could be drawing current over-spec, and that's even worse than having just one slot over-draw! Furthermore: Not all mobos have multiple PCIe 16x slots. It also depends entirely on the quality of the motherboard: An older, aged, budget motherboard will struggle MUCH more with over-spec power draw, compared to any new or high end board.
1- how much real power consumption at 1500 1450 exp and what the lower voltage you can set at let say 1400 1450 1480 cause 1.3 for vga too much max 1.22 1.25 2- what about unlocking cores 40 cu ? or 4gb to 8gb 3-can you test it in quantum break cause heavy game and to see shaders decrease effect vs r9 390 + the highest oc effect in it
I assume it's been said many times before, but thank you so much for putting out videos in english as well.
you're welcome :) Trying my best to share the info I have. This time English version was first and German will be up tomorrow.
This guy makes reviewers look like they have no idea what they're doing :S
my thoughts exactly
Oh RX 480 might burn your PCIe slot even on stock, because its drawing 82W out of it... OCs RX 480 so it draws ~310W, 155W out of PCIe and it works.
Really most reviewers have no Idea about specs, limits or electronics. Otherwise the nVidias Founders would have been a scandal as well. nVidia has better chips than AMD, but the Founders PCB is horrible and the crappy VRM can't feed the 1080 chip under temps, so it throttles...
Another example is the higher energy conumption 165W on RX 480 instead of 150W TDP adverticed. All manufacturers haven't got the fans in their TDPs (not AMD, not nVidia and not Intel). The crappy Fan on the RX 480 is rated at 15W. So the card is probably running exactly on 150W with Powerlimit on +0%.
Reference cards from AMD and nVidia are both terrible like always, but which reviewer got that? It's really sad to see such a lack of knowledge on TH-cam. The end user might have forgotten what he learned in physics, but the guys making their money with reviews should remember it, it's basically their job. And a bit of research with the web is so easy, at least if you're not searching fo answers in the gaming community, they seem to be immune to facts.
yo maybe you want to read this first, because it seems that you deluded by your fanboyism
oc.jagatreview.com/2016/07/klarifikasi-dari-amd-mengenai-radeon-rx480-pcie-power-draw-issue/
run at 1.425 Ghz on the cheap motherboard for 7 hours stress test no problem, the reviewer is lucky n00b btw, he is world class overclocker
5:26 Best part..."...might just shut some media up."
This should be at the top!
The first step anyone wanting to use a RX 480, is to check the motherboard's PCI-E specification to see if it's 3.0 or higher, the latest version if 4.0 slots, which is likely ONLY found in high end motherboards (mobo manufacturer uses 'spare' and surplus parts for mainstream boards because it's cheap).
PCI 2.0 should also be able to handle it. Also the AMD fix added one option to enable compatibility mode which throttles power so it doesn't exceed the 150 TDP in case you are too worried about it.
Awesome video. Thanks for showing what the polaris Gpu is actually capable of and shutting up these tech reviewers.
Excellent video, i would love to see an advanced overclocking video for the GTX 1080 without any voltage limitations. I have seen several videos that say that the voltage is limited on the 1080 and it would be very cool to see what the GTX 1080 could do without any limitations.
Deutsches Video folgt morgen :)
Das ist doch kein GTX980 Niveau. Nur eine GTX980 @1500Mhz erreicht ungefähr denselben Wert bzw etwas höher.
Also wenn, dann eine 980@OC Niveau.
Haare spalten
*****
Hahahaha kann sein :D
Wann genau? ^^
Finde es toll das du dir mit den englischen Videos so viel mühe gibst alleine deswegen gibt es von mir ein Thumbs up
you have just won a subscriber, this video was the best i have seen in a while, and i watch frequently LTT and HWC and all the major channels, what are you doing is pretty amazing! keep it up!
Yeah! Someone saying that 480 isn't causing problems!
There are probably 3-4 people around the world that breaked a motherboard. It was probably someone trying to use 480 on a 2006 mobo or a crypto currency miner using 5-6 cards/mobo.
+1 sub
Very professional presentation, you guys have mastered the art of overclocking well. i really enjoyed this insight into the RX480.
brilliant video its nice to hear from someone who actually has a good understanding of the technical aspects of what he is talking about and straight to the point.
A overclocked RX480 slightly edging out the R9 Nano in performance... for a sub $250 card that's incredible.
i get your point but he did custom water cool it but yeah hopefully aftermarket cards are just as good!
i think the sapphire nitro could do it.
Well if I buy this card I'd probably water cool it too. EK already has full cover blocks for it
ccricers wait for the 490 or 480x
u mean 485 :P
FINALLY, some info from somebody that knows a thing or two about how things work. Thank you? +1
This dude really nows what he is talking about. Damn nice video keep up the work. You should make more videos. all the information and concepts were very well explained.
This guy is so much better than the pretentious reviewers out there. He reminds me of my electronics teacher in school...
thanks :D
Man, you put all the other reviewers to shame. Serious great knowledge and great video. Keep creating awesome videos.
thanks :) appreciate that
Wow, die Fakten einfach 'mal in unter 15min auf den Tisch geknallt... sehr geil! I'm just gonna hope pcper, redgamingtech, gamersnexus, jayz2cents, techyescity&Joker, Paul and his Hardware, Bang4buckgamer, tek syndicate and even the guys @ Linus media group take and notice of this... techoftomorrow seems done after breaking the nda for a few hours, aren't they? Roman, keep the real stuff rockin' :-)
I love the low level solution you used to overclock. Subscribed!
I did something similar for my RX480 (bad 70%ish asic quality)
I used a bios editor (SRBPolaris) that set the voltage by messing with the I2C, unlike polaris bios editor who just change the pstate and voltage past 1.15 don't set.
1510Mhz@1.35V, TDC limit set to 170A, the other limit to something like 260W, stay at about 60ºC, thank to my custom loop :)
Note that having the voltage set that way, prevent all userspace tool to work. Changing the clock with afterburner, even just 1mhz down or up, crash the system.
Super advanced, glad someone out there is doing stuff like this.
It's the first video I watch from you, but I must say Im really impressed with your knowledge about overclocking. I mean, compared to you the "popular" youtubers dont know a thing about overclocking. I really like these kind of videos!
thanks :) appreciate that
You looked like you knew what you are doing. Then you gave us the firestrike score, instead of the Graphics score which is the actual GPU performance instead of the whole system.
15300 Graphics score is above a stock R9 Fury, and way above a stock GTX 980 which starts at ~13500.
Graphics score is not everything :) Combined test is really hard on the GPU and not limited by the CPU. Also with a slow CPU you will also limit the graphics score.
Yes. But not everyone has the same CPU and the total score is also influenced by the Physic score, which is purely a CPU test . Graphics score is the only score that varies based only on the power of the graphics card.
Why don't you test using a i7 6950x then? Are you trying to make the card look bad? Obvious sarcasm.
I used the 6950X on my previous OC Video. The total score is the same. Physics score is weighted really low and has almost no impact on a card like this
/fs/3200194 GTX 970 + i3
/3dm/4515846 GTX 970 + i7
Both cards at comparable clock rates.
One has 10k in Firestrike, with Graphics score of 11.5k
The other one has 8k in Firestrike, with a Graphics score of 11.9k.
Are you telling me that CPU has nothing to do with it?
You are testing a graphics card, not review-ing your test bench. Stick to graphics scores.
Yeah Fire Strike GPU score please, and if possible a passmark GPU score. Graphics score is everything in this context. A combined system score is not relevant
Nice video der8auer. I've used your info on overclocking.guide on the shunt resistor mod on my gtx 770's and it works out great. They been running on the reference coolers at 1300 MHz for over a year now no issues. Only use solder on the back of the cards, the max TDP only gets to 65% leaves a lot of headroom for a water block. Keep up the good work you know you stuff man.
I'm glad to see that I can get a better overclock out of an RX 480 with a little modding. Thanks!
It's scary, how close you came in your prediction.
My Sapphire RX480 Nitro+ 4GB, runs at CORE: 1400MHz, MEM: 2000MHz with an added plus to core voltage of +78mV(stable).
As soon as I try to push it to 1410 or above, it fails...
Nice video - cheers and thumbs up, from Denmark ;)
Guess i should feel happy with my MSI gaming X model runing @ 1440mhz/2140mem , using a 240mm AIO :p
What a great video! Someone who actually knows what he's talking about, unlike a lot of channels on here
thanks man
Great analysis - Really impressed with your efforts and the way you imparted your knowledge so showing the limitations for the RX 480 overclocking to 1500Mhz - Many thanks
Very thorough review on OC'ing. Thanks you cleared up a lot of info for me to be able to OC safely
awesome video! thank you, I've been looking for a detailed properly done water cooled overclock just to see the limits of the RX 480.
graphics score was 15,363. that is huge.
weird its not huge in % but boost performance better i don't if smaller chip make oc values better
Naah ... not really. Guy with 1360 MHz did 15,526 :)
www.3dmark.com/fs/9048815
Notice that he has modified tesselation for that score.
....with tessellation disabled
No, the second benchmark with the 16 612 score is without tesselation.15,363 points were made with standard settings
I think it would've been better if you added some game benchmarks or any chart to do some comparisons of performance
Great video. I'm waiting on my Sapphire Nitro pre-order atm. I suppose keeping it a constant 1400mhz should be fine? Or should I only clock up b4 gaming?
That's how I treat my overclocks. There's really no reason to run an OC while on the desktop or browsing TH-cam.
Yea, but from what I can gather from screenshots you can OC by phases in Wattman so if you just OC on the higher phases you can actually keept it a low clock values in the lower ones (browsing, etc), ye?
Oh, I don't know. I'm not familiar with Wattman.
The gpu downclock if its not running hi demanding 3d load, no need to adjusting the setting for light work.
you are doing this for 5 years and you dont have a ton of subscribers???? wow man you are strong! a lot of youtubers get 100k subscribers in not more than a year! You are awesome!
I've been doing XOC for 10 years but only started active on youtube about a year ago :D
+der8auer you have just won a subscriber for life!
+der8auer keep rolling out the good content!
thanks mate, really appreciate someone who gives this much detail on how extrem oc works. i probably will never do anything like this, because i just dont have the money to risk a dead 1080 ( and doing something other then the top end cards is stupid for normal usage cause you can just drop a few € more to have the same result as a crazy oc midrange card. ) but knowing how this stuff works get my inner enthusiast going. only thing i notice it, i less enjoy playing games, then running them for seeing my performance these days :(
Outstanding review. Deep review, using actual science and circuitry knowledge. subscribed and liked.
You guys are crazy! Thanks for pushing things to the limit and debunking some of the attention seeking people who publish misleading info! Keep clocking!
Also noticed that you publish in two languages; should have assumed that :). Nice touch and greatly appreciate the English translation. My German is too rusty.
Nice tehnical video, I'm glad I found your channel.
der8auer - thank you for your videos and explanations and for you unlocked air BIOS. I am running a 240mm AIO water loop on my card and I want to know what is a good and safe voktage for the GPU. Is 1.3v ok under this conditions? I am around 43C under load with the AIO cooler. Also, regarding VRM - 90C measured with laser thermometer - is this ok compared to the rated 125C for the VRMs? Thank you!
Very nice video.
RX 480 might not scale so well, but it's not that important, most people don't overclock.
Have you tried undervolting this card as far as possible?
What are lowest stable voltages and power draw with such setup?
To this day this video helped me squizze out an extra 6% currently lilmited to 1530 mhz at 1.3v
thanks for all the info m8! really interesting video about maximizing the rx 480 potential OC!
hope we see some strong stuff from asus this time and hit that 1450 gold mhz you talked about!
I've subscribed in hope you will do this with the GTX 1080
Can't wait to see what that is capable of
Great video btw
nice Vid again Roman :D really appreciate the inside views and interesting facts on the card :)
seems like elmores tool will be very popular ;> glad i have one in my tool box waiting for the right project.
Immer mehr von solchen videos.... einfach nur hammer!!!!
can you test the FPS in various games with that 1.5ghz 480? it'd be interesting to see how this one performs against other cards in a real world scenario.
you can expect the same FPS increase percentage-wise. 3DMark is just a 3D application similar to any other game.
der8auer other tech youtubers said that these synthetic tests only strain one or 2 parts (like tessalation or rendering etc) of a gpu therefore not replicating a real world scenario.
or is 3dmark diferent?
That simply makes no sense. 3DMark is using the GPU the same way a game does. Also for example GT1 is very GPU heavy while GT2 has a higher load on the memory. For a real analysis you have to take several applications into account. Some games work better with Nvidia, some games better with AMD. If I'd take one random game for comparison it has the same validity as 3DMark - not much but it's a start.
der8auer ok, thanks for clearing that up
What's nice is that 580 you can top with 1480-90 on good chips with 1.15volts :) and run it near silent on air. Good improvement in my book for AMD!
Nice heatsinks on the VRAM, how did it not catch fire?
Because they are specified up to 125°C and are only 100°C on load?
der8auer well if they are only 100c at load
I'm working on an article about power consumption and VRM temperature as well at the moment :) Will be online soon on overclocking.guide
der8auer Good stuff!
I must have missed the heatsinks on the VRM in the video, because I didn't see any.
Also i really appreciate this you saved us a lot of waiting and wondering.
Amazing video! now this is a true extreme overclocking! impressive by all means
good, german, no-bullshit video, makes me very happy - subscription!
Great video, reminds me more of a research paper/case study, but amazing!
thanks :)
Now this is seriously awesome =)
nice vid. can you do more of these hard mod vids for different cards pls? would be so nice to see it.
Looks like j4003 on the soldermask is an i2c header aswell. On pretty much all radeon cards. That 480 had it just to the right of the core at 4:29
see, I used the Arctic Accelero Hybrid III 120mm and flashed tie vbios to your unlocked one and I use Asus gpu tweak II, and I set the voltage to 1.3v, but when I check in HWinfo64, it measures quite a bit lower for the gpu core. is this reading affected by the 80mm fan I have plugged into the pwm connection on the gpu PCB?
oh and my core clock is at 1430MHz.
Awesome! Love your channel and the technical know how you have. Subbed!
Subbed, some quality OC work, thanks for sharing.
Thank you for educating many of these fools trying to say that the power delivery of the RX 480 card was flawed...idiots! Great video!
Wow. Amazing. I'm really impressed with you work, subscribed.
would you do any gaming benchmarks for this? :D
Dat hair tho, have you traveled from 2010 to 2016 to see how technology advanced?
wait wut?
hahahaha I lol'd hard on your comment xD
Nice video! Shared it in Overclockers.com on AMD RX 480 Review list thread!
thanks man :) appreciate that
you are legend.. keep up with good work. cheers from Croatia!
so is the voltage controller programmable in such a way where AMD can release a BIOS update to shift the load balancing around
as I understand buildzoids video they ram and vcore and wired to physically separate planes I am not sure if program-able load balancing is even possible ?
The RAM is hardwired to one plane. The Vcore is split 3:3 between the PCIe slot and the external connector. So the latter one can be rebalanced within a broad range.
the question is whether or not the controller can actually do that
someone on OCN did confirm that the controller is programmable in such a way where it can be made to shift the load
www.overclock.net/t/1604979/a-temporary-fix-for-the-excess-pci-e-slot-power-draw-for-the-reference-rx-480-cards
Not bad... Not bad....
I have the world record RX 580 /w 9900K.
It's the fastest aircooled RX 580 in the world. I absolutely promise that. It does 1570mhz/2500mhz stable (UberTimings 3.4)1.430v GPU and 950mV IMC.
It has copper shims installed between the memory and heating, liquid metal on the core, and thermalright TFX putty on the VRM.
I'd put it up against anyone's watercooled RX 580 as well.
It's so good, I'm probably going to watercool it just so I can make sure it lasts. 75c after gaming for a few hours.
I'd like to get it down to 50c or less and maybe aim for 1600mhz @ 1.5v.
Wow! That's cool, I just got my GPU only water cooler work, I'd like to see how far I can get on RX480. Reference Card will arrive soon, it's for competition, and I'm waiting for Nitro+ launch in China, can't wait but the only thing I can do is waiting...
Excellent video, keep up the great work!
Thanks, you have access to some incredible knowledge here. As I understand it the people at PC Perspective were claiming the power coming through the PCIe (each individual pin) was exceeded the allowed / rating. They were indicating that there could be damage to the Board as a result of prolonged use and or worse (saying the protection for such PCIe slots might not kick in for one lane violation, only if all lanes exceeded the supposed limitation protection they were claiming. So in other words they were pretty much indicating melting / fire etc. Do you have any comments about what they say / did?
Usually there is no protection for PCIe slots. There is just a straight trace going from 24-Pin to the PCIe slot. In theory it can get hot and burn. If you keep the card at stock I don't see any problems tho. Overclocking might always exceed several limits so AMD can't be in charge here.
Great Video, thanks for the information.
Exactly what i was looking for.
Awesome work der8auer, Thank you.
Do you know if any water-cooling blocks for the 480?
full cover blocks from EKWB will be available soon
That chair looks gloriously comfortable. What kind is it?
DXRacer :)
Gut gemacht!
Disappointed , this means the card is not going to be a good overclocker, the mods and extreme voltage plus very expensive water cooling used here far exceed what any aib card will do plus this setup is pushing the card too hard for a retail card to even come close. I don't expect to see a 1400mhz aib card unless it is oc to an inch of its life..
wow! i love your videos. so technical. i got a question, whats the max safe voltage for rx 480 for daily use?
Thanks. I have learned something today.
nice work dude
I overclocked my MSI Gaming X 4GB from 1303MHz to 1400MHz on stock voltage, completely stable. My stock voltage never goes above 1.1250V. I was able to get 1450MHz stable with +30mV, but it can definitely go higher, I'm just not comfortable with such voltage for heavy use.Do I have a good overclocking card, because wherever I read people have trouble getting much above 1.4GHz, even with extra voltage?
Could you provide some benchmarks for performance with factory settings versus your custom overclocking? I'd love to see what kind of Fps increases this could supply.
You know what would be cool?
Have the separate PCB be integrated into the main GPU PCB.
i'd like to point out a technical error. when you said that SLI/Crossfire motherboards that are meant to handle more than one card won't have a problem force feeding more than the maximum 66W @ 12V the PCI-E specs allows is not true. PCI-E slots have five 12V leads that are rated at 1.1Amp each, providing a maximum of 66W (5 x 1.1A x 12V = 66W). we're ignore the 3.3V line.
so while the motherboard may indeed be able to provide power for as many as 7 PCI-E slots, or 462W @ 12VDC, that doesn't mean that ANY ONE PCI-E slow can provide all that power. I'm sure a board like the Asus X99 deluxe can handle the RX480 even at 1500MHz, but people who are buying $200 GPUs are not in the market for $300-$400 motherboards. people with $200 GPUs will be rocking $100 motherboards that cost about $15 to build. there is no fiscal room on a product like that to go way up and above the mandatory specs.
Seems even cheapcomponents themselves are overspec'd by oem's (from who these components are sourced by board makers). Pci-e spec or platform spec to be more precise has little bearing on the actual capability of individual components like the pcie-slot connector.
Klar lässt sich diese Karte nicht so gut übertakten, wie die gtx 1070 bzw. 1080, aber dafür, dass sie mit 8gb nur 270€ kostet ist das schon mal ein großer Schritt in die richtige Richtung ,wenns ums übertakten geht, bei AMD.
Danke dass du so ehrlich bist und eine Objektive Meinung uns mitteilst :) Denn es gibt fast nichts schlimmeres als Enttäuschung
BuildZoid ROCKS
wenn man etwas länger wartet und die yield rate besser wird kann man evtl. mit 1350-1400 stabil betreiben bei einer guten luftkühlung das währe echt super
Em...I checked from intel website, it explains that "All intel motherboard only supply 75W on PCIE slot". It means each PCIE slot can only supply 75W. Multi GPU get around 150W is no problem, only if each pcie slot draw lower than 75W.
Cool! 4096 Fury X shaders at 1,6 would be fun 😎
you mean vega. nvidia gonna kill it as usual, but dx 12 and vulkan are the secret weapon from amd :)) not tomb raider piece of gamework
nvidia is not going to kill vega
trust meh
@The-BGR Spot trust you? a random nobody on the internet. suuuuuuure....
solomonshv either me, or the other random guys on the internet
why did you not quote for example - nhozdien, but you say that to me ?
lil nvidia fan
The-BGR Spot did you read my comment slowly :))
Please do a video when bios mods can be done!
''ohh no its gona fry your mobo''
Glorious review. Subbed.
Top overclock tumbs up! gj
great video , the information was very interesting
nice work like gamer nexus !
afterburner supports i2c controll via software with commandline switches. maybe you could try to reach similar results without the external pcb.
Could be that the driver/BIOS will not allow it or reset it
Its like I guessed. The slit provided the r2c parameters to use with Afterburner in order to move the bigger ammount of the powerdraw to the Peg.
www.overclock.net/t/1604979/a...e-rx-480-cards
Now it should be totally clear that amd shouldn't have a Problem to Balance the powerdraw in a different way. Either by a driver which performs the i2c commands at startup or with a new bios. I'd prefer the last. But not eb is willing to risk his card flashing a new bios even if the driver install routine does it on its own
You could even use i2c commands for your Overclocking and surpassing AMDs boundaries. Would live to see you trying to get the 480 to the limits without any hard mod.
Ist es denn in der Regel so, das sich PCI-E-Slots eine große Powerlane teilen, die dann dementsprechend für 75Watt*Anzahl der PCI-E Slots ausreicht?
Dann wäre ja ne einzelne RX wirklich kein Problem. Aber grundsätzlich finde ich die Vorstellung schon bedenklich, weil ich mir denke das die stromführenden Kontakte auf nem Mainboard ja schon deutlich dünner sein müssen als z.B. ein 6-Pin-Powerkabel - ob sich da also nicht an ungünstigen Stellen im Mobo zusätzliche Wärme entwickeln könnte? Das könnte ja schon irgendwo Ausfälle begünstigen.
Naja, würde gerne deine Meinung dazu hören. Gutes Video, weiter so!
It's a very silly excuse to say that PCIe power draw isn't an issue just because a motherboard has multiple PCIe slots. Just because a board might be able to output >150W across /all/ of the slots at once, doesn't mean it's okay to draw above-spec on any individual slot. That's why the specification talks about power draw for one slot, not total! - And at any rate, what if you had multiple 480's? Then all of the 16x slots could be drawing current over-spec, and that's even worse than having just one slot over-draw! Furthermore: Not all mobos have multiple PCIe 16x slots.
It also depends entirely on the quality of the motherboard: An older, aged, budget motherboard will struggle MUCH more with over-spec power draw, compared to any new or high end board.
Nice video.I was really hoping for 1500 mhz from custom cards really tho.
Im kinda dissapointed.
Dude had builted an (monstrosity
Mach doch auch so ein video über die 1080 oder 1070, würde mich sehr interresieren wie weit die sich von einem pro pushen lassen :)
1- how much real power consumption at 1500 1450 exp
and what the lower voltage you can set at let say 1400 1450 1480 cause 1.3 for vga too much max 1.22 1.25
2- what about unlocking cores 40 cu ? or 4gb to 8gb
3-can you test it in quantum break cause heavy game and to see shaders decrease effect vs r9 390
+ the highest oc effect in it
they can try bios hack to find out if polaris 10 does have 40cu, but it's just rumor.
hast du irgendwas gehört wann die custom karten kommen? ich kann nicht warten...^^