He means just the slot, it's in the video posted on the home page of HardOCP.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
AMD Radeon RX 480 Video Card Review @ [H]
- Thread starter FrgMstr
- Start date
He means just the slot, it's in the video posted on the home page of HardOCP.
Caveat, the 126W pull through the slot is with the PL set to +50%. It *shouldn't* be pulling evenly from the PCIe slot, but it looks like it is, instead of drawing all overlimit power through the 6pin. It would be nice to see what the power draw is with the power limit set to default instead of +50%.
This is called making things worse. Tell it pull more power and then run furmark. Of course it's not throttling under furmark. I think people are trying to burn their shit on purpose at this point to make a point. Lol
I checked my local NCIX here in Vancouver. The 8GB version of this is available for $329 and most of them say its a sale price from the regular price. I can also get an Asus Strix GTX 970 for $339.... I will buy the vendor with the better value usually and this isn't the great value proposition the article states (here anyways) And I'm thinking to myself that Nvidia is going to release a 1060 soon that will likely be 20-30% faster than this (guessing) and cost $349 - $400 soon in Canadian.
Too bad video cards software do not display power usage with PCIe slot and power plug(s), better yet amps. Showing clearly when going over any spec which if you ignore and end up with a melted paper weight will be your fault. Then again you do push the specs for that extra bit of capability the hardware has like a dragster knowing it may not turn out well at the end of the quarter mile. Maybe we should have speed limit signs on dragster strips.
When you push the power limit to +50% what in the hell do people think power does? duh! If the card is in spec out of the box and someone over volts it and over clocks it yep it will go out of spec, software warns you as well that it will do just that. Maybe AMD should follow Nvidia and limit it to +20% like on the 1080 or +12% on the 1070 and give less choices to those wanting to push things.
Now why AMD choose to release a card right at the power rated power specs and then give options to basically blow it up (with warnings as if people will read them) is maybe a little bit naive. That is if the 480 is in spec out of the box with software designed for.
The real question are what options do AMD have at this point:
When you push the power limit to +50% what in the hell do people think power does? duh! If the card is in spec out of the box and someone over volts it and over clocks it yep it will go out of spec, software warns you as well that it will do just that. Maybe AMD should follow Nvidia and limit it to +20% like on the 1080 or +12% on the 1070 and give less choices to those wanting to push things.
Now why AMD choose to release a card right at the power rated power specs and then give options to basically blow it up (with warnings as if people will read them) is maybe a little bit naive. That is if the 480 is in spec out of the box with software designed for.
The real question are what options do AMD have at this point:
- Tell the truth, get a hundred random cards or more to independent labs to test
- Bios update
- Driver update (kill furmark clock speeds, actually the hardware should be able to respond to a software load but in this case no)
- Revision to the 480 - 8 pin connector as long as they can shift power from the pcie slot to the power connector more
- With the revision maybe a trade out program where folks can turn in their 6pin and get a 8pin card in return if they choose
- Kill all driver OCing options: let utilities from third party tackle that like what Nvidia does
- Restrict power options in the bios like what Nvidia does for example 20% and 12%
I never tought i was going to say this, but here it goes: AMD did not disappointed! We now have a DX12 card with 8GB of RAM for $239. If the Crossfire perfomance delivers, it will totally replace SLi GTX 970 as the cheapest 4k solution, even more so now that Freesync monitors are common and cheaper than g-Sync ones.
chameleoneel
Supreme [H]ardness
- Joined
- Aug 15, 2005
- Messages
- 7,601
They released a brand new software called Wattman (which integrates into CCC), for more control over power and overclocking. And the 480 is the poster child. They are going to do everything they can to fix this, without breaking the marketing.
TheLAWNoob
Limp Gawd
- Joined
- Jan 10, 2016
- Messages
- 330
Hey Kyle, try replacing the thermal paste on the RX 480 and see how much cooler it runs.
According to chiphell AMD is using toothpaste for TIM (not literally). 1 guy claimed 15C temp drop by re-applying TIM.
rx480换了硅脂温度爆降十几度。。。更新测试图 - 顶级图形卡 - Chiphell - 分享与交流用户体验的最佳平台 - 手机版 - Powered by Discuz!
According to chiphell AMD is using toothpaste for TIM (not literally). 1 guy claimed 15C temp drop by re-applying TIM.
rx480换了硅脂温度爆降十几度。。。更新测试图 - 顶级图形卡 - Chiphell - 分享与交流用户体验的最佳平台 - 手机版 - Powered by Discuz!
lolfail9001
[H]ard|Gawd
- Joined
- May 31, 2016
- Messages
- 1,496
You mean the screens where GPU-Z reports 85W power consumption on GPU in Furmark? 0/10
By who?
'Who' is determined by some long, arduous legal process... over a $200 video card.
Better to just return it to the store and buy something else.
Wow nice quality control. Worthless. I hope everyone is waiting for the 1060.
Last edited:
seen another site do it and it made no difference. I don't believe this crap about tm dropped temps 15c. Not happening lol. Unless there was a big change in ambient temperature.
pendragon1
Extremely [H]
- Joined
- Oct 7, 2000
- Messages
- 52,226
a few degrees maybe but not 15+ unless the heat sink was not seated properly to begin with. maybe 15c IF the card was old, dirty and the tim was dry...
They aren't planning on scaling Polaris up. Vega isn't the same thing just made bigger.
I guess Polaris is meant to be a pipe cleaner part. That can explain the lack of performance and high power consumption.
The 'AMD secret sauce' that they always have up their sleeve NEVER ends with the internet. NEXT YEAR they are always gonna destroy Nvidia once and for all. They have 10 cents in the bank and their products are clearly falling behind but there's always that secret sauce ace up the sleeve... it's just always next year. Heard this for decades with this company and the glory days and one-hit wonders are long over.
Never count anyone out! No matter what you may expect.
Has anybody tested and not found it?
But is there actually a testing?
Five years ago PCI-SIG approved equipment without any actual tests. Source
German site ComputerBase.de reached some nice results by simply undervolting the card. If I understand it correct they were even able to overclock more with the voltage reduced.
Just as with ESD most mobos (PCIe slots) won't die immediately. There's a great possibility of reduced life span though, and how much reduction you don't know until it dies.
Here are a few reports showing that it might be a matter of hours.
Most PSUs nowadays use a combined 6+2 pin plug. When used in 6-pin mode each of the four power cables (2xGND and 2x+12V in use) is to carry only 3.13A on average.
In 8-pin mode the cables need to handle a whopping 4.17A on average, so just there's a 33% extra.
Then consider that the external connectors and cables are substantially sturdier than the PCIe-slot and PCB tracings, and should be able to handle >15A without any problem. The real limitation for the external connector is the PSU's OCP.
* Overclocking does indeed increase the bus draw.
* Very few cards (of the one's I've seen tested) come close to the limit. Most cards that have an external connector are below 35W from the slot.
In the "used" market only?
I have yet to find a review of the 4GB RX480, which is now sold in Sweden at "only" $235 plus sales tax (about the same as GTX 970). (The 8GB version is $284 plus tax.)
I'm also eagerly awaiting reviews on the two versions of GTX 1060, which might hit the same price point as RX480 here in Sweden.
D
Deleted member 93354
Guest
And you know this how?
So you are telling me Vega isn't based around GCN with a different memory controller?
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
Different graphics IP block
D
Deleted member 93354
Guest
How different is the question. I seen no official statements or even rumors.
For as I can tell it's still GCN 1.4 and the ratio of SP and TU are the same. That indicates a similar stack. Other than the memory controller, that also implies a scale up with more SP and ROPs. There's no need to re architect from the ground up. Just adjust the memory controller and the scheduler to account for the memory speed differences. (The big difference being the memory controller) Looking at the performance bump from Tonga to Fiji appears similar to the scale up between Polaris and Vega will be (looking at AMD's performance growth graph)
If I were to hazard a guess we are getting:
- a small vega ( RX490) with GDDR5X
- and big vega (Fury) of HBM2.
Last edited by a moderator:
Well we don't know, lol, I would gander not that far away for Polaris though, cause I just don't see AMD doing to architectures at the same time, complicates driver builds 
I think they will both use HBM2, I don't see AMD's messing around with their stack like that by using two different memory controllers, unless Polaris (a bigger Polaris) is small Vega.
I think they will both use HBM2, I don't see AMD's messing around with their stack like that by using two different memory controllers, unless Polaris (a bigger Polaris) is small Vega.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
We know there are two gpus though, 10&11 and 10 is smaller one this time.
I dunno razor, I could see them having Vega 10 on g5x 384b, and Vega 11 is hbm2
From LinkedIn we have fairly credible evidence that Vega uses a new version of the graphics ip block, and I remember reading they would use in house developed cell libraries for the first time - can't find source though, I tried.
Anyway, if they do develop their own cell libs the chip has potential to have completely different performance characteristics, although considering its the first time and it's amd, there is potential for disaster too
I dunno razor, I could see them having Vega 10 on g5x 384b, and Vega 11 is hbm2
From LinkedIn we have fairly credible evidence that Vega uses a new version of the graphics ip block, and I remember reading they would use in house developed cell libraries for the first time - can't find source though, I tried.
Anyway, if they do develop their own cell libs the chip has potential to have completely different performance characteristics, although considering its the first time and it's amd, there is potential for disaster too
I just don't see the benefit of doing 2 separate buses that's all. And with P10's power consumption needs, if Vega isn't that far away from it, GDDR5x won't be good for it from a power point of view. They will have a tough hill to climb and to convince consumers to buy their products over nV.
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
Shouldn't it be prohibitively expensive for them to do what they did in past gens and compete with relatively bigger dies?
Pushing clocks for more perf, thus sliding out of ideal range
Polaris 10: "There's nothing hidden on that product to unlock" - interview
AMD confirms 36CU only, so no higher end part with 40 or 44 cu's on this thing..... What did Adorned say lol.?
AMD confirms 36CU only, so no higher end part with 40 or 44 cu's on this thing..... What did Adorned say lol.?
5150Joker
Supreme [H]ardness
- Joined
- Aug 1, 2005
- Messages
- 4,568
My question is why do people even give any credibility to that adoredtv guy? He's some nobody from what sounds like Ireland or Scotland that makes up pro-AMD b.s. The only ones I see worshipping at his altar are the often fooled and ultra gullible AyyyMD fans. Maybe I should start my own YouTube channel and make up a bunch of pro AyyyMD bs for subs and views...hmm
He was a longtime troll on various forums. One of his accounts at AT was "SiliconWars". He's definitely an AMD diehard, admits it himself. But he's your typical AMD fan. He only bought from them when they had "5850" or "7850" bang for buck cards. He was never a high end buyer from what I can tell looking at his old posts on AT. I don't watch his channel but I noticed in one of his videos he claimed that AMD was going to take the perf/watt crown with Polaris, which didn't happen. So he's delusional himself and probably believes what he spouts.
I honestly think he's crazy, just as this [H] poster concluded.
Interesting Article: Waiting on 20nm graphics cards from Nvidia and AMD? Dont bother
This is AdoredTV's account here on H jimbo75
Does that make me irregular AMD fan as I bought the High-End card from them and it was not even faster than what the competitor was offering
I had my reasons why I bought this card. The reason can be seen from my display setup on my rig.
Adored says dual gpu consoles as a way to win the pc war against Nvidia. Which is quite the flimsy limb to go out on, a fanboy's wetdream. Basically AMD's master plan to consolize gaming (yay?). He was all fired up about Polaris but no one cared about performance we already had. People want to see 1080s and insane performance 970/1070 stuff with good thermals... evidenced by how many 970s sold.
Then the whole "expanding the TAM" stuff. Yeah, Nvidia already did that for them.
AMD's fanboys kill AMD. You can see it in the sentiment shown by AdoredTV and other AMD fanboys everywhere.
"Nvidia is overpriced", "NV is marketing" etc. NV NEVER is good enough for them, AMD can never lose because it's always "next year!". Yet some RX480 level product is just perfect no matter what.
Then the whole "expanding the TAM" stuff. Yeah, Nvidia already did that for them.
Yeah, that's not your average AMD buyer. AMD is dying because it's own fans kill it. They wait until the 'new 7850' level of value card is released and ignore Nvidia until AMD creates another card, like the 480. Otherwise they're not buying and starve their favorite company of cash.Does that make me irregular AMD fan as I bought the High-End card from them and it was not even faster than what the competitor was offering
I had my reasons why I bought this card. The reason can be seen from my display setup on my rig.
AMD's fanboys kill AMD. You can see it in the sentiment shown by AdoredTV and other AMD fanboys everywhere.
"Nvidia is overpriced", "NV is marketing" etc. NV NEVER is good enough for them, AMD can never lose because it's always "next year!". Yet some RX480 level product is just perfect no matter what.
Last edited:
5150Joker
Supreme [H]ardness
- Joined
- Aug 1, 2005
- Messages
- 4,568
This is pretty much true, most of the most vocal one's I've seen on forums and reddit tend to never buy new cards from either company, they just howl about it loudly. There's this really active guy on Anandtech named RussianSensation or something and all he does is constantly defend AMD and his last card was a 7970.
N
NemesisX
Guest
Ieldra
I Promise to RTFM
- Joined
- Mar 28, 2016
- Messages
- 3,539
Does that make me irregular AMD fan as I bought the High-End card from them and it was not even faster than what the competitor was offering
I had my reasons why I bought this card. The reason can be seen from my display setup on my rig.
There's no reason to justify your purchase man, I guess you're an aberrant AMD fan - which isn't a bad thing !
Today may be remembered as the end of the AMD rebellion against PCIe, the great battle of our time... may be over !
>arse technica
seems legit xD
N
NemesisX
Guest
They're a reputable site. I was waiting to see actual benchmarks for the 1080 compared to the 480 in CrossFire and the 1080 beats it in almost every test ... or it's a little above equal. It's still impressive for the 480, but AMD's claims didn't hold true.