AMD Polaris Thread: Radeon RX 480, RX 470 & RX 460 launching June 29th

[Bacon1]

there have been plenty of legit looking leaks with GPU-z data & all you really need to do is look at the cooler - there's no way it won't throttle the card

38ish hours to go - I'm fairly confident throttling will be one of the major drawbacks mentioned in the reviews

Ok so no actual source of throttling while gaming then, when multiple leakers have had no throttling issues. Just what you think because you've seen a picture of the card and a screenshot from furmark. Gotcha.

 

[PeckingOrder]

Ok so no actual source of throttling while gaming then, when multiple leakers have had no throttling issues. Just what you think because you've seen a picture of the card and a screenshot from furmark. Gotcha.

there have been screenshots from games, too (XFX, imgur album, GTA V and other games)

I believe in thermodynamics, too and I'm pretty sure the ref cooler doesn't cut it

but you can keep drinking the AMD Kool-Aid until the reviews are out

 

[Silverforce11]

Because this is about trying to get as many folks as possible to BELIEVE their lies BEFORE 480 is released, so that as many as possible will BUY NVIDIA cards BEFORE 480 launches.

ONCE 480 releases, the truth of 480's performance/stock/OC-ability/etc will be known, and this fud and its associated lies evident to all.

Couple this with (this rumor) :

https://semiaccurate.com/2016/06/24/serious-manufacturing-problem-hits-nvidia-consumer-gpus/

And this makes it imperative (from nVidia's perpective) to get folks to purchase as many nVidia cards as possible BEFORE the 480 Release, and BEFORE these rumored supply issues--IF TRUE--become visible and known to the GPU buying masses. Taken together, this would also explain why nVidia pulled forward (by ~1.5 to 2 MONTHS) the 1070/1080 "launch" and the (pure cash grab) of the Founders Edition pricing as well.

If 480/470/460 cards perform and OC well and stock is plentiful, and the rumored (nVidia) supply issues play out, AMD will gain significant market share, and thus nVidia's income stream and market share will suffer.

I don't need a Semiaccurate article to know NV rushed their paper launch way before they were ready.

What did we see in late April? Die shots of GP104, QUALIFYING SAMPLES on AIB cards. What else? GDDR5X from Micron... again, QA samples. Still in the qualifying stage in late April and they want to launch a month later?

How it works is qualifying, everything is good to go, mass production of chips begin. It takes time for mass produced GP104 and GDDR5X to be assembled into cards, tested, and then shipped to retail over a period of time to build up volume for launching.

Nobody would launch while qualifying is just wrapping up, that unless they want to rush their launch ahead of schedule. It's 5 weeks after the "launch", 1070 and 1080 stocks are nonexistant. The shills would love to believe demand is just so crazy that it's the only reason. They are in dream land if they think demand for $699 GPU is mass volume. I know for a fact many retailers here only received a handful for the launch, and every few days, another small handful arrives.

The dual MSRP & FE pricing is just that. By rushing the launch, they capitalize on a period of no competition and can charge inflated prices. A GTX 970 replacement suddenly goes from $330 -> $449. A GTX 980 replacement goes from $500 -> $699. Few people on team green seem to care.

Yet when AMD is launching a $199/$228 mainstream part, they feel so threatened they have to constant post fud, even so far as bringing up "AMD has bad drivers" dogma while constantly mis-representing the mainstream segment by comparing it to $499 & $699 GPUs from NV.

I'll say it again for those who pretend not to know:

RX 480 if it can match a GTX 980 & 390X, brings performance of $499 & $399 down to $199/$229. That's not good enough?

 

[Silverforce11]

there have been screenshots from games, too (XFX, imgur album, GTA V and other games)

I believe in thermodynamics, too and I'm pretty sure the ref cooler doesn't cut it

but you can keep drinking the AMD Kool-Aid until the reviews are out

If you're wrong when the reviews are out, will you be man enough to apologize for your blatant fud spreading?

 

[RussianSensation]

just saying the ref cooler on the RX 480 is worse than what Intel includes with their lowest end desktop CPUs (and they got about 1/4 TDP of this gpu) and it will make RX 480 throttle hard.

Pretty amazing stock CPU cooler for a $315 processor.

Can you please tell me why NV is selling a $699 FE videocard with a GPU cooler that's miles worse than a $27 CPU cooler?

No one is forcing anyone to buy a reference RX 480 when Sapphire Nitro will produce a great RX 480 card.

Why not let the miners buy all the reference RX 480 and wait until after-market RX 480 cards show up?

If noise and temperatures were a primary concern, then it's best to skip all AMD/NV reference cards (with the only exception AIO CLC reference cards like R9 295X2/Fury X, etc.)

 

[PeckingOrder]

If you're wrong when the reviews are out, will you be man enough to apologize for your blatant fud spreading?

ofc I will

No one is forcing anyone to buy a reference RX 480 when Sapphire Nitro will produce a great RX 480 card.

That's exactly what I'll do, wait for a decent non-ref card to come out. I just hope I won't be waiting too long.

// btw I think the RX 480 will be a good card, not a great card tho (not 8800 GT great). but imo the AMD marketing dept pushed it as a revolutionary jump in performance in its price segment - and that won't happen. having unrealistic expectations and then seeing the real performance might leave a lot of folks disappointed

 

[Silverforce11]

http://videocardz.com/61475/amd-radeon-rx-480-rumors-part-7

Again more leaks point to 390X/980 class of performance on stock.

Looks like the reference cooler is good for stock clocks, but not good for OC. Hmm.

 

[RussianSensation]

Furmark much?

"Furmark is an application designed to stress the GPU by maximizing power draw well beyond any real world application or game. In some cases, this could lead to slowdown of the graphics card due to hitting over-temperature or over-current protection mechanisms. These protection mechanisms are designed to ensure the safe operation of the graphics card. Using Furmark or other applications to disable these protection mechanisms can result in permanent damage to the graphics card and void the manufacturer's warranty."
https://forums.geforce.com/default/...her-stress-tests-with-geforce-graphics-cards/

November 11, 2010
"NVIDIA has officially released the successor of the GTX 480: the GeForce GTX 580. This card is powered by the GF110 GPU, which is a refresh of the GF100 GPU. For more detail about new things brought by the GF110, check this page out.

But the real new thing is somewhere else: the power draw is now under strict control. Like AMD with the Radeon HD 5000 series (see )ATI Cypress (Radeon HD 5870) Cards Have Hardware Protection Against Power Virus Like FurMark and OCCT), NVIDIA has added dedicated hardware to limit the power draw. In short, when FurMark is detected, the GTX 580 is throttled back by the power consumption monitoring chips."
http://www.geeks3d.com/20101109/nvidia-geforce-gtx-580-the-anti-furmark-dx11-card/

August 27, 2008
"Interesting piece over on Geeks3D - it seems that the latest Catalyst 8.8 drivers released a few days ago by AMD have a built-in feature that detects when the user has the FurMark OpenGL benchmark running and downclocks the GPU on the Radeon HD 4850 and Radeon HD 4870 to keep the temperature down."
http://www.zdnet.com/article/amdati-catalyst-8-8-downclocks-gpu-on-detecting-furmark/

Both Nvidia and AMD are strictly against using synthetic power viruses for testing videocards.

Furmark is designed to load every single component on the GPU and the PCB, including the Mosfets and VRMs. Can any real world application load all the shaders, geometry, pixel shaders, textures, ROPs, cache, memory, VRMs/power delivery to 100% simultaneously? The way Furmark works is that if you design a GPU with 2x8-pin power, even if the ASIC itself (the chip) is designed for only 180-190W, but the board is designed for 375W, Furmark will go all the way to the board's limits as set in the BIOS. That's exactly how a power virus behaves.

The same 980 videocard hits 204W in games.

Using Furmark for GPU stress testing is downright dangerous and irresponsible. Using Furmark only increases the changes of RMA for a PC gamer due to the risk involved. What exactly are we trying to prove by showing that a Gigabyte GTX 980 G1 SKU can handle > 340W of power? Unless we are using LN2 for overclocking to see if we are going to be board/VRM power limited if we were to use 1.8-2.0V GPU voltage, this is a meaningless power test for 99.999% of PC gamers.

 

[JDG1980]

Furmark is designed to load every single component on the GPU and the PCB, including the Mosfets and VRMs. [...] Using Furmark only increases the changes of RMA for a PC gamer due to the risk involved.

If the VRMs and MOSFETs are crap, I'd rather find out right away while the card can still be returned, rather than years down the road after sustained usage causes it to degrade.

 

[Yakk]

Yeah, not many people do hardware burn ins anymore...

 

[RussianSensation]

If the VRMs and MOSFETs are crap, I'd rather find out right away while the card can still be returned, rather than years down the road after sustained usage causes it to degrade.

So if someone made an RX 480 with 2x8-pin and 14+1 power phases, and you fired up Furmark and the card used 350W of power over 24 hours and burned out overnight, you'd blame the manufacturer?

Yeah, not many people do hardware burn ins anymore...

There is nothing wrong with stress testing a videocard with real world applications. Ethereum mining = makes $. Distributed computing = hobby. Rendering/video encoding = work/producitivity/hobby. Games = hobby, etc. Furmark = useless synthetic power virus.

 

[rgallant]

ofc I will



That's exactly what I'll do, wait for a decent non-ref card to come out. I just hope I won't be waiting too long.

// btw I think the RX 480 will be a good card, not a great card tho (not 8800 GT great). but imo the AMD marketing dept pushed it as a revolutionary jump in performance in its price segment - and that won't happen. having unrealistic expectations and then seeing the real performance might leave a lot of folks disappointed

not to peeps who IN other parts of the world who would have to spend alot of money [1,2,3,4 of weeks wages] to own a 390x or gtx 980 class of gpu.
now that will be cut in half. I'm almost sure they will be crying with joy and not disappointment

 

[Shivansps]

not to peeps who IN other parts of the world who would have to spend alot of money [1,2,3,4 of weeks wages] to own a 390x or gtx 980 class of gpu.
now that will be cut in half. I'm almost sure they will be crying with joy and not disappointment

Try 3 months..

 

[zinfamous]

What source do you have for any of this?

he keeps posting Xfire data and neither claiming it is single card or xfire, just saying "this is bad; I do not want it" despite being perfectly within the norm of expectations for what it is.

He and the other have done this about 5 times now. Over and over, called on it, but keep doing it.

an obvious shill.

 

[sm625]

One thing I always wondered about furmark... how exactly do they know it is running? Does the driver search for an exe with furmark in the name? If you renamed your furmark executable file, would it still detect it?

 

[Elixer]

One thing I always wondered about furmark... how exactly do they know it is running? Does the driver search for an exe with furmark in the name? If you renamed your furmark executable file, would it still detect it?

That is one way to know it is running, by checking the name, the other way is to have a hash that is unique to furmark, and there are other ways available as well.

 

[JDG1980]

So if someone made an RX 480 with 2x8-pin and 14+1 power phases, and you fired up Furmark and the card used 350W of power over 24 hours and burned out overnight, you'd blame the manufacturer?

Why would anyone do that? Even the most powerful Pitcairn cards, like the MSI HAWK and PowerColor Devil, only had 2x6-pin connectors. That means the board is only allowed to draw 225W (75W slot + 2x 75W PCIe), and those boards are overbuilt enough that they have no trouble handling that. (According to TPU, the 7870 HAWK actually draws 244W under FurMark, which is illegal per the PCIe spec, but it should still be fine if you have a solid PSU.)

Yes, I do run Prime95 + FurMark for 24 straight hours whenever I update the CPU, GPU, or other related components. I will probably be running it again later this or next week when I pick up a RX 480. And I suspect this card will be fine, as the previous ones have all been.

 

[JDG1980]

One thing I always wondered about furmark... how exactly do they know it is running?

They don't. The power limiter on AMD and Nvidia cards doesn't care what application you use. It just monitors power draw and throttles back if it hits the limit.

 

[crisium]

I wonder if AMD will regret not having AIB custom coolers for launch day, again (they learned so well with custom only 300 series).

This not only sets an initial impression on July 29th, but a continuous impression as:

1) Many sites will continue to compare this card in upcoming reviews (against 1060, Vega, etc)
2) Many people will google 480 reviews and get the launch day reviews.

Even if, by chance, all the 480s overclock the same including reference, you are still leaving a relatively worse noise and heat impression. Nvidia can get away with it, but AMD cannot.

And right now the common belief is OC results will be markedly better on aftermarket, so that's just another metric that will make the card look less desirable than if we had Nitro there day 1.

I'm surprised they would do this again. Unlike 290 series this looks to be just fine for stock speeds, but AMD needs every edge they can get and a blower just doesn't leave a great impression most of the time.

 

[3DVagabond]

If the VRMs and MOSFETs are crap, I'd rather find out right away while the card can still be returned, rather than years down the road after sustained usage causes it to degrade.

Note what the 1080 does when running Furmark. It doesn't even run at it's base clock. It's a useless comparative bench when the IHV sets reduced clocks for it.

 

[nkdesistyle]

I wonder if AMD will regret not having AIB custom coolers for launch day, again (they learned so well with custom only 300 series).

This not only sets an initial impression on July 29th, but a continuous impression as:

1) Many sites will continue to compare this card in upcoming reviews (against 1060, Vega, etc)
2) Many people will google 480 reviews and get the launch day reviews.

Even if, by chance, all the 480s overclock the same including reference, you are still leaving a relatively worse noise and heat impression. Nvidia can get away with it, but AMD cannot.

And right now the common belief is OC results will be markedly better on aftermarket, so that's just another metric that will make the card look less desirable than if we had Nitro there day 1.

I'm surprised they would do this again. Unlike 290 series this looks to be just fine for stock speeds, but AMD needs every edge they can get and a blower just doesn't leave a great impression most of the time.

Not a mistake. They had to have a reference model because they have lot of oem's getting these cards as well and they had to price it cheap. The first impression they have to set it bang for buck. If that 229 stock it delivers 980 stock performance they have done their job. 229 sounds way better than not having a base price on launch day. Give it breathing room and you can have AIBs launch ASAP in like 2 weeks. Almost every rumor has been mid july for AIB cards that is pretty quick.

Remember the first impression is $$$ and performance for it. Second impression is aesthetics. If they have AIB cards and no reference they fail at the first impression. AIB cards are going to be all above msrp.

 

[3DVagabond]

Not a mistake. They had to have a reference model because they have lot of oem's getting these cards as well and they had to price it cheap. The first impression they have to set it bang for buck. If that 229 stock it delivers 980 stock performance they have done their job. 229 sounds way better than not having a base price on launch day. Give it breathing room and you can have AIBs launch ASAP in like 2 weeks. Almost every rumor has been mid july for AIB cards that is pretty quick.

Also by not giving the AIB's access to the silicon earlier it reduces the leaks. I'm sure that AIB's who make both AMD and nVidia cards would have given nVidia a heads up months ago about price and performance if they knew. The last thing they would want is to over pay for their nVidia product.

 

[Yakk]

So if someone made an RX 480 with 2x8-pin and 14+1 power phases, and you fired up Furmark and the card used 350W of power over 24 hours and burned out overnight, you'd blame the manufacturer?



There is nothing wrong with stress testing a videocard with real world applications. Ethereum mining = makes $. Distributed computing = hobby. Rendering/video encoding = work/producitivity/hobby. Games = hobby, etc. Furmark = useless synthetic power virus.

I burn in to test all components. I much rather it fail right away under extreme circumstances then take longer under average conditions. After that, the cards can mine or play for years of uptime.

 

[Silverforce11]

Note what the 1080 does when running Furmark. It doesn't even run at it's base clock. It's a useless comparative bench when the IHV sets reduced clocks for it.

Well, it's up to AMD to detect Furmark and throttle it hard to prevent it from power loading to the extreme.

If their GPUs look bad with Furmark, they only have themselves to blame.

 

[3DVagabond]

Well, it's up to AMD to detect Furmark and throttle it hard to prevent it from power loading to the extreme.

If their GPUs look bad with Furmark, they only have themselves to blame.

My point was that they all throttle during Furmark. It's not because of the quality of the heatsink. It's done in software.