where does AMD go now?

File: amd2.jpg (52KB, 568x612px) Image search: [Google]

52KB, 568x612px

>>55694513
Vega will fix everything and beat 1080ti nvidiot.

>>55694513
just buy what you want

>>55694513
nVidia doesn't have a 14nm card

File: 1467970861102.jpg (104KB, 960x540px) Image search: [Google]

104KB, 960x540px

>>55694513

Delete this NOW

This is your final warning...

Sales for the RX 480 seem to be doing well. Higher tiers are much smaller so I doubt them having a competitor for the 1070/1080 is as damaging as you're making it out to be.

Big question is if Zen succeeds.

>RX 480 is being throttled by the power delivery / temperatures, non-reference cards will offer much better performance
>aftermarket card with 3% higher boost block ends up performing 3% better than the reference card
>RX 480 can't OC because it's being limited by the amount of power it can get, non-reference cards with an 8 pin will OC much better
>Maximum overclock of our sample is 2250 MHz on the memory (13% overclock), which is limited by the adjustment limit in AMD's drivers. GPU overclocking works slightly better than with the AMD reference design, reaching 1355 MHz (3% overclock).
How many of you got your hopes up? All this spiel was basically a repeat of 290X's launch, so if you were around then, you shouldn't have fallen for it this time.

>>55694513
>RX480
>flagship
inb4 Fury
>Fury
>this year

>>55694848
290x didn't have any power limit issues, it was all thermal throttling. Had like 20% of OC headroom with custom cooling.

>>55694513
>>55694848
>3.4% core clock bump
>3% performance boost
Nearly perfect scaling with overclocks.

Reminder the STRIX is the LOWEST clocking of the aftermarket AIBs. Most are clocked at 1350-1400Mhz, and that's a 10% performance boost without any memory overclocking, which scales even better.

If they're priced under 300$ the AIB 480s will absolutely devour the 1060, even moreso than the base card has.

>>55694513
The hell is this? I've seen reference PCBs overclock better.

ASUS has really turned to shit on the AMD side

File: perfrel_1920.gif (74KB, 400x1091px) Image search: [Google]

74KB, 400x1091px

>>55694893
I didn't claim it did. I did mention temperature throttling as a factor with RX 480 as well, and I said that this stuff was "basically a repeat". Basically, not exactly.

Anyway, the stuff people were saying with 290X were mostly that stock vs stock comparisons with reference cards are unfair, because 290X is throttling. Was there a huge difference? Pic related. A slightly overclocked (+50 MHz) aftermarket card couldn't reach performance more than a few % higher, basically a result of it's overclocking, not fixing the throttling. The same thing happens with the Sapphire and MSI cards. You can go check them yourself if you want.

The performance increases the aftermarket cards got from overclocking were 6.5%, 11.1% and 10.2%. As a comparison, the reference card got a 7.2% performance increase on uber mode. The differences were not particularly huge.

>>55694955
Looks like you still have your hopes up.

>Nearly perfect scaling with overclocks.
Anon, anon, anon. That is not how it works. You completely ignored the memory overclock. You can't OC both the core and memory and then claim that all the performance increase came from OC'ing the core. While memory OC doesn't lead to as large performance gains most of the time, they're still there, and they're especially there when your memory OC is 4 times your core OC.

Example of memory OC benefits in comparison to core OC benefits
http://www.corsair.com/en-us/blog/2014/october/overclocking_the_gtx_980

>Most are clocked at 1350-1400Mhz
Powercolor's card is 1330 MHz.
XFX's card is 1338 MHz.
Sapphire Nitro+ is 1342 MHz.

Stop fucking lying. Seriously. You people wouldn't be hated so much if you stopped fucking lying.

>>55695066
You're comparing factory OC like that's the true limits of the card. My Sapphire Tri-X could hit 1130Mhz easily. That's quite a jump form the 1030Mhz stock. Slapping on water or custom AIR you could hit 1200Mhz.

Also of course you list the Asus which has several threads showing how shitty it was.

>>55695066
>Anon, anon, anon. That is not how it works. You completely ignored the memory overclock.
I didn't though, the baseline for the STRIX vs Reference is 3.4% overclock for 3% performance boost.
1310/1266 = 1.034
Baseline STRIX has no memory overclock.

File: perfrel_1920.gif (71KB, 400x1057px) Image search: [Google]

71KB, 400x1057px

>>55695116
>You're comparing factory OC like that's the true limits of the card.
Holy fucking shit. Are you illiterate?

>The performance increases the aftermarket cards got from overclocking were 6.5%, 11.1% and 10.2%. As a comparison, the reference card got a 7.2% performance increase on uber mode. The differences were not particularly huge.

>Also of course you list the Asus which has several threads showing how shitty it was.
Seriously, how fucking illiterate are you?

>The same thing happens with the Sapphire and MSI cards. You can go check them yourself if you want.

>>55695130
oh fug, I got things mixed up there since I had the overclocking results in mind.

That said, 1266 MHz isn't exactly the value you should be using, by the way. Reference RX 480 doesn't quite stay 1266 MHz. For example, in hardocp's BF4 benchmark, it averaged 1260 MHz, so we're looking at a ~4% OC for 3% performance boost, so ~75% scaling.

http://www.hardocp.com/article/2016/06/29/amd_radeon_rx_480_video_card_review/3

>>55695179
Well maybe I'm tired but I'm failing to see your point. You're saying not to get your hopes up because aftermarket OCs weren't particularly high right? I'm saying that's nonsense because they're far more capable with the right cooling. Not sure if that will be the case with the 480 but I'm still waiting on reviews from people that know how to OC and not just moving a couple bars in Wattman.

File: cooler2.jpg (263KB, 900x600px) Image search: [Google]

263KB, 900x600px

>>55694513
I'm really loving the heatpipe contact on this cooler.

I give it 2.5/5.

>>55695197
I'm saying

1. Aftermarket cards did not give any large performance increases at stock like many people were hoping. Reference RX 480's performance wasn't significantly limited by its power or by its temperatures.
2. It's unlikely that you'll get huge performance increases with overclocking. TPU's overclocking results have been in line with other sites' before. No real reason to think that things are different now.

File: perf_oc.gif (32KB, 400x377px) Image search: [Google]

32KB, 400x377px

>>55695289
Ok well the graph you posted was just the factory OC and not an actual OC like I was talking bout. TPU got the Tri-X to 1135Mhz, pretty much what I said.

https://www.techpowerup.com/reviews/Sapphire/R9_290X_Tri-X_OC/27.html

Bigger difference than what you're claiming.

Also they could only get the Asus to 1090Mhz
https://www.techpowerup.com/reviews/ASUS/R9_290X_Direct_Cu_II_OC/27.html

>>55694513
What's with this non-stop nvidida shilling on /g/?

It was never even close to this bad with the last release of AMD cards.

>>55695355
AMDtards are the ones shilling, there were 10x more RX 480 threads than GTX 1060 threads on here. Now that the verdict is out that the 1060 handily beats the 480, with market price being very similar for both cards, you have to be an absolutely retarded fanboy to go AMD this generation.

>>55695355
half of /g/ is shillbots. Don't even bother arguing with them.

File: 1466282708985.jpg (43KB, 486x317px) Image search: [Google]

43KB, 486x317px

>>55694513
>the overclocker dream

File: 98797987987.png (8KB, 363x364px) Image search: [Google]

8KB, 363x364px

>>55694513
>Flagship

>>55695404
I feel half of them are butthurt freetards that just want to see GPU threads burn.

>>55695778

it is AMD's flagship for this generation. vega is not even being released until next year.

>>55695340
>TPU got the Tri-X to 1135Mhz, pretty much what I said.

Can
You
Fucking
Read

>The performance increases the aftermarket cards got from overclocking were 6.5%, 11.1% and 10.2%. As a comparison, the reference card got a 7.2% performance increase on uber mode. The differences were not particularly huge.

>>55695391
>1060 handily beats the 480
The only people that say that are toms and shilling sites like them and must i remind you that almost all the games that toms benchmarked were gameworks games exept for 3

But then again what to you expect from the people that shilled Pentium 4 and the Nvidia FX series at periods in time where AMD/ATI were hands down better.

>>55696903
>AMD/ATi beats Intel/Nvidia for a period totalling 3 years from 2003-2006
>therefore you should buy AMD even if they're much worse than the competition nownow
Madshill logic

>>55695391
I really dont give a fuck about gpus, i dont really enjoy games as much i did before and my laptop is good enough for me.on 1080p

But the constant amd retardation on g is so fucking annoying

http://radeon.com/radeon-wins-3dmark-dx12

>>55694513
To the grave. Prepare for glorious monopoly. Game developers won't have to optimize their games for both nVidia and AMD.

>>55695250
>I don't understand thermodynamics

>nVidia buys RTG

File: 1310483412100.jpg (36KB, 413x395px) Image search: [Google]

36KB, 413x395px

>>55694513
Post yfw literally everything kyle said in his blogpost was true

>>55697564
Only it doesn't, even by their own graph.

>the 480 is a flagship
It's becoming increasingly easier to spot shills.

>>55695391
>T-they started it

>>55696996
Good job pretending you didn't notice what he said before about how the tests were almost all Gameworks titles.

>>55698424
Are you simply not going to play any Gameworks game because you have an AMD card. That already puts you in a massive disadvantage.

File: 1466869089545.jpg (1MB, 3165x4080px) Image search: [Google]

1MB, 3165x4080px

IT'S OVER FOR THE CURRY-BASED-KEKOLDS OF AMD CORPORATION AND ITS SHILLS

MY

FUCKING

GIGASIDES

GNU/NEETS ON SUICIDE WATCH

What about Free/g-sync?
FreeSync monitors are actually affordable.

>>55698463
Name one GameWorks game that is actually worth playing.

>>55694513
In the loo. I mean designated shitting streets.

>>55694513
>what can AMD do to recover?
Do people buy rx480?
Will people buy rx480 in the future?

if answer is yes to both of this questions, then what are they going to recover from?

>>55698547
>moving the goalposts
GPUs are made to play games, people will play those games whether you like them or not.

>>55698632
The margins AMD gets from RX 480 may not be particularly high.

>>55698352
480 is a flagship, flagship just means best option.
right now it's not only best it's also the only one.

>>55698704
maybe, who knows, if only there was actual stock to know what is selling.

>>55698114
Does it really not matter though? I don't know that much about thermodynamics but it seems like they're not as useful as they might be with a copper base.

Aren't most direct touch cpu coolers worse than others anyway?

>>55696996
Its about toms not amd you stupid nvidiot, lrn 2 reading comprehension.

File: whatrecoverignlookslike.png (115KB, 1098x602px) Image search: [Google]

115KB, 1098x602px

>>55694513
>what can AMD do to recover?
Whatever they are doing right now is working pretty well.

for 1080p gaymen and streaming, is a 1060 good enough for top graphics or should I go for the 1070? Will need a new psu for that though.

>>55701110
RX 480 Nitro

>>55701167

Nope, needs to be NVidia since I use the gamestream on my shield tv, and the hacked drivers for amd cards are terrible.

>>55696393
1135Mhz is a 13.5% OC from reference clocks. You're going from factory OC to their OC to make it look like less than it is.

Still you're misguiding everything here because your main point was there wasn't a huge difference which is wrong.

>>55701217
>he bought a shield tv
>>>/v/

or

buy a steamlink and don't be a faggot

>>55701217
If you're stuck on shield a 60 could do if you don't mind lower settings.

>>55701217
Not even memeing, selling that useless shield and getting a steamlink + RX 480 would be way cheaper for more or less the same thing

>>55701289
>useless shield

i use it as a plex server. which can serve 1080p video to 4 remote devices, 4k netflix box and it's pretty good for casual vidya. There is nothing else available on the market at a similar price point that does anywhere near as much. So no, the shield ain't useless.

>>55701696
>gimic after useless gimics
yea it is useless just like all the other shit that garbage company makes

What are you talking about? Their shares are up and they just posted earnings hours ago with projection of 15% revenue increase next quarter due to 480 shipments.

>>55701838

wew lad, how can it be a gimmick if it's used. the plex server is used often, serving video to my phone, or even my tablet, whilst on a train, netflix in 4k is great on my 4k screen.

you really are a shitty currynigger working for amd, aren't you?

>>55702250
>>55702250
>you really are a shitty currynigger working for amd, aren't you?
thanks for confirming that you are an nvidia shill/fanboy

File: 1464366265571.png (2MB, 1000x967px) Image search: [Google]

2MB, 1000x967px

>>55694513
>Flagship

>>55699181
I wish I had had some money to invest earlier this year.