They only started coming out Fiji chips for desktop.. don't think they'll pursue laptop market with Fiji so fast. There's probably still one more Fiji based SKU which is dual chip Fiji (R9 Fury Rage Roar Wow?)..

A very good news to Amd R9 300 series, seem like they finally fix the crash problem.
https://community.amd.com/message/2671041#2671041

But very weird, I din heard anything from our forum saying the card did crash the game?

Performance /rm is insanely high

i think it should be LOW?

in any case, paying Fury X price for something that performs like a 390X.

if one is looking for performance then it's horrible. if your purchase is to factor in the size then it's different.

What iI meant is performance /rm since our rm is nose diving everyday lol

I think goldfries is right. Performance/RM ratio will be low since RM is increasing against USD. If RM(price)/performance ratio, then it will be high.

Yeah u got it right was meaning rm/performance silly mistake

Real performance gains should be next year.

I think even with R9 290, the performance is good enough to wait for the next generation GPUs

thats true, theres rumor that they dont send sample to Techreport and TPU because they both use a price/performance chart.
Since this card costs more than double a 970 it would look pretty bad in that scenario.

R9 390 8GB performs close, and can outrun GTX 980 on some tests.

Want performance? get 2x R9 390 8GB CFX!!!

It'll be super power hog of course but that's quite some performance for RM 2.8k or so.

It reach its power target before thermal limit thats why the temperature is low. You can do it even on Hawaii cards.

Anyone tried the new dota 2 engine?

My R9 290 can't maintain 120fps, GPU usage around 30%+, CPU slightly below 60%. All settings maxed 1080p

While in my cheap laptop with GT 610M, lowest settings at 768p and game render 100%. The fps on average 80+, while on source 1 engine it is only 30fps

It's harder to run on max settings while it is friendly to old pcs

120FPS min or average? Mine can't maintain over 120FPS when there's alot of effect on the screen.

Its the frame cap. I didn't unlock the cap because having 120fps coincide nicely with my 60hz monitor. It's like vsync without the vsync latency. Works well with source 1 dota2. But source 2 dota2 can't maintain 120fps. Maybe need to lower settings zzzz. Not like my card is low end

This post has been edited by terradrive: Sep 11 2015, 03:06 PM

For Source 2 will have support for DX12 right? Let's see what they bring to us then.

The problem is......it isn't overclocking reasonably well enough.
There was one overclocker who able to overclock (volt modding?) Fury (cut down of Fury X) to 1400mhz and the cooling and power requirements are insane.
Fiji chip efficiency is thrown out of whack once it reaches 1ghz.

I wonder how well will this better binned Fiji/Nano will fare if it is using water cooling. (The lower power phases might be an issue though)

Eh, I am starting to think the other reason that AMD reorganizes the Radeon division might be easier entity separation (sold off? spun off?) if thing suddenly get bad for AMD.

This post has been edited by JohnLai: Sep 11 2015, 05:19 PM

Oh, i increased the cap to 200 since i didn't notice any tearing.
Just tried latest client and i got average 130-ish FPS give or take with some rare split secs dip to mid 100s/high 90s. Last week client perform slightly better though with rare dips into mid 110s.
DX11 mode perform lower for me btw.

Yes and Vulkan.

What is this "reasonable overclock" for Fury-X according to you?

Any high end card that need volt mod will use insane power consumption. Heck, an unlocked GTX780 with all safety bells and whistles removed can consumed 400w ALONE when overclocked.

Yes the board power limit will be an issue to overclock a Nano over the top especially since board partner not allowed to create custom PCB for it.

That is one of the possibility in case AMD CPU division going under if Zen fail.

Something like Intel CPU or Nvidia Maxwell V2 level of overclocking.

Example, increasing GTX980 base clock from 1126MHz to 1377MHz (251Mhz differences, unless if you consider max boost clock) only resulted in extra 47-67 watts (based on anandtech gtx 980 review)


EDIT:

Oh, someone finally asked amd about the 'binned' controversy.

My question: I plan on water-cooling and overclocking the Nano in a tiny case. What kind of performance can I expect while watercooling+overclocking the Nano compared to the Fury X especially since the Nano is a higher binned card?

There has been some misconception out there that the Nano is an uber binned part. Actually, both Nano and Fury X share the same ASIC configuration. What differs is the audience and form factor the products are targeted for. If you are interested in water cooling, aftermarket waterblocks are available for both the Nano and Fury X however the Nano is designed to be smaller and more power efficient with lower overall clocks and power delivery circuit. The Fury X is the card you want if you intend to push the Fiji ASIC to the absolute limit.

- Victor

Source

This post has been edited by JohnLai: Sep 12 2015, 11:35 AM

Generally, every chip has a sweet spot (best performance per watt (power efficiency)).

For Fury X, I believe they already pushed the card to almost max (1050mhz) on stock, so the power efficiency is not ideal.. If they released FuryX at 900mhz stock, I believe the power consumption would match R9 Nano and NVIDIA's power efficiency, and the "OC" headroom would be 25% to 1120mhz. The R9 Nano actually gives a very good view on the power efficiency of the Fiji arch.

But normally reviews are tested based on stock clocks, so if they released Fury X at 900mhz, they would get disappointing reviews.. as it would be so far behind NV's 980Ti stock clocks..

Radeon's currently have so much untapped performance potential especially at DX11 games.. If they can somehow "fix" their DX11 driver, all their GCN based cards would be major performance boost.

How do you guys define power efficiency? To me how much power is wasted as heat; power draw vs. actual power consumption; is what dictate whether power efficiency is poor or good. Just because power consumption is high doesn't necessarily mean power efficiency is not good.