IT'S SO FUCKING CUTE
>>55364199
It's no nano~~
>>55364199
Thats not a nano dude.
>>55365570
It's much more affordable though.
Also, pic related is pretty cute too
>>55364199
The bare die tho
>>55365592
I sincerelly dont care about that since i can afford all this stuff.
I enjoy my nano.
I hope amd makes a nano successor
Also, what gpu is that?
is that a piece of white lint on the gpu die ?
>>55365633
Gtx1070 itx iirc
>>55365661
Ye
>>55365661
Sure is.
>>55365688
>Gtx1070
stop shilling
>>55365707
KILL YOURSELF YOU FUCKING DEGENERATE SCUM
>>55365737
>>55365633
The image has rx480 written on it.
>>55365592
8-pin rx480?
If so then I would also look forward to buy this card for a mini pc build.
>>55365707
I know nvidia shilling is on its peak since ya know, the release and it is also summer, but hey, we talking about mini itx-cutey-squishy-lovable gpu's.
>>55365806
But it's a gtx1070 ;_;
>>55364199
HOT HEAD OH NO
>>55365592
Janky ass OEM cooler.
>>55364199
I like to imagine there's wildfire in the VRs just waiting for sustained overcurrent to unleash itself
>>55364199
are those 8 things the vram?
why is vram so much smaller than sticks of dram
>>55364199
Yup.
using it right now.
>>55364199
XSPC is making a cheap water block for it, and EK is making one also.
>>55366826
Your ram has similar chips on it, it just works a little different
>>55365629
L-lewd!
Has anyone tested the rx480 with better cooling yet?
Global foundries 14nm apparently runs a little too warm but I haven't seen how it responds when kept cool.
>>55368424
https://www.youtube.com/watch?v=2kjIHgq2zBU
Hits 1390 Mhz with low temp/noise, limited by 6 pin power.
>>55365904
The Underrsterd Store
>>55368424
>Global foundries 14nm apparently runs a little too warm
lol no
>>55368424
That's not the problem, it's the amperage draw that's messing up the card, whether that's AMD, foundry, or something else we'll have to wait and see.
>>55369096
The reference 480 uses a jank ass piece of shit flimsy heatsink like you find on old, old cpus, and a weak little fan that's supposed to cool the whole card. If you put that same cooler on a 280 i bet it would get 95+c.
>>55369175
Who thought this was a good idea?
>>55365629
it's so future-core
>>55369096
>how does electricity work
Seriously, look it up.
The voltages for the pstates are higher than necessary to ensure tons of chips passed validation. There is a little section of registers on die that record what is essentially die quality, and its a rating of how well the die performs per volt. 480 cards shipped to reviewers and ones in consumer hands have a pretty wide sweep. AMD didn't do any clocking binning with these dies. They basically said
>90% of dies run with these voltages, so ship em all with these settings
The RX 480 is a factory overclocked card. GCN doesn't favor high clocks because of how short the architecture's pipeline is. The Polaris 10 die was designed to yield high, and be as cheap as possible, so among other things it was limited to 32 ROPs. They clocked it as high as they did to make up for this lowered pixel throughput which is why power consumption is what we see. Were it clocked around 800mhz-1000mhz it would be around 100w~ but then it wouldn't have the pixel throughput necessary to compete against the 970. If they gave it more ROPs they'd have a large more complex die that wouldn't be as cheap.
The balancing act of trying to make a profitable $200 card isn't as simple as everyone seems to think.
This is all separate from the issue of the cards placing a heavier draw on the PCI-E slot than on the 5 pin connector. On that specific point AMD said it was an issue affecting some cards, and they were looking into it.
>>55369227
Its pretty incredible compared to all prior GCN based dies.
The SIMD lanes, l1 caches, scalar unit, TMUs, ROPs, everything is separate and distinct. The structures themselves were fabbed so they're less dense, less chance for defect. Balanced vs die area you can see they put a ton of work into making sure they'd have solid margins on the chips.
Unfortunately no one on this board can appreciate technology.
>>55369323
Besides you, you mean?
>>55369337
Of course.
>>55369323
That picture gives me feelings I've never felt before in my nether regions.
>>55369323
I think it's neat, even when AMD's products aren't always the best for gaming they're still very interesting. The R9 290X was a very impressive piece of hardware and both the PS4 and Xbox One APUs are actually pretty neat pieces of hardware.
>>55369435
Amd gpu's are more like for compute.
In fact the firepro line is kinda useless in the low end, since the gayming counterpart is on par.
>>55370471
Isn't the 290x better at compute than the Titan x?
>>55369323
I appreciate how sexy it looks, anon-senpai
>>55370546
Yes.
In fact the 295x2 is king.
I havent seen compute related benchmarks for the Radeon Pro Duo
>>55369096
>That's not the problem, it's the amperage draw that's messing up the card, whether that's AMD, foundry, or something else we'll have to wait and see.
The VRM design is strange, to me it looks like the entire GPU is powered by the PCIe port. Notice the rx480 mosfets and chokes are close to the pcie 12V pins. Around the power connector there's only one VRM it probably only powers the fan and other small things.
They must want to reuse the PCB on lower end models that run off PCIe only.
>>55370800
Maybe they intend to use the same pcb for the 470 and possibly 460. I wouldn't be surprised if they were just cut down 480s to decrease costs.
>>55370546
>>55370881
I have never been able to find much use for my R290X compute power. Lux render is the only thing that works, blender cycles is slower on opencl than CUDA, none of the commercial programs have any openCL rendering support. 4GB isn't enough either 16gb and 32gb is around where GPU ratracing will become convenient.
>>55371601
I boinc on the side when not gaming so I get some usage out of it.
