Fury X has joined the game
#11
From ze latest beta
atom Wrote:Type: Driver
File: Host
Desc: added support for all AMD Catalyst >= v14.9
#12
(06-18-2015, 12:19 PM)Flomac Wrote: Btw. the latest AMD drivers are these Omega 14.12, which are not working under hashcat.
Did this change with the latest beta drivers? Otherwise it might be a bit difficult to test the Fury Wink

Edit: AMD claimes these 15.15 drivers to be needed for the new GPUs
http://support.amd.com/en-us/kb-articles...eries.aspx

I like how they support 64 bit only Smile
#13
(06-18-2015, 06:40 PM)atom Wrote: I like how they support 64 bit only Smile

If that leads to freed up manpower put into solving more bugs than yeah, it' very welcome Wink

So the newest beta (15.5) would work under hashcat? Maybe I organize myself a Fury X just for fun and give it a try.

Btw. AMD says the Fury Nano will have a TDP of 175 Watts, though it's yet unknown how many cores it serves, but a 2x better perf/TDP over the 290X is claimed resulting in 3584 cores. But then, again, no one knows in which way powertune will kick in.
#14
They can always downclock and undervolt the hell of the GPU, and that would work.
But, whatever's the case, we'll know for sure in 6 days (and they start selling em 3 weeks after that).
#15
Okie dokie, from what I've seen on Techpowerup, the stock variant exists with water hybrid water cooling, non serviceable.
The card was too hot for air cooling (surprise surprise), so no way it's gonna be <=275W.
The fact that AMD said their cooling system can dissipate up to 500W of heat is scary.

Also, REVIEWS TOMORROW!
#16
I think card is too hot because of form factor of design (to make things smaller), not because it's impossible to cool it by air. Make cooling better, get more quiet system.
#17
So it's finally there. A lot as expected. In games on the same level as the 980Ti. No problem to hold speed in games, a steady 1050MHz without throttling. Power consumption is ok, noise level with 40db very low, temp under 55° and no problem. So far that's a green light in games.

The situation under hashcat will be surely different. Under very heavy stress the clock goes down to 950-970MHz, which is still very good. The power consumption though goes up to an average 350W and has its peaks in 450W. Temp goes up to 65°. The card seems to be able to handle that and it doesn't throttle too much but one tester claimed this temp as way too high for a water cooled system.

One way to improve performance is changing the fan of the water cooler. It's under control of the Fury and is being hold back not spinning at its limit. With changing it you lose this control which is a good thing since the fan runs then at the speed you want. One website tried different fans and if you go extreme, the temp goes down from 65° to 52° while noise goes up from 40 to 50db. Same level as the reference 980Ti.

So it does not look bad at all. Changing the fan will be a must for hashcat users to get the temps under control but someone should definitly give it a try and see what it can pull off. I guess it might be a +40% from Hawaii and that wouldn't be too shabby, though still slower than the Titan X.
#18
450W under hairy doughnut is quite scary for a single GPU card, but, well, competition made AMD desperate.
So far, the card looks better than the abomination I expected.
Heck, if it won't throttle in any way under oclhc, I would say the communist engineers have done a good job this time, as long as they don't push the max TDP even higher and use freon cooling next time.

3 weeks before some practical tests with oclhc, which is what most people care about.

Oh, and it's GCN v1.2, so no LOP3.LUT or anything fancy besides HBM1 this time.
#19
At the same price it will have roughly the same speed like the GTX980 Ti, which draws 100W less. I will now surely upgrade from 980 -> Ti.
#20
(06-24-2015, 05:17 PM)Rolf Wrote: 450W under hairy doughnut is quite scary for a single GPU card, but, well, competition made AMD desperate.
So far, the card looks better than the abomination I expected.
Heck, if it won't throttle in any way under oclhc, I would say the communist engineers have done a good job this time, as long as they don't push the max TDP even higher and use freon cooling next time.

3 weeks before some practical tests with oclhc, which is what most people care about.

Oh, and it's GCN v1.2, so no LOP3.LUT or anything fancy besides HBM1 this time.


Are you sure this card is GCN v1.2 ??
I see here that they write that is v1.3
http://www.hardocp.com/image.html?image=...82X2wuZ2lm

I think best will be wait to get one card to atom and after we will know all.