What's wrong with 40+ fps on 4K?
>>63612909
Colton Wright
Other urls found in this thread:
Samuel Clark
Is this tested with Crimson ReLive Redux?
Brandon Sullivan
they used a new version Crimson 17.11.4 since in the old drivers there was (is) a bug that sets some parameters as in the reference design.
James Bailey
>OC
>3 more fps than reference Vega
literally Fury 2.0
Time to wait™ for Navi
Jonathan Jones
A custom overclocked vega 64 can't barely beat the regular non oc vega.
Just no
This is not the point
Lincoln Johnson
Just wait for the magic drivers kind sirs
bad nvidya is finish
pls buy amd radeon premiun grafiks
Adrian Powell
Is this the Vega with the missing paste on the HBM?
youtube.com
Jose Gray
*Insufficient paste to cover the HBM on a pre production sample
but yes, thats the card
David Ortiz
>4K
so its a CPU bottleneck
David Bell
No shit. It's not Raja's fault though. It's not like he designed GCN or made the call to continue using it. They had no choice because they don't have the resources to develop a new architecture. AMD are finished in the GPU space.
Lucas Sanchez
Vega seems to do alright in situations where you don't need to get the highest possible ammount of performance out of your silicon, see Raven Ridge.
GCN ist just shit at the high end in general: to get NVidia competing performance you need very high clocks, for that you need a lot of voltage and for that you get a very high power consumption and basically kill your OC headroom.
We saw it with the 200, 300, Fury and 400 series, now we're seeing it again with Vega.
AMD needs to ditch GCN already and work on something new.
On paper, in terms of TFLOPS, Vega64 should be 10-15% Faster than the 1080Ti, yet it just about competes with the 1080, which should be 30% slower.
Imagine what would happen if AMD gets their shit together and get a similar level of performance/FLOPS to NVidia
Aaron Williams
>i7-8700K
Xavier Stewart
How about the noise and temperature?
Ian Ortiz
Well Raja fucked off to Intel now. You happy? Wait™ for Navi.
Jeremiah Brown
>4,7 GHz
>not 5,2
you might as well underclock it
Angel Mitchell
On paper, GP102 hits ~13 TFLOPS on boost 3.0 alone.
Besides, how do you "ditch" GCN?
No one does this.
No, not even "We are fucking rotten corpse of G80": the company.
Justin Reyes
G80 was amazing
Justin Wright
The point is the stock Vega vs the overclocked aib Vega
Mason Green
So is GCN.
In fact, Vega pushes the boundary far and beyond, adding TB(D)R and some neato hacks aka NGG.
Some don't work yet, but it's a question of time and software, but hardware design.
Jayden Evans
A shame the chinks are fucking RTG up.
John Bell
How?
Vega10 is an excellent piece of hardware.
It's the problem of software, and software alone, given the even nVidia struggled to make TBDR work.
Besides, the chinks started with Cypress.
What's wrong with them?