| craighopkins said: AMD ryzen brought good value so let's hope Vega does the same |
It's using HBM2. If you are hoping for value, it ain't happening. Polaris was their value proposition.
| shikamaru317 said: If the 3 SKU rumor is true, I'm expecting the lowest to be 5-10% above the 1070, the middle to be 5-10% above the 1080, and the highest to be somewhere in the neighborhood of the 1080ti. Not sure what pricing will be like though, hopefully AMD will aim for lower prices than the current crop of Nvidia cards, somewhere around $350 for the lowest, $450 for the middle, and $650 for the highest. |
The 3 SKU rumor might be true anyway. If we get two Vega GPU's this year which is what leaked details and benchmarks are hinting at... Then we might get that rumoured new Vega chip in 2018.
haxxiy said:
How so? Maybe not entirely... but there is still a fair amount of unreasonability going on there. The Fury X clocked at 1050 MHz on a 28 nm architecture. The jump to 14/16 nm alone increased clocks by some 30% on both AMD and Nvidia GPUs. And unlike Nvidia, for some reason the bigger AMD GPUs are the ones with the highest clocks. No reason for it to be below even the RX 460, a first generation GPU held back by the desire to make it work on a PCI-e without extra power connections.
|
1.2Ghz wouldn't be below a Radeon RX 460.
But to answer your question... The wider the chip, the more energy it tends to consume. If the chip is extremely wide and power hungry, then that can come at the expense of clockspeeds, it's a balancing act.
Sometimes a smaller chip is having better power characteristics than expected, thus it will ship out with higher clockrates to hit a TDP target.
Thus a low-end GPU having a higher clockrate than a high-end GPU is not entirely uncommon... As you allude towards, the RX 460 was a high clocking part at 1090Mhz base, which is higher than the Radeon RX 470 at 926Mhz.
www.youtube.com/@Pemalite
