When Nvidia or AMD launch a new GPU, there’s a typical rollout pattern. The first cards out the door are reference designs, based on a package Nvidia and AMD provide. Later, companies like MSI, Gigabyte, EVGA, Asus, and PowerColor debut their own custom designs. These custom boards are generally clocked higher, have better, quieter coolers, or may be built on a custom smaller PCB. This time around, however, AMD’s board partners are reportedly having major trouble sourcing GPUs.
That’s the word from Tom’s Hardware, which spoke to a number of companies. XFX and Sapphire have cards in the works, but no firm release date. PowerColor has a design but has yet to receive the DRAM it needs. Asus has pushed back the launch date on its own cards, Gigabyte hasn’t firmly committed, and MSI is apparently skipping the party altogether.
THG writes that the problem is twofold. First, the quality of AMD’s Vega GPUs has been variable enough that OEMs don’t feel comfortable establishing an overclock frequency they can guarantee and ship. Second, there have been temperature reporting discrepancies between what the board reports and what AIBs are measuring.
Molded vs. unmolded GPU packaging.
Finally, there are mounting differences that’s making it difficult to mass produce Vega. Unlike previous GPUs, not every Vega card uses the same mounting. There are three variants of Vega in-market: One with HBM that’s the same height as the GPU (molded, Samsung HBM2), one with HBM that’s 40 micrometers lower than the GPU (unmolded, Samsung HBM2) and apparently a third variant with Hynix memory.
Was HBM a Mistake?
I’m not going to pretend to have a secret source inside AMD on this one, but it’s hard to look at Vega and not wonder if HBM was a fundamentally bad call. The company’s molding issues are directly related to the height of the HBM2 memory. Meanwhile, it took AMD over two years to launch Vega. That’s the longest gap AMD or ATI has ever gone between high-end GPU refreshes, which previously took 12 – 16 months at most. HBM2’s rollout has been slower than anticipated in general, particularly the highest speed memory.
We don’t know for certain that HBM is the culprit here, but it certainly seems like the most obvious place to look. Because there’s always a considerable lag between when a GPU design project kicks off and when the final product tapes out, it’s entirely possible that AMD was too far along in the design process to start over when HBM2’s growing pains became clear.
The big signal here will be whether AMD’s future GPUs are based on GDDR6. If HBM2 was the culprit for the problems Vega seems to have, AMD won’t keep using it. AIBs are still expected to bring custom Vega cards to market in general, but those samples may not be in-market until closer to the holiday season.