In the CPU room AMD is currently shooting on all cylinders and has recovered in the performance sector with its usual Ryzen processors and high-end desktop (HEDT) Threadripper chips. Both are now in their second generation. So what's next? AMD goes into a 7 nanometer production process, both for a new generation of CPUs and GPUs.
Of course we've known this for a long time, but what's interesting about this change is that Intel is still trying to get its 10nm chips out of the door, in volume. Cannon Lake has seen many delays, but according to Intel, everything is on its way. We have to wait and see if that's really the case.
Meanwhile, AMD is throwing forward and knocking on its 7nm chips that pay big dividends by keeping the moment in motion. AMD is not just hope for this, it needs 7nm to work out as expected.
"We knew 7nm would be a big challenge, so we put the effort, we moved our resources to the new node," AMD CTO Mark Papermaster told CRN in an interview. "We did not just die in the water. We went all in."
The "all-in" approach seems to work, at least in the case of AMD's roadmap. According to Papermaster, the Sunnyvale chip designer had originally planned to release 7nm Vega GPUs in 2019. However, they are now scheduled to arrive once a year with the release of AMD's upcoming Radeon Vega Instinct GPU. AMD's "huge focus" at 7nm is what allowed the company to shorten the timeline, said Papermaster.
In addition, AMD is ready for 7nm Epyc server chips for this year, followed by a 7nm Ryzen update once more. AMD sees a window with the ability to roll out products on advanced production processes, while rival Intel continues to adjust its 10nm process knot.
"I think we have the opportunity to be positioned much stronger than we originally expected," added Papermaster. . "But I have to say that our original plan was to be placed very strongly, so any delay from our competitor could only strengthen the value AMD brings to the market."
It takes a lot of planning and coordination to get where AMD is currently on. By going all-in by 7nm, the company had to make sure that design teams, foundries and manufacturers were on the same plan. Juggling the latter is one of the more difficult tasks with this kind of thing because it means widespread change to even the automation tools they use. Unlike Intel, AMD is a fabless company.
"In order to really utilize what's needed to succeed in a semiconductor industry, it needs 100 percent focus," said Papermaster.
Although it may sound like a gamble for AMD putting all the eggs in a 7nm basket, it may be more out of necessity than choice. Intel content like its 10nm process is more advanced than the competition, as it packs more transistors and does a better job of scaling. In fact, Intel has said a desire to introduce a new calculation to reflect exactly where chip design technologies really are.
"What really needed is an absolute measure of transistors in a given area (per mm2). In the other extreme, just take the total number of transistors of a piece and sharing of the site but not due to the large number of design decisions that can affect IT factors such as cache sizes and performance targets can lead to large variations in this value. It's time to revive a metric that was used earlier, but dropped out by favoring several nodes since. It is based on the density of the transistor of standard logic cells and includes weight factors as the account for typical design, "said Intel last year.
Regardless of how you shake (or measure) it, 2019 will be a very interesting year in CPU design and technology in general.