AMD’s CPU and GPU roadmap for 2020 and beyond is full of big upgrades

One of many greatest tales in PC {hardware} over the previous few years has been AMD’s surge in mindshare and recognition. No simple job—for years, AMD’s CPU and graphics efforts have been principally met with disappointment. However with the introduction of its Zen CPU structure in 2017 and, to maybe a lesser extent (for now), its Navi GPU structure, there may be real pleasure and optimism for its merchandise, which hadn’t existed for the reason that Athlon 64 days.

Wanting forward, AMD appears poised to maintain the momentum going, each with its upcoming Zen Three CPU structure and RDNA 2, which is able to underpin its next-gen “Navi 2X” GPUs (in any other case often called “Huge Navi” in media circles). Nevertheless, whereas the street results in Zen Three and RDNA 2, it would not cease there.

Throughout its Financial Analyst Day presentation this week, AMD shared some updates about its roadmap and what to anticipate by way of options and efficiency.

Zen Three And Zen Four Are Each On Monitor

(Picture credit score: AMD)

COPYRIGHT_BP: Published on https://bingepost.com/amds-cpu-and-gpu-roadmap-for-2020-and-beyond-is-full-of-big-upgrades/54938/ by - on 2020-03-08T07:13:52.000Z

Beginning on the CPU facet, AMD reiterated that its future variations of Zen are on monitor. Zen Three will come out later this 12 months, whereas Zen four is slated to reach in 2021. That stated, there are fascinating (and delicate) particulars to notice within the slide above.

In Search Of New {hardware}?

The preliminary model of Zen was launched within the first quarter of 2017 and comprised AMD’s first technology Ryzen 1000 collection processors, and was constructed on a 14-nanometer manufacturing course of. Then got here Zen+ in April 2018, which shifted to a 12nm node and made up the second technology Ryzen 2000 collection. We’re now on Zen 2 (7nm), the structure behind AMD’s third-generation Ryzen 3000 collection. Keep in mind, the smaller the manufacturing course of, extra energy environment friendly the transistors, and the extra you’ll be able to match onto a chip.

Be aware that I am solely referring to desktop CPUs right here—issues get needlessly complicated if throwing cellular APUs into the combo as a result of they do not comply with the identical mannequin scheme. What’s fascinating, although, is AMD’s shift away from the “+” designation. AMD now clumps its Zen+ line into the preliminary Zen household, though they signify two generations of merchandise.

Considerably associated, discover that Zen Three is not labeled as 7nm+ because it had been in beforehand shared roadmap slides. What offers? This has to do with how its manufacturing associate TSMC labels its course of nodes.

As highlighted by our associates at Anandtech, there are three:

• N7—preliminary model utilizing ‘DUV’ solely instruments (no EUV)
• N7P—second-gen model of N7, additionally DUV solely
• N7+—EUV model of N7

DUV (deep ultraviolet) and EUV (excessive ultraviolet) seek advice from the kind of lithography used, and particularly the wavelength vary. Samsung really has a very good explainer on lithography strategies, however briefly, EUV paves the way in which for extra superior processors.

Because it pertains to AMD’s roadmap, Zen Three is seemingly not being constructed with TSMC’s N7+ node. So to keep away from confusion, AMD dropped “7nm+” from its present slides and easily says 7nm (it isn’t clear if Zen Three will probably be based mostly on N7 or N7P).

Regardless, AMD is on schedule. Through the presentation, AMD acknowledged its aim is to at all times have a aggressive product within the wild, one other one in improvement, and yet one more one within the design section. Zen four falls into the latter class and can see a shift to 5nm.

(Picture credit score: AMD)

One of many issues I discovered fascinating about AMD’s presentation was its candid remarks on course of expertise. The corporate’s CTO, Mark Papermaster, stated AMD did not anticipate being a frontrunner in course of expertise when it shifted to 7nm. He is referring to being forward of Intel, which is struggling to get correct yields from 10nm (we’re nonetheless ready on 10nm Intel desktop CPUs).

Evaluating nodes from two completely different producers (TSMC and Intel) is just not so simple as trying on the quantity. Nevertheless, it is plain that AMD is forward of Intel proper now, because it pertains to manufacturing.

See within the slide above how the traces converge in 2022? In no unsure phrases, Papermaster stated AMD assumes Intel will “come again sooner or later” and be forward once more, simply not instantly.

“After you have a niche and a few of the points which will trigger a delay in a brand new expertise, it is gonna take a while,” Papermaster stated.

In the meantime, AMD is assured it may preserve its present trajectory, which is sweet information for customers. If that occurs, even when Intel jumps again forward (with 7nm or 5nm), AMD is actually saying it will not be just like the previous, when its merchandise have been merely much less succesful and fewer fascinating than Intel’s.

Subsequent-generation Radeon Playing Cards To Deliver A 50 P.c Performance-per-watt Enchancment

(Picture credit score: AMD)

Transferring on, AMD believes it’s on stable footing with its RDNA graphics structure as nicely. RDNA is what powers AMD’s present technology Navi GPUs within the Radeon RX 5000 collection. In response to AMD, it introduced a 50 p.c uplift in efficiency per watt in comparison with its earlier technology GCN structure.

RDNA 2 is now proper across the nook. It would energy the next-gen model of Navi. We already know this will probably be featured in Sony’s PlayStation 5 and Microsoft’s Xbox Sequence X consoles, and Microsoft not too long ago bragged that the GPU in its system will ship 12 TFLOPS of graphics efficiency.

That is extra TFLOPs than what Nvidia’s GeForce RTX 2080 Tremendous affords, although we won’t discern a lot about efficiency from that alone—the Radeon RX Vega 64 is able to 12.66 TFLOPs, however is slower than a 2080 Tremendous.

For the PC, AMD has already stated it plans to launch RDNA 2 merchandise this 12 months. In response to yesterday’s presentation, RDNA 2 will once more ship a 50 p.c efficiency per watt enchancment. AMD can also be already looking forward to RDNA 3, however is not providing up too many particulars for the time being (it is going to arrive by 2022).

AMD is just not adopting the Huge Navi nomenclature that’s generally utilized in headlines, and is as a substitute rolling with “Navi 2X” for its RDNA 2 GPUs. And the corporate’s message right here is evident—it has each intention of competing within the excessive high-end by releasing “top-of-stack GPUs with uncompromising 4K efficiency.”

(Picture credit score: AMD)

AMD is establishing massive expectations for Navi 2X (and after that, Navi 3X). Along with promising “excessive efficiency with environment friendly energy,” Navi 2X GPUs will deliver hardware-level help for real-time ray tracing, and variable fee shading.

If this all comes collectively, AMD will deliver parity between its GPUs and Nvidia’s lineup, each in efficiency and options. Whether or not or not that truly occurs stays to be seen, and naturally Nvidia is not standing pat—its Ampere lineup is rumored to deliver massive efficiency positive aspects as nicely.

What does this all imply for PC gaming? Properly, if AMD can execute on each fronts (CPUs and GPUs), avid gamers could have a wealth of choices to select from, not simply within the speedy future but additionally past. The larger query is graphics. We have already seen AMD assert itself again into the CPU house, particularly with its latest-generation Zen 2 processors. The identical has solely been partly true in graphics, although. Nvidia owns the top-end of the market. We’ll have to attend a bit of bit longer to see if that modifications.