Ryzen 7000 and Radeon 7000 rumored to share late 2022 launch window

midian182

Posts: 9,769   +121
Staff member
Rumor mill: Could AMD launch its next-generation Zen 4 processors and RDNA 3 graphics cards simultaneously or very close together? That’s the new rumor floating around, though we’ll be waiting until the end of next year to find out. At least the chip shortage should have improved by then—maybe.

The rumor comes from Twitter user Broly_X1, who does have a history for accurate leaks; VideoCardz notes that they leaked the whole AMD Computex keynote in advance, tweeting about the Radeon Pro launch before the card was revealed.

We’ve already heard that the Zen 4 architecture, which will be used in the Ryzen 7000 (Raphael) mainstream desktop CPUs, is set to arrive in 2022. It’s expected to be based on the 5nm process and feature an AM5 socket that switches from AMD’s usual pin grid array (PGA) design to the same land grid array (LGA) design favored by Intel, though the chip itself is expected to retain the same 40mm x 40mm size as those used in the AM4 socket. There’s also rumored support for dual-channel DDR5 memory and PCIe 5.0.

The RDNA 3-powered Radeon 7000 series, meanwhile, is also set to use the 5nm process node. There are rumors it will utilize a Ryzen-style chiplet design that could bring substantial performance improvements over the Radeon RX 6000 series. Broly_X1 adds that AMD will “tape out” RDNA 3 later this year, which is the point that the design will be complete.

Whether AMD does launch its Zen 4 processors and RDNA 3 graphics cards simultaneously or close together could all depend on the global chip shortage. One would imagine the issues we’re experiencing right now will have eased by the time Q4 2022 arrives, but one manufacturing giant has warned they could continue until 2023.

Last month saw part of an alleged AMD roadmap that included 3nm-based Zen 5 processors, slated to arrive in 2023.

Permalink to story.

 
Good to see AMD still moving forward and pushing Intel/nVidia! Competition is good!
 
I can already tell you I know they wont: not sure why would anyone think AMD can release both new cpus and new gpus given the current state of availability.

Not that they will stop to consider even for a millisecond that maybe they shouldn't repeat their 2020 paper launches for RDNA 2.0 but well, "war never changes" and such.
 
I can already tell you I know they wont: not sure why would anyone think AMD can release both new cpus and new gpus given the current state of availability.

But the problem is demand. AMD is making tons of CPUs/GPUs. They just all get snapped up. They are likely selling more than ever before, and so moving on to the next gen makes complete sense. Not being able to personally get a hold of one is a secondary point.
 
But the problem is demand. AMD is making tons of CPUs/GPUs. They just all get snapped up. They are likely selling more than ever before, and so moving on to the next gen makes complete sense. Not being able to personally get a hold of one is a secondary point.
Not really just "demand"

AMD is juggling CPUs for Consoles, Servers and mobile and desktop markets at the same time as they are using access to the same fabs for GPUs

It is true that "demand" is strong and supply is constricted (Artificially and intentionally by fabs but that's an entirely separate discussion) but AMD is not large enough to outbid and reserve a ton of fab space like Apple yet it similarly has their hands on too many markets.

AMD could allocate more of their reserved space to RDNA but honestly why would they? Consoles are a strong commitment and the most profitable for them it's still CPUs so if there's an area that necessariy needs to suffer, it should be RDNA where they're the least profitable in terms of ROI and such.

So the fact that AMD insist on telling the public they can not only handle all these lines but launch products concurrently or very close to, it's on it's face just not possible. They just want to piggyback the hype they generate from their CPU products towards their RDNA products without considering they'll be creating really significant distrust when it inevitable becomes their second paper launch in a row for GPUs.
 
Not really just "demand"

AMD is juggling CPUs for Consoles, Servers and mobile and desktop markets at the same time as they are using access to the same fabs for GPUs

It is true that "demand" is strong and supply is constricted (Artificially and intentionally by fabs but that's an entirely separate discussion) but AMD is not large enough to outbid and reserve a ton of fab space like Apple yet it similarly has their hands on too many markets.

AMD could allocate more of their reserved space to RDNA but honestly why would they? Consoles are a strong commitment and the most profitable for them it's still CPUs so if there's an area that necessariy needs to suffer, it should be RDNA where they're the least profitable in terms of ROI and such.

So the fact that AMD insist on telling the public they can not only handle all these lines but launch products concurrently or very close to, it's on it's face just not possible. They just want to piggyback the hype they generate from their CPU products towards their RDNA products without considering they'll be creating really significant distrust when it inevitable becomes their second paper launch in a row for GPUs.

Maybe. But perhaps the biggest mistake AMD could make right now is sitting on current generation hardware rather than moving ahead and keeping the technological position they've achieved relative to Intel and others.
 
Maybe. But perhaps the biggest mistake AMD could make right now is sitting on current generation hardware rather than moving ahead and keeping the technological position they've achieved relative to Intel and others.
Relative to intel = Agreed.

Relative to Nvidia = That's a bit more of a complicated question to answer. To me, just brute forcing it isn't going to do it they really should prioritize software support for RDNA 2.0 far more than a new gen.

I can concede partially that there might be some updates on the hardware that might be needed to support greater software support iterations so there's that part that might be not much of a choice at all but to move forward.
 
IIRC, it was based on the 3000 series

My post meant that they cleared that mobile-desktop naming mess with the 5000 series, so they wont do the same mess when they just cleaned it.

The question that matters is what are they going to release next as the 6000, ZEN 3+ or ZEN 4.
 
I just look forward to casually walking into a store and buying the stuff, when everything is in stock and priced around (or maybe slightly less) than MSRP. The way it was when I built my last computer. Ahh, the good old days.

Ok, time to go, got to get some damn kids off my lawn! :)
 
Relative to intel = Agreed.

Relative to Nvidia = That's a bit more of a complicated question to answer. To me, just brute forcing it isn't going to do it they really should prioritize software support for RDNA 2.0 far more than a new gen.

I can concede partially that there might be some updates on the hardware that might be needed to support greater software support iterations so there's that part that might be not much of a choice at all but to move forward.
Also... when I refer to "others", I also mean Apple. Sitting on tech isn't going to work well as Apple continues to develop it's Mx processors.
 
If AMD can make GPU chiplets a working reality this will make Graphics cards more affordable and more profitable for AMD.
 
Not really just "demand"

AMD is juggling CPUs for Consoles, Servers and mobile and desktop markets at the same time as they are using access to the same fabs for GPUs

It is true that "demand" is strong and supply is constricted (Artificially and intentionally by fabs but that's an entirely separate discussion) but AMD is not large enough to outbid and reserve a ton of fab space like Apple yet it similarly has their hands on too many markets.

AMD could allocate more of their reserved space to RDNA but honestly why would they? Consoles are a strong commitment and the most profitable for them it's still CPUs so if there's an area that necessariy needs to suffer, it should be RDNA where they're the least profitable in terms of ROI and such.

So the fact that AMD insist on telling the public they can not only handle all these lines but launch products concurrently or very close to, it's on it's face just not possible. They just want to piggyback the hype they generate from their CPU products towards their RDNA products without considering they'll be creating really significant distrust when it inevitable becomes their second paper launch in a row for GPUs.

Alternatively, when their high-end products move to 5nm node, that'll free up capacity on the N7 so they can keep manufacturing RDNA2 and Zen3(+) to better fulfill the demand for people who don't require the best hardware, and just need a reasonably priced GPU. After all, they already did the same with Zen(+) and Zen2, with 1600AF continuing to be manufactured long after Zen2 launched.
 
Alternatively, when their high-end products move to 5nm node, that'll free up capacity on the N7 so they can keep manufacturing RDNA2 and Zen3(+) to better fulfill the demand for people who don't require the best hardware, and just need a reasonably priced GPU. After all, they already did the same with Zen(+) and Zen2, with 1600AF continuing to be manufactured long after Zen2 launched.
That's a strategy that would work for us as consumer, but I have a sneaky yet strong suspicion AMD would like to concentrate on a new product vs re-launching and old one. They might still do it don't get me wrong, but knowing full well that a lot of people won't even consider buying a previous generation while the current one is out: they've kept the Ryzen 3000 series and even the 2000 series on retail without affecting sales for 5000 at all as most people would rather wait and even on heavy budget constrains most people try to fit in at least a 5600x
 
That's a strategy that would work for us as consumer, but I have a sneaky yet strong suspicion AMD would like to concentrate on a new product vs re-launching and old one. They might still do it don't get me wrong, but knowing full well that a lot of people won't even consider buying a previous generation while the current one is out: they've kept the Ryzen 3000 series and even the 2000 series on retail without affecting sales for 5000 at all as most people would rather wait and even on heavy budget constrains most people try to fit in at least a 5600x
Amazon still has 3700x, 3600 and even 1600AF in its top ten best selling list. Most people seem to be fine with getting a last gen CPU, if it means better price/performance. Especially when you can't even get a 5000-series CPU for less than $300.
 
Amazon still has 3700x, 3600 and even 1600AF in its top ten best selling list. Most people seem to be fine with getting a last gen CPU, if it means better price/performance. Especially when you can't even get a 5000-series CPU for less than $300.
That's entirely dependent on how you (Or Amazon) are selecting the timeline and whenever or not there's compelling options.

So if AMD could have had far more volume of 5000 chips at launch they would have probably impacted those other chip sales more. For a long while it was simply flat out not possible to order a 5000 chip so many people got what was available. They didn't decided "I'll skip the newest and clearly superior product" there was no choice at all.

So what you said still fits with what I am talking about here: AMD would like to focus on new products and abandon the older ones or not relaunch/support them anymore. Whenever they actually can do that or not due to supply constrains is a different story however.

So if we had no supply issues at all, AMD would just kill the older chips. Maybe they would do some rebranding to relanuch certain 3000 series chips as 5000 series chips as an economical option like rebranding the 4350g to 5200g and the 4650g to 5400g for example with the chip being exactly the same but now as a wider release even if it has the previous architecture but still shares the "5000" moniker. Believe it or not that can do the trick to get people to buy older tech.
 
Back