AMD Releasing “Zen 3” CPUs on Nov. 5: Flagship Ryzen 9 5950X (16C/32T) to Cost $799

Tsing

The FPS Review
Staff member
Joined
May 6, 2019
Messages
11,214
Points
83
amd-zen-3-logo.jpg
Image: AMD



It might be time for that CPU upgrade. During today’s Zen 3 event, AMD CEO Dr. Lisa Su unveiled her company’s latest 7 nm desktop processors: the Ryzen 9 5950X, Ryzen 9 5900X, Ryzen 7 5800X, and Ryzen 5 5600X. These CPUs boast higher max boost clocks and an impressive IPC uplift of 19 percent.



The performance of the Zen 3 architecture is so great, in fact, that AMD is billing the Ryzen 9 5950X and Ryzen 9 5900X as the fastest gaming processors on today’s market. Benchmarks presented by red team show that its Ryzen flagship edges out Intel’s Core i9-10900K in titles such as Shadow of the Tomb Raider and Total War: Three...
Click to expand...

Continue reading...
 
Nice! Can't wait to see those reviews. Gaming performance was the only thing they were still lagging on when compared to Intel. If I didn't just build a new PC 2 years ago I'd be looking at one of these.
 
Decided to pull the trigger on a 3700x when it hit $275. Figured I'd regret it later, but I have yet to see an equivalent 8 core 65w TDP part and the MSRP on the 5800x is a bit high. I'll see what they look like in 6 months maybe.
 
raz-0 said:
Decided to pull the trigger on a 3700x when it hit $275. Figured I'd regret it later, but I have yet to see an equivalent 8 core 65w TDP part and the MSRP on the 5800x is a bit high. I'll see what they look like in 6 months maybe.
Click to expand...
The 3700x is still a great processor and, yep, for that price you just can't beat it. I know I love mine.
 
Armenius said:
Nice! Can't wait to see those reviews. Gaming performance was the only thing they were still lagging on when compared to Intel. If I didn't just build a new PC 2 years ago I'd be looking at one of these.
Click to expand...

The fact they talked specifically about single-thread performance and that games still rely on a heavy single-thread, impressed me. If single-thread performance is as up as they claim, then this could be really good for games.
 
Brent_Justice said:
The fact they talked specifically about single-thread performance and that games still rely on a heavy single-thread, impressed me. If single-thread performance is as up as they claim, then this could be really good for games.
Click to expand...

Indeed. Obviously, I don't have a full breakdown of the architecture, but a couple ways this might have been accomplished is by increasing the boost clocks, but this would imply that the boost clocks may be more consistent than the last generation. Also, a single core will have access to more L3 cache without crossing CCX's or CCD's. While AMD wasn't super specific, they did mention FPU improvements, and game engines like that, so this is another potential reason. It's likely not any one thing, but a combination of factors which lead to the increases.

It's too early to tell, but AMD has generally been very straight forward and honest about Ryzen, (Boost clocks not withstanding), and I don't doubt the numbers we are seeing here. Even if those numbers are averaged. That said, "trust, but verify." And we do intend to verify. :)
 
Riccochet said:
If it's that much of an improvement I'll be selling my 3700X and buying a 5800X.
Click to expand...

19% increase in single-threaded applications is a nice jump. Let's hope it really holds true.
 
I look forward to release, reviews, and hopefully some price drops. $500 too rich for my blood for a 2700x replacement. $300 maybe.
 
Glad to hear the news, now the real question. Will they actually be in stock anytime soon after being released?
 
Those numbers are very high if they hold up. Well they really completely reworked this thing.
I guesd the zen 3plus will be with ddr5 and zen 4 will be 5nm and ddr5 and yet another re work? AMD gone crazy, but the good kind.
 
I've been recommending AMD CPUs for gaming outside of more unique situations (or demands) since they got the 3000-series more or less provably stable. Performance on Intel could be higher, but the difference wasn't worth the cost for most people.

And I'll probably wind up recommending these. While Intel hasn't gone full Bulldozer yet, they have really shot themselves in the foot with their fab misadventures!
 
Uvilla said:
Those numbers are very high if they hold up. Well they really completely reworked this thing.
I guesd the zen 3plus will be with ddr5 and zen 4 will be 5nm and ddr5 and yet another re work? AMD gone crazy, but the good kind.
Click to expand...

According to AMD, Zen3 is a completely new architecture. It was partially developed along side Zen2 and obviously incorporates lessons learned from previous Zen architectures into it, but it is supposedly a new design. AMD's roadmap (which has been out there for months) suggests there won't be Zen3+. The roadmap indicates that this time next year we will see Zen 4, not Zen3+. That will be the one that will have DDR5 and potentially support for PCIe 5.0. Now, that doesn't necessarily mean that it won't be iterative to some degree but we are a long way off of finding that out.
 
Dan_D said:
According to AMD, Zen3 is a completely new architecture.
Click to expand...
I have to wonder how much this is really the case. Zen (through Zen+ and Zen 2) always seemed like an outstanding architecture wrapped in layers of compromise. Process limitations, bus speeds, memory speeds, memory compatibility, chipset woes, UEFI woes, board partners doing their best ECS impressions... but the cores themselves were solid.

I was mostly just hoping that Zen 3 would be the version where they finally take the brakes off it!
 
LazyGamer said:
I have to wonder how much this is really the case. Zen (through Zen+ and Zen 2) always seemed like an outstanding architecture wrapped in layers of compromise. Process limitations, bus speeds, memory speeds, memory compatibility, chipset woes, UEFI woes, board partners doing their best ECS impressions... but the cores themselves were solid.

I was mostly just hoping that Zen 3 would be the version where they finally take the brakes off it!
Click to expand...

Until I have a full architectural breakdown from AMD, I won't be able to say. I'm not sure what you mean by process limitations. TSMC's 7nm process is fairly efficient. In fact, AMD was able to gain a higher transistor budget as it were and leverage that for more performance in the same power envelope that we saw with Zen+. If you are referring to clock speeds, clocks aren't determined by process node alone. Look at Intel, they still have a fairly large clock speed advantage over AMD for all the good that's done them. Intel is still at 14nm. Intel also hit 3.8GHz back in the Pentium 4 days on much larger process nodes than we have today.

It's not process limitations that limit AMD's clock speeds. It's the overall design of the CPU that does that. You can specifically design CPU's to achieve higher clock rates by doing things like increasing your pipeline stages. When you do that, you get Netburst and Bulldozer. That's why neither company does that anymore.

Memory speeds went up, and memory compatibility improved. There really aren't any chipset woes with X570. UEFI / AGESA code woes, certainly. Those account for the vast majority of issues relating to X570 or AM4 motherboards in general.

AMD is capable of roughly the same memory speeds as Intel systems are. The only difference is that Intel's memory controller does better with four DIMMs than AMD's controllers do. Intel's memory controller also benefits less from clock speed and tighter memory timings than AMD does. AMD's board partners get away with more than Intel's partners can because AMD doesn't have the same level of control that Intel exerts over them.
 
I had no reservations about this CPU. It came out focused on the area’s it was the weakest and that’s great. Benches are hopefully coming soon and let’s just say I’m ready for a 5950X long as the benchmarks confirm the presentation.
Edit: on second thought even a 5900X
 
Last edited:
Dan_D said:
I'm not sure what you mean by process limitations. TSMC's 7nm process is fairly efficient.
Click to expand...
7nm is; I was speaking to their challenges with 12nm before that in the context of the challenges they'd had at getting clockspeeds up on Zen.
 
You must log in or register to reply here.
Become a Patron!
Back
Top