AMD Unveils 5 nm Ryzen 7000 "Zen 4" Desktop Processors & AM5 DDR5 Platform

[TheLostSwede]

I was thinking of going to ADL now with more of it's bugs ironed out with bios updates so far, also same with win 11.... however... I'm keen to know when AM5 will actually be launched for retail globally?
Watching the DDR5 market indicates prices are coming down in my part of the world, ever so slowly, but by the time AM5 is really actually available I presume the kits will be cheaper still.
Also, happy my AM4 cooler mounting kits will work on AM5!

August/September maybe?

 

[jesdals]

Looks good but will still wait for gen 2 AM5 motherboards because of DDR5 implementation - but hopefully we will se Intels counter showing the true performance with DDR5. But nice to see progress.

 

[Durvelle27]

I’m making the switch

 

[Verpal]

The scary thing here is if AMD decided to make the Radeon RX 7400 with PCIe 5 x2 connectivity. Then we are all screwed all along

I wanted to laugh at this statement but I also laughed at the idea of AMD will launch something with pcie x4 before so......

 

[Pumper]

Disappointing. Only 15% up when the CPU is boosting 400-500MHz higher is really not impressive.

And what is up with that comparison to 12900K? In 5950x vs Zen4 they are using 3600CL16 vs 6000CL30, but vs Intel they use the same 6000CL30 for Intel and 6400CL32 for Zen4, for whatever reason - why not use the same RAM, especially when you already have the Zen4 +6000CL30 RAM system at hand?

MT looks impressive, but then again they are using some in house benchmark for that instead of Cinebench, which they could have used instead, so it's impossible to tell if it will perform that good in the real world (5950X is only ~5% faster than 12900K in rendering, according to TPU tests).

No gaming numbers also shows that AMD is not confident in the performance advantage vs the current Intel CPU, not to mention their own 5800X3D. No wonder they chose not to release 5900X3D and 5950X3D as these would just end up beating Zen4.

Was waiting for Zen4 to see if I should upgrade from my 3900X to 5950X, now that these are going for 500€. Looks like there is no point in going for Zen4, especially when it requires a new mobo, new ram and new cooling on top of the CPU.

 

[zlobby]

We now need the mobile parts!

 

[ARF]

Disappointing. Only 15% up when the CPU is boosting 400-500MHz higher is really not impressive.

And what is up with that comparison to 12900K? In 5950x vs Zen4 they are using 3600CL16 vs 6000CL30, but vs Intel they use the same 6000CL30 for Intel and 6400CL32 for Zen4, for whatever reason - why not use the same RAM, especially when you already have the Zen4 +6000CL30 RAM system at hand?

MT looks impressive, but then again they are using some in house benchmark for that instead of Cinebench, which they could have used instead, so it's impossible to tell if it will perform that good in the real world (5950X is only ~5% faster than 12900K in rendering).

Was waiting for Zen4 to see if I should upgrade from my 3900X to 5950X, now that these are going for 500€. Looks like there is no point in going for Zen4, especially when it requires a new mobo, new ram and new cooling on top of the CPU.

Well, I guess the potential negative reviews will push AMD to decrease the prices of the platform.
Because otherwise, AMD will not sell well. The prices are too high as is.

 

[Denver]

Maybe but the wording in their slide saying ST performance and not IPC leads me to believe they are factoring in the clock speed increases already.

Also the MT performance increase was pretty underwhelming the 5950X is already about 20% faster in Blender vs the 12900k so 11% beyond that is pretty meh.

Hopefully I'm wrong or they are underselling it.

Nope.

 

[Daven]

The biggest missing spec is the L3 cache configuration, specifically 3D V-cache. I wonder if AMD is still deciding on what to do.

Edit: oh and also AMD is staying max power is 170W but most sites think this means base power. To me it seems like 170W is the power at max boost which would compare to the 240W of Alder Lake.

Edit: upon further looking at MSI AM5 mboard marketing material, it compares the 65-105W TDP of AM4 to 65-170W TDP of AM5. Nothing released today insinuates 170W base TDP so I’m guessing this is future proofing for higher core counts.

Edit: even more confusing, we have a 5.5 ghz clock in a gaming demo and a 15% ST gain in cinebench. Impossible to tell if clocks were the same in both cases.

Edit: I’m gonna go out on a limb here and say clocks were similar between the 5950X and the Zen 4 processor. The 15% is probably mostly IPC. Since IPC is usually the mean or median clock for clock increase in performance between two parts over like 30 applications, I don’t think AMD has this number yet. They are still tuning for more apps and final specs may not be set in stone yet.

 

[Denver]

I realized now that the difference is 45%, not 31%. 12900k is 45% slower.

Ps. The AMD marketing team's terrible.

 

[Valantar]

I realized now that the difference is 45%, not 31%. 12900k is 45% slower.

Ps. The AMD marketing team's terrible.

Those two numbers are literally the same thing. 204s is 31% faster than 297s; 297s is 45% slower than 204s. They chose the more conservative wording, which uses the existing product as the baseline for comparison. That's the only sensible, good-faith comparison to make - especially as a "slower than" wording in marketing is guaranteed to be flipped into a "faster than" wording by readers who don't consider how this changes the percentage. And that would be a shitshow for AMD.




While I haven't watched the presentation, I have to say this sounds a tad underwhelming. If that 15% is IPC, that's pretty good even when accounting for DDR5. If it's ST performance? That's underwhelming, especially when you take into account a 10% clock speed increase. Also a bit disappointed to not see any major packaging changes here - the CCDs are stacked closer, but other than that it looks like we're still getting through-package IF with its high power draw. And that's a damn shame. Here's hoping 7000 APUs will be the true focus of this generation, with MCM APUs finally entering the ring.

 

[Daven]

Those two numbers are literally the same thing. 204s is 31% faster than 297s; 297s is 45% slower than 204s. They chose the more conservative wording, which uses the existing product as the baseline for comparison. That's the only sensible, good-faith comparison to make - especially as a "slower than" wording in marketing is guaranteed to be flipped into a "faster than" wording by readers who don't consider how this changes the percentage. And that would be a shitshow for AMD.




While I haven't watched the presentation, I have to say this sounds a tad underwhelming. If that 15% is IPC, that's pretty good even when accounting for DDR5. If it's ST performance? That's underwhelming, especially when you take into account a 10% clock speed increase. Also a bit disappointed to not see any major packaging changes here - the CCDs are stacked closer, but other than that it looks like we're still getting through-package IF with its high power draw. And that's a damn shame. Here's hoping 7000 APUs will be the true focus of this generation, with MCM APUs finally entering the ring.

The 5.5 Ghz number was in a game demo. The 15%+ number was in a list with a 5 ghz+ number with footnotes that it was a cinebench test. It’s crazy confusing.

 

[Chrispy_]

So this is the official reveal, when is the official launch (and review embargo date?)

As cool as Zen4 is likely to be Zen2 and Alder Lake were both troubled launches, Zen3 was smoother because it was so similar to Zen2 and fully platform-compatible. I'd be inclined to let some other guinea-pigs pay through the nose to beta-test the new platform for me for at least a couple of months.

 

[529th]

I got the same impression as most sentiments here. This will be an interesting launch to hold back on, purchase wise, and watch what things really boil down to let alone AGESA changes. I have a feeling the 3D will be close to AM4 performance in gaming.

 

[thegnome]

Err they must be sandbagging or something for later... Zen 4 has a better node and higher clocks that would alone make up ground to Alderlake mostly... Zen 3 was 19% higher IPC compared to Zen 2, and that was with a few design changes and slightly higher clocks, thats without a new node, memory, cache, etc.

 

[Arkz]

So B550 will be PCI-E 4.0 on the 16 lane slot, but then 5.0 on the M.2 slots. Hmm, well that's probably still fine. I don't think any GPU comes close to saturating all 16 lanes in 4.0. Hell a 3080 on a 3.0 slot runs fine.

 

[ARF]

So this is the official reveal, when is the official launch (and review embargo date?)

Sometime September or October this year.

I wonder why they hurried so much to reveal it today... is it anything to do to get the partners ready for the software ecosystem update?

So B550 will be PCI-E 4.0 on the 16 lane slot, but then 5.0 on the M.2 slots. Hmm, well that's probably still fine. I don't think any GPU comes close to saturating all 16 lanes in 4.0. Hell a 3080 on a 3.0 slot runs fine.

You probably talk about the B650, and by the way, are there any PCIe 5-ready M.2 SSDs?
Even if there are, they would be price-wise prohibitive. Just get a good-old PCIe 3 NVMe M.2 SSD

 

[Arkz]

Sometime September or October this year.

I wonder why they hurried so much to reveal it today... is it anything to do to get the partners ready for the software ecosystem update?



You probably talk about the B650, and by the way, are there any PCIe 5-ready M.2 SSDs?
Even if there are, they would be price-wise prohibitive. Just get a good-old PCIe 3 NVMe M.2 SSD

Yeah B650 ha ha. And yeah probably not, although there will be eventually. But until all new games start supporting Direct Storage I'm not that fussed with NVMe drives anyway. Generally the performance in games isn't even that different between a decent NVMe and a SATA SSD.

 

[Valantar]

Err they must be sandbagging or something for later... Zen 4 has a better node and higher clocks that would alone make up ground to Alderlake mostly... Zen 3 was 19% higher IPC compared to Zen 2, and that was with a few design changes and slightly higher clocks, thats without a new node, memory, cache, etc.

Zen3 was a ground-up redesign, not "a few design changes".

 

[R0H1T]

Come on it's obvious they have something up their sleeve, these are either zen4c cores or they have x3d designs lined up for later!

It's still quite underwhelming considering a 100% increase in L2 cache, a lot more memory bandwidth and 10% higher clocks. And Zen 3 was launched almost two years ago.

I guess you've benchmarked them, right?

 

[ratirt]

I'm really confused with some comments here. 15% ST gain is quite a bit in my opinion. Then you have the IPC gain which some people misinterpret. AL has the same IPC as 5000 Series AMD CPUs. According to GURU 3d.

So 15% increase is not nothing I would think. Also you have the frequency boost. I'm puzzled how AMD measures the IPC to be honest.
Consider this. The 5800x3d has the same IPC as 5800x and 12900k according to guru3d and yet it is way faster in games due to 3dvcache but lacks in other apps like MT apps in comparison to 5800x due to lower frequency. So IPC is one thing, ST performance is another and general performance is totally different thing. I haven't watched the presentation yet but I'm really going to refrain from speculations and guess what it will be like. Especially, if this is supposed to be something totally different than 5000 series CPUs.
What I'm trying to say is, the IPC and frequency etc. is misleading in any way. You have to look at the bigger picture here.

 

[R0H1T]

The 5800x3d has the same IPC as 5800x and 12900k according to guru3d

For CB15, Intel & AMD generally use a suite of benchmarks to avg IPC gains across a variety of workloads. ADL is definitely faster than zen3, but not that much (purely) on IPC.

 

[noel_fs]

I also feel it won't be enough especially if the rumored 10-15% ST performance improvements for Raptor Lake are true... In some applications Zen 3 is 30% behind in ST performance vs Alderlake.

what applications 30% behind?

they need to work on marketing jesus christ


Anyway, seems good enough, maybe would have expected a little better but im positive that benchmarks will only make it look better


Some people saying amd is behind intel is quite delusional imo. Intel is the one a year behind, Alder lake barely outperformed zen3 while being considerably less efficient. There is no way intel will come out with a >15% ipc improvement when they struggled to get alder lake out.


I believe its actually around around 30% performance improvement when taking clocks into account, thats really good.

Pcie5 also makes the platform more future proof for early adopters.

 

[phanbuey]

what applications 30% behind?

Just some:
Adobe Premier, Handbrake, Microsoft Office (Word, Excel etc.), older single threaded titles (age of empires 4 etc.).

Still though, I would be surprised if Zen 4 isn't faster than Raptor lake. 3Dcache +15% IPC + massive clock boosts + 16 full fat cores vs big.little should be able to beat Raptor Lake, but we'll see.

If intel boosts cache size, ring speeds, pumps clocks and adds more cores they might be able to keep up in some things, but I just don't see Raptor Lake beating a 7950 X3D in anything except a few ST outlier apps.

The mid range will be a price-performance fight tho, which is exciting.

 

[ratirt]

For CB15, Intel & AMD generally use a suite of benchmarks to avg IPC gains across a variety of workloads. ADL is definitely faster than zen3, but not that much (purely) on IPC.

Well, You say purely on IPC but on GURU3d it looks like these 2 are equal. IS there a different way to measure IPC than GURU3d did? If so, which is is the right one since they can't be two ways and each one gives a different result. Maybe different application gives different impression of the CPU's speed. If that is the case (i believe it is) maybe we should wait for the general performance metric instead of believing something that cannot be clearly quantified with numbers.