AMD Ryzen 7 9700X

[W1zzard]

It has 88W power target. It should be compared directly to the 7700 with 88W power target.

But still a pass for most people.

Just to clarify: 65 W TDP equals 88 W PPT. That's the default setting

You are right, technically, this is really more similar to the 7700 non-X, except for the positioning and the price

 

[Crackong]

All goes to power efficiency

 

[john_]

More like a Zen4+ than a Zen5.

 

[BoggledBeagle]

BTW this gives Intel a slight chance for their upcoming CPUs, they may be better than these Zen 5 chips.

 

[neatfeatguy]

Granted, I have not checked every single review out there but still quite a few sites and TPU are literally the ONLY one who are saying that the 9700X can hold a candle to the X3D CPUs of the 7000 generation.

In fact, TPU said that the 9700X is faster than the 7950X3D (0.8% at 1080p). I can not even remotely find this result and these findings reflected in a single other review.
"All" other reviews show 7800X3D/7950X3D with a considerable lead over the 9700X.

This is why I would like to ask the slightly provocative question: What went wrong with benchmarking at TPU?

For the record, I think you are usually a great source with reliable data. But something does not add up here. Literally no one else but you sees 9700X even in the same ballpark as the 7800X3D and 7950X3D.

I don't want to speak for W1zzard, but if I remember correctly he has his own spots in games where he runs benchmarks - spots that are more liken to what a gamer would experience in gaming and not just the included benchmark.

If that still holds true, that's why I think numbers on TPU tend to be slightly different from most other places that tend to just run with the included benchmark or a beginning spot where the game(s) may not be as demanding. And that is why I like seeing W1zzard's results.

When I used to screw around with benching my 980Ti vs my 570 SLI setup I would find heavy GPU spots and run those a handful of times on each setup to get a better idea of what to actually expect because sometimes the benchmarks included with a game are a very poor representation of what the game will actually demand from your system.

 

[RGAFL]

I agree, on paper this looks meh. Performance per watt is good though. Put the Intel CPU's down to the same power levels and it would be embarrassing for Intel in single thread. Let's see how the real big boys get on next week.

 

[RogueSix]

No.... but Intel are taking care of that themselves at this point...

Exactly. Anyone even considering a Raptor Lake CPU at this time must be clinically insane. And I would even go so far as saying that the same is true for Arrow Lake because no one knows if Arrow Lake will not also croak after a few months of usage due to massive degradation.

Intel does not deserve the benefit of the doubt imo. The only way they can regain trust is by releasing a couple of flawless CPU gens again.

Personally, I won't take any chances. I'm ready to replace my 13900K with Zen 5 as soon as the X870E boards become available (hopefully with the 9800X3D chip in tow). I really wish they would have released the boards today as well. Then I could already grab a X870E board with a 7800X3D (9700X is out after the reviews) and upgrade to 9800X3D later.

 

[Hardware1906]

well, one of AMD good point is power efficiency, so yeah
i guess we will need to wait the 9000X3D

 

[Vario]

Exactly. Anyone even considering a Raptor Lake CPU at this time must be clinically insane. And I would even go so far as saying that the same is true for Arrow Lake because no one knows if Arrow Lake will not also croak after a few months of usage due to massive degradation.

Intel does not deserve the benefit of the doubt imo. The only way they can regain trust is by releasing a couple of flawless CPU gens again.

Personally, I won't take any chances. I'm ready to replace my 13900K with Zen 5 as soon as the X870E boards become available (hopefully with the 9800X3D chip in tow). I really wish they would have released the boards today as well. Then I could already grab a X870E board with a 7800X3D (9700X is out after this review) and upgrade to 9800X3D later.

I mostly agree with your position but I'm skeptical of big change from 7800X3D to 9800X3D.

 

[W1zzard]

and not just the included benchmark.

Confirmed, I'm always using in-game actual gameplay, not the integrated benchmarks. Scene selection goal is "demanding, but not worst case"

 

[AnotherReader]

Performant is not a word in english, do not use it!

It is only in french and Linus uses it because he is from Canada, country of snail-eating ancestry, and because he is confused.

Unfortunately, this word is now in both the Oxford English dictionary and the Cambridge dictionary. You can't blame Canada for that.

 

[R0H1T]

Pardon the French then

 

[SRB151]

I would never get used to 95C by design.

Just wait for Arrow Lake....Intel upped Max. to 105 deg. Be interesting to see if it bounces off 105 constantly after 40-60 seconds like Raptor Lake.

 

[Xuper]

anyone knows what clock was that during testing Blender ? It's 3.8ghz on all cores?

 

[R0H1T]

More like a Zen4+ than a Zen5.

Come on 15-16% IPC >>>> 3-4% from way back when zen+ launched.

The main reason this is looking bleak is that zen 3 - zen4 had IPC+high clock speeds!

 

[AMDK11]

Overall, this is a major disappointment and does not feel like the CPUs are worthy of a new architecture name. At best Zen4+

ST performance bump is ~12%, but it only translates to 5% MT advantage. The only notable improvement is that it's up to 9% faster than the 65W 7700, but in the end, you might as well just save the $80 and get the 7700 instead of 9700X.

AMD made a big mistake by not including their NPU in these, losing the only reason to recommend Zen5 over Zen4.

It's not a question of whether the processor deserves the next-generation designation or not. The Zen5 microarchitecture is completely new from scratch, and the fact that it doesn't achieve much greater IPC growth in the first generation is another matter. Zen5 is a new step in AMD's development and is expected to enable further steady IPC growth.

I understand that many may be disappointed with the performance increase of the Ryzen 9000.

 

[Vincero]

Come on 15-16% IPC >>>> 3-4% from way back when zen+ launched.

The main reason this is looking bleak is that zen 3 - zen4 had IPC+high clock speeds!

That fact that single thread IPC is so high, but multi-thread isn't suggests that even with the additional efficiencies in design and new process node, they are still hitting the power / thermal limits of the silicon.
I guess if that constraint was taken away, i.e. free-up the package power limits and use better cooling then it would look a lot better all around... but then you start getting into more exotic requirements.

 

[BoggledBeagle]

Just wait for Arrow Lake....Intel upped Max. to 105 deg. Be interesting to see if it bounces off 105 constantly after 40-60 seconds like Raptor Lake.

Arrow lake CPUs will have modest temps, if Intel wants them to last more than a few months.

It is good to see that AMD keeps the CPUs cool at stock setting this time.

 

[BoGGe]

It's not a question of whether the processor deserves the next-generation designation or not. The Zen5 microarchitecture is completely new from scratch, and the fact that it doesn't achieve much greater IPC growth in the first generation is another matter. Zen5 is a new step in AMD's development and is expected to enable further steady IPC growth.

I understand that many may be disappointed with the performance increase of the Ryzen 9000.

+ it's performance uplift is probably designed for the server world where the big bucks are.

 

[RandallFlagg]

This thing is getting trounced by a default power 14700K on a majority of workloads.

Yet its real competitor is the not yet released Arrow Lake.

This looks like a real flub for AMD to me. Zen 5 has a lot of architectural changes that should have yielded more performance. The 9600X review shows similar results vs the 14600K.

Unless Intel hoses up bigtime with Arrow Lake (entirely possible), this may be the end of AMDs rise in the desktop space. They've left Intel a big opportunity to leapfrog them.

Their saving grace may be that, because of the nature of the changes to Zen 5, compiler optimizations may not be there to take advantage of those changes.

Imagine the chart below if Arrow Lake merely provides a 5% uplift vs 14700K.

Now do 10%.

 

[R0H1T]

That fact that single thread IPC is so high, but multi-thread isn't suggests that even with the additional efficiencies in design and new process node, they are still hitting the power / thermal limits of the silicon.
I guess if that constraint was taken away, i.e. free-up the package power limits and use better cooling then it would look a lot better all around... but then you start getting into more exotic requirements.

Just increase the TDP then, it's probably power starved.

 

[RogueSix]

I mostly agree with your position but I'm skeptical of big change from 7800X3D to 9800X3D.

Yeah, I'm not expecting a huge boost. It's more like wanting to buy/own the latest instead of the "old" gen at this point in time. Unless AMD has worked some miracles on the cache optimizations, I would guess that 9800X3D vs. 7800X3D is going to end up in the +5% to +10% range max.

The writing for X3D is on the wall after these reviews. These CPUs really should have been on TSMC's 3nm node instead of "4nm" (which is just an optimized 5nm node). AMD's 9000 series is basically Zen 4+ and 5nm+. It's obviously a very incremental generation because AI is sucking up all 3nm capacities at TSMC.

 

[Vincero]

Just increase the TDP then, it's probably power starved.

I sense an XT version may do just that ... probably in a couple of years time.

They could do but don't forget they make SKUs based on demands from OEMs, not just enthusiasts. At the end of the day, if its not worse and potentially improves their operating costs then 'it's an improvement' as far as the AMD bean-counters are concerned.

 

[AnotherReader]

The creator of Y-cruncher, Alexandar J Yee, has analyzed the differences between Zen 4 and Zen 5. It's an interesting read. The summary table explains some of the results we are seeing:

If we look at pure homogenous CPU workloads with no memory bottleneck, here's what I get from just my own tests (largely taken from my own projects):

Workload	IPC Improvement: Zen4 -> Zen5 (Granite Ridge)	Application
Scalar Integer	20% 30 - 35%	C++ Code Compilation Basecase large multiply. Scalar 64-bit NTT kernels.
x87 FPU	10 - 13%	PiFast, y-cruncher BBP (00-x86)
128-bit SSE	-1% (regression)	y-cruncher BBP (05-A64 and 08-NHM)
256-bit AVX	5 - 8%	y-cruncher BBP (19-ZN2)
512-bit AVX512	96 - 98% (basically 2x)	y-cruncher BBP (22-ZN4) and various internal kernels

 

[R0H1T]

We'll see how Turin performs, should give us some more info whether 3nm would've helped or not.

I sense an XT version may do just that ... probably in a couple of years time.

They could do but don't forget they make SKUs based on demands from OEMs, not just enthusiasts. At the end of the day, if its not worse and potentially improves their operating costs then 'it's an improvement' as far as the AMD bean-counters are concerned.

AMD Announces Zen 5-based EPYC “Turin” Processors: Up to 192 Cores, Coming in H2’2024

www.anandtech.com