Leaked Ryzen 9 7950X review

[Super Firm Tofu]

Ryzen 7 7700X Edges Past Core i9-12900K In New Benchmark

Zen 4 strikes hard

www.tomshardware.com

View attachment 262988
i9 12900K does not score 8460, it scores 12xxx
View attachment 262989

This new release is looking pretty bland.

The CPU-Z test results in that chart are for 16 threads only - so in other words, the 8 P-Cores against the 7700X. Add in the extra 8 E-Cores of the 12900K and you get the 12000+ multi score.

 

[Vario]

The CPU-Z test results in that chart are for 16 threads only - so in other words, the 8 P-Cores against the 7700X. Add in the extra 8 E-Cores of the 12900K and you get the 12000+ multi score.

If that is the case, it makes no sense to artificially limit thread count for multithreading tests.

 

[wheresmycar]

7950X is only slightly better in Final Fantasy 15.
In everything else it is much slower

View attachment 262966
View attachment 262967
View attachment 262969
View attachment 262970

and those hot hot temperatures

View attachment 262971

View attachment 262972

I think i'll wait for the 7600X/7700X reviews... i dont like the idea of 7000-series being beaten by last Gen chips so fingers crossed less cores, less power and less temps = MORE FPS!!

i have to admit the 5800X3D looks like a beast and a potential upgrade if our current B450 can manage it.

 

[kane nas]

Do you really think that the 5700g and 5800x are faster than the 7950x? The Intel friends remind me of the guy who is drowning and holding on to his hair to save himself.

 

[Gungar]

Does fps number in eco mode (65W) are really nice

 

[fevgatos]

Why is the ram performance on the 12900K so poor?

Here is mine

Saw similar garbage leak at Tom's. It doesn't edge past shit.

Ryzen 7 7700X Edges Past Core i9-12900K In New Benchmark

Zen 4 strikes hard

www.tomshardware.com

View attachment 262988
i9 12900K does not score 8460, it scores 12xxx
View attachment 262989

This new release is looking pretty bland.

Are you using the sammy dies? You can run them at 6000c30-34-34-58 1t with 65535 trefi and hit around 51ns latency. 1.435 volts and 0.95 on sa

 

[mahoney]

So that's why they're not including coolers - you need a proper cooler to cool them.

 

[Waldorf]

@CallandorWoT​

maybe you like playing on screens below 32, me and many others dont,
so anything less than UHD res on +32in screens is useless,
unless you like the screen-door effect or sit 100ft away from it.

completely ignoring that not every game requires me to run +1000Hz@0.0001ms,
to be enjoyable (at that res), or the fact that i can lower
1. in-game res (as screen is still UHD, so game still looks better than on same size screen that runs a lower native res)
2. IQ settings in the game, and increase fps thru lower visuals

 

[Vario]

These things will probably perform higher after first round or two of bios update.

 

[mahoney]

These things will probably perform higher after first round or two of bios update.

Since when does a bios update improve cpu performance? I kept hearing the same bs when ryzen 1st gen came out. People expecting miracle bios performance updates. The only thing helping here is a tuned ram.

 

[Dr. Dro]

It would seem most game engines really, really like the ample L3 in the 5800X3D far more than they like raw IPC, in a way, this somewhat makes sense with modern games being so large and handling so much data (textures, audio, logic, etc). If anything, this proves it's not worth withholding your purchase over FOMO if you want a gaming processor, as the 5800X3D is poised to be a gaming heavyweight throughout this next generation and possibly even beyond. It's like the Core i5-2500K in a way, it was the gamer's favorite processor throughout the entire 2010's.

With lower costs on AM4-based boards and DDR4, it seems like a no-brainer to me.

I'm still waiting for W1zzard's review here on TPU as it's the review source I trust the most, i'm fairly sure it will be a great processor overall - I think people are way too obsessed with chart toppers where in reality, you are going to have a killer chip if you have Zen 2 or newer for pretty much anything out there.

 

[gffermari]

The 5800X3D owners after seeing new next gen cpu reviews

(I’m joking..these cpus are obviously powerhouses on every app.)

 

[BNSoul]

well do you know why the 5800X3D was made? it's literally THE Gaming CPU from AMD that is worse in basically every single existing workload than the 5800X non 3D.

In what planet do you live? Planet stone age Windows I guess, here's a list of productivity apps in which the 3D humbles the regular 5800X and sometimes even the 5950X, including engineering tools, physics simulation, file, photo and video compression and editing, 3D rendering etc etc... in some of them the 5800X3D is 2x faster than the regular 5800X.

Here's an appetizer:
Widely used professional engineering apps for physics and fluids simulations, the regular 5800X takes 177 seconds to complete the task, the 3D just 81... so much for "every single existing workload".

Have a nice look at the link below to check many other apps (there's many more) where the 5800X cannot match the 3D version, these are widely used by professionals, there's more out there than Windows and Cinebench.

AMD Ryzen 7 5800X3D On Linux: Not For Gaming, But Very Exciting For Other Workloads - Phoronix

www.phoronix.com

 

[gffermari]

The people who are using these apps may have a company laptop with intel shixxxxxy processor or desktop connected to an Intel based server.
There’s no one on earth who would buy a 3D to use it in these apps. If there was, he would buy an epyc with 768MB cache instead.

 

[Dr. Dro]

In what planet do you live? Planet stone age Windows I guess, here's a list of productivity apps in which the 3D humbles the regular 5800X and sometimes even the 5950X, including engineering tools, physics simulation, file, photo and video compression and editing, 3D rendering etc etc... in some of them the 5800X3D is 2x faster than the regular 5800X.

Here's an appetizer:
Widely used professional engineering apps for physics and fluids simulations, the regular 5800X takes 177 seconds to complete the task, the 3D just 81... so much for "every single existing workload".

View attachment 263027

Have a nice look at the link below to check many other apps (there's many more) where the 5800X cannot match the 3D version, these are widely used by professionals, there's more out there than Windows and Cinebench.

AMD Ryzen 7 5800X3D On Linux: Not For Gaming, But Very Exciting For Other Workloads - Phoronix

www.phoronix.com

It performs well at these workloads because you're basically getting a processor which contains a single CCD, at a high frequency, of the same kind that goes in AMD's most exquisite (and uber-expensive) EPYC Milan 3D processors. No doubt they're going to sell well to hobbyist-level users of such pro apps or even professionals with comparably low system demands, but the 5800X3D was very much marketed as the company's ultimate gaming processor. Which it is.

Windows users are the vast majority of all desktop segment clients, by the way. It'll forever remain that way until Linux distro makers can agree on basic standardization and come up with an intuitive, cohesive system interface, which is never since they'd rather club each other to death before agreeing that something should be done some way across every distro because it doesn't match their vision of any given feature.

 

[SL2]

If you want to upgrade your ancient[sarcasmmm] 5800X3D you have to wait for its successor..


If you don't know what I'm talking about: there will be 3D cache models in the Ryzen 7000 series, and they will show up later. Have a look!

Spoiler

 

[freeagent]

The time to buy AM5 is when they release X3D

 

[Hifihedgehog]

so i was right with "AMD locks 5800X3D Overclocking to not make Zen 4 look even worse."

Yeah, sure. If you believe that review 100%, then I have a bridge to sell you. While the review is informative and it has confirmed my decision to wait for the X3D version of Zen 4, it is not a slam dunk in evaluation professionalism. It shows the giant chasm of quality level that separates the countless small review sites from the trusted big operations who know their stuff. The biggest red flag is the gaming tests where the 7950X (non-X3D product) is shown to be worse than the 5950X and 5900X (non-X3D products). If you did not have the alarm bells go off in your head during your first pass of reading the review, I encourage you to take a second perusal noting how the performance of the 7950X is below even last generation which does not pass a basic sniff test. At worse, I would expect parity but we know from an overwhelming chorus of official and leaked gaming results that the 5950X is in fact inferior to the 7950X in gaming. The problem here is this lesser known reviewer most likely does not retest their hardware under a perfectly equal testing environment. In fact, we are looking a mulligan stew of application iterations, driver versions, and OS updates.

 

[mahirzukic2]

The first quite complete leak. The results are from September, 13th, so they could be not for the final BIOS version.

Looks legit but please take it with a huge grain of salt.

Videocardz has CineBench scores.

Wow it really shows the efficiency in that 65w ECO mode. In gaming it's mostly equal or better than @170w, and in other programs is obviously slower but not a whole lot. It is remarkably efficient in the 65w mode.
Very good to see it.

Faster than a 32 core 3970X in rendering, nice. As for 5800X3D, it's only good in gaming, while Zen4 is good at everything.

When I saw this, I was thinking Okay, this is definitely my processor for the next build which should last a long time.

 

[Waldorf]

@[B]mahoney[/B]
because thats what happened in the past with ryzens, just naming things like changes related to clocks/voltages/boost/latency etc,
or possibly even "botched" stuff on the side of amd/aega.

not so much as "faster" (e.g. from 100 to 101%), but more as in bringing perf to the level it should have (e.g. from "98 " to 100%).