AMD Ryzen 9 7950X

[Icon Charlie]

I'd like to see performance comparisons at a normalized wattage (65, 88, 105, 142, etc..) between Zen2, Zen3, Zen4 vs. Intel 10th, 11th, and 12th gen.

I agree. This Eco stuff is market speak. Show me REAL data against undervolted CPU's from previous generations.

I am really thinking of under volting my CPU just because of this AMD nonsense. And I am already running a very cool rig right now with great performance and stability.

 

[Mussels]

Wait wut
View attachment 263420

12900K use 1% more power while being 4% faster than 7950X, that means 12900K is more efficient in gaming

.... no. Thats a performance graph at 720p and you cant go using power figures out of nowhere.

Maybe try using w1zzards own results that were measured while in game?

I agree. This Eco stuff is market speak. Show me REAL data against undervolted CPU's from previous generations.

I am really thinking of under volting my CPU just because of this AMD nonsense. And I am already running a very cool rig right now with great performance and stability.

That's really hard to achieve, it's not like reviewers have 50 different platforms ready to go with identical setups to go testing that sort of thing, and then of course individual variance of each CPU happens

 

[nguyen]

.... no. Thats a performance graph at 720p and you cant go using power figures out of nowhere.

Maybe try using w1zzards own results that were measured while in game?


That's really hard to achieve, it's not like reviewers have 50 different platforms ready to go with identical setups to go testing that sort of thing, and then of course individual variance of each CPU happens

How about this chart?

I really love TPU way of using avg FPS/avg power consumption (across 12 games), no room for cherry picking result.

 

[Mussels]

How about this chart?
View attachment 263699

I really love TPU way of using avg FPS/avg power consumption (across 12 games), no room for cherry picking result.

Nice crop, It's not like the rest of the lineup - the 5600x, 5700x and 5800x are more efficient or anything


The only 12th gen that makes a good showing is the 12600/K, it seems everything above that destroys efficiency

The gaming performance is basically identical on all the Zen3 parts.
No one ever buys a 5950x for gaming. No one.
It's like you went right for the worst performing Zen3 part and tried to make up facts to fit the opinion instead of the other way around.

 

[nguyen]

Nice crop, It's not like the rest of the lineup - the 5600x, 5700x and 5800x are more efficient or anything

The only 12th gen that makes a good showing is the 12600/K, it seems everything above that destroys efficiency

The gaming performance is basically identical on all the Zen3 parts.
No one ever buys a 5950x for gaming. No one.
It's like you went right for the worst performing Zen3 part and tried to make up facts to fit the opinion instead of the other way around.

Lol I thought you were talking about direct competition 5950X vs 12900K, but if you like to cherry pick CPU to fit your narrative, how about the 12400F which is Intel 12th?

What else do you have? that Ryzen 5000 can be tweaked for better efficiency? that 12400F can't be tweaked right?

 

[Richards]

Lol I thought you were talking about direct competition 5950X vs 12900K, but if you like to cherry pick CPU to fit your narrative, how about the 12400F which is Intel 12th?
View attachment 263731
What else do you have? that Ryzen 5000 can be tweaked for better efficiency? that 12400F can't be tweaked right?

Exposed him lol

 

[R0H1T]

What else do you have? that Ryzen 5000 can be tweaked for better efficiency? that 12400F can't be tweaked right?

No actually it can't & yes Ryzen chips have a lot more room because they're all unlocked!

Depending on the workload you can expect anywhere between at least 5-40% more efficiency for Ryzen, though you will need to cap power consumption as well.
Which is why locked Intel chips are generally so efficient.

AMD Ryzen 9 7950X in faster or more frugal: Gaming and workstation undervolting with light and shadow | Page 11 | igor´sLAB

Anyone who buys a new AMD Ryzen 7000, no matter which model, will certainly curse the high temperatures and thus also get annoyed about the usually high power consumption under full load. However…

www.igorslab.de

 

[JustBenching]

Nice crop, It's not like the rest of the lineup - the 5600x, 5700x and 5800x are more efficient or anything


The only 12th gen that makes a good showing is the 12600/K, it seems everything above that destroys efficiency
View attachment 263700


The gaming performance is basically identical on all the Zen3 parts.
No one ever buys a 5950x for gaming. No one.
It's like you went right for the worst performing Zen3 part and tried to make up facts to fit the opinion instead of the other way around.

Aren't these graphs SYSTEM power consumption? So wtf are you comparing then? If one CPU gets more fps the GPU will also pull more wattage.

I don't know why this argument keeps on going, it's been demonstrated repeatedly that in gaming, zen 3 efficiency sucks. Or better put, in all lightly threaded workloads like games / autocad / premiere and these kinds of stuff.

 

[Dr. Dro]

No one ever buys a 5950x for gaming. No one.

Hi, I have been summoned

But I am also out of my mind!

Anyways, I agree in general. Zen 4 is a solid improvement when power-normalized IMO, but we've reached a point where processors have gotten plenty fast. No one left Sandy Bridge for Ivy Bridge, hell, most people stayed on Sandy Bridge well into Haswell's lifetime. We're just hitting that point now - I strongly believe that Zen 2 and Zen 3 are the new Sandy and Ivy processors, these will stick around as a performance standard for some time.

It seems obvious to me that if you're after the absolute fastest gaming parts, Raptor Lake will be the way to go, but buyers of Raptor Lake should also be aware that it's a sealed deal. Not that i'm placing too much stock on AMD honoring X670E's long life if they begin to get too far ahead of Intel in performance. I won't forget, or forgive the treatment and cold shoulder given to us X370 adopters.

 

[nguyen]

No actually it can't & yes Ryzen chips have a lot more room because they're all unlocked!

Depending on the workload you can expect anywhere between at least 5-40% more efficiency for Ryzen, though you will need to cap power consumption as well.
Which is why locked Intel chips are generally so efficient.
View attachment 263810

AMD Ryzen 9 7950X in faster or more frugal: Gaming and workstation undervolting with light and shadow | Page 11 | igor´sLAB

Anyone who buys a new AMD Ryzen 7000, no matter which model, will certainly curse the high temperatures and thus also get annoyed about the usually high power consumption under full load. However…

www.igorslab.de

This will be the only time I response to you, every Intel CPU can be undervolted and it takes about 10s doing so.

 

[InVasMani]

Yes, but if you want efficiency just run a 5800X3D on a single core with the rest disabled and have nice big cache for it while it sips power punches well above it's weight for a single core.

 

[R0H1T]

This will be the only time I response to you, every Intel CPU can be undervolted and it takes about 10s doing so.

So can Ryzen & no you need a compatible board for Intel.

 

[Mussels]

Lol I thought you were talking about direct competition 5950X vs 12900K, but if you like to cherry pick CPU to fit your narrative, how about the 12400F which is Intel 12th?
View attachment 263731
What else do you have? that Ryzen 5000 can be tweaked for better efficiency? that 12400F can't be tweaked right?

I'm not arguing against the 12400F at all.
I fully agree some of the intel chips have better power efficiency - but it's not the high end ones people go on about being for 'gamers'

We have maybe one xx400 or xx600 cpu in each generation, and then every model above that is a shitshow

 

[nguyen]

I'm not arguing against the 12400F at all.
I fully agree some of the intel chips have better power efficiency - but it's not the high end ones people go on about being for 'gamers'

We have maybe one xx400 or xx600 cpu in each generation, and then every model above that is a shitshow

What I'm trying to say since the beginning is that making assumption about gaming efficiency from cinebench is just wrong, games are definitely not designed to utilize every core to 100% (it would be a stuttering mess).

I'm glad W1zzard cleared that up by separating productivity and gaming efficiency in the new review format.

Ever since Cinebench become a popular benchmarking tools, people just got the wrong impression about how much CPUs use during gaming and just over-spent on motherboards for absolutely nothing. Now we have expensive X670 motherboards with overkill VRM for nothing.

 

[PurchaseTBD]

Hello, I had originally went over the 7600X review and its comment section which mentioned fresh OS and whatnot but I came across some things that I'm not sure explain it away.

WinRar Compress in the 7600X review has an 11400F faster than anything reviewed in the 7700X&7950X reviews?

Adobe Premiere Pro 7950X review & 12900KS review in april, "A lot of memory is consumed and accessed in the process, over 10 GB for our test scene." was this the same test scene? because windows 11 shouldn't be taking twice as long for the same scene? Adobe Photoshop also had a substantial push back to time to complete which suggests to me they were a different "battery of typical editing tasks", were they different? The gaming FPS numbers I looked over briefly didn't jump this much.

STV-AV1 was mentioned in the comment section also, new build, a 12900K going from 99.6s to 12.8s, large improvements made there for that compressor if the footage was the same, which it appears to be the case when I look at x264, x265, and mp3 results, which made me question the Adobe tests.

 

[fybyfyby]

wtf. All we got is overclocked 5950x with lower efficiency.
5950x is 120W TDP and this monstrocity is 200W+ TDP everywhere.
I guess sticking with my 5950x until 2030 then.

So compare them at the same wattage and you will see. Reason new Ryzens have higher TDP is to make possible to have high frequency on all cores. And thats cool! No one forces you to use that frequencies. You can set TDP lower. You still get high ST performance but also higher MT performance. Thanks to better IPC and higher frequencies on same TDP as 5950x. On the other hand, if you have 5950x, you allready have beast cpu. I have 5900x, so for me 7950x is very much good upgrade.

 

[Dr. Dro]

So compare them at the same wattage and you will see. Reason new Ryzens have higher TDP is to make possible to have high frequency on all cores. And thats cool! No one forces you to use that frequencies. You can set TDP lower. You still get high ST performance but also higher MT performance. Thanks to better IPC and higher frequencies on same TDP as 5950x. On the other hand, if you have 5950x, you allready have beast cpu. I have 5900x, so for me 7950x is very much good upgrade.

I would argue it's not worth changing platform and memory to upgrade from the 5900X processor yet. If you desperately want more gaming kick - flip it and sidegrade to the 5800X3D. You'll lose four cores but optimize the internal architecture and gain all the benefits from the huge cache. Or, add a little extra money and get a 5950X processor, which is immune to the 6-core CCD's limitations and is effectively a dual-5800X setup, having your cake and eating it too, at the cost of being a small bit behind in gaming performance CPU wise (who cares lol)

 

[W1zzard]

WinRar Compress in the 7600X review has an 11400F faster than anything reviewed in the 7700X&7950X reviews?

This is definitely a testing error, will retest and update the reviews accordingly

Edit: this is fixed now

Adobe

I think they should be the same workloads, not 100% sure. Both programs have been updated though. Maybe Adobe optimized their tracking algorithm to be more precise, but runs longer. Photoshop usually gets slower with each new version, the 10% increase in runtime seems plausible.

My benchmarks are not designed to be comparable across test systems over the years, this gives me much more freedom to pick newer/better workloads and scenarios.

AV1

I selected a shorter input video this time to cut back on total runtime of the benchmarks. Running SO many tests takes forever, and I also have A LOT of comparison CPUs. The duration for AV1 does seem rather short now on the fastest CPUs, guess I'll pick something longer for next rebench.

 

[veleciraptor]

If you drop the power usage to 65 watt, it is one of the most efficient CPU ever made.

 

[PurchaseTBD]

would you consider adding VMAF (Video Multi-Method Assessment Fusion) to your selection of tests? perhaps as a synthetic or rendering?
powershell would measure the time it takes ffmpeg to perform the task on a video file

for example, I generated colorbars at 4K, exported a lossless x265 600 frame video under a megabyte and told ffmpeg to compare it to itself using the vmaf model, took 318.4561 seconds on an 6T FX6300 while a 12T 2600X took 46.5407 seconds, I don't believe the contents of the video matter, just the resolution and frame count.

like cinebench you could potentially measure single threaded and multithreaded performance

 

[Denver]

More Zen4 specific updates:

https://twitter.com/x/status/1576335723044622337

 

[blazeddd]

Any chance we could get some updated relative performance charts with the new Nvidia RTX 4090?
Would it most likely be similar for 4K gaming? Mostly concerned about 5600x paired with a 4090 at 4K.

 

[Mussels]

Any chance we could get some updated relative performance charts with the new Nvidia RTX 4090?
Would it most likely be similar for 4K gaming? Mostly concerned about 5600x paired with a 4090 at 4K.

future CPU reviews might use the new GPU's, it takes a lot of time to re-test every combination


Having a 5600x tested would be useless without every OTHER CPU also being tested to compare it against

 

[AnotherReader]

Chips and Cheese has a pretty comprehensive two part deep dive into Zen 4's microarchitecture:

1. Part1
2. Part 2

 

[DMD]

Good morning.
Regarding the problem of slow boot up, is there currently any motherboard more suitable for 7950X CPU or should we wait for AGESA bios update that can improve the boot performance?
Thank you and merry Christmas to all.