AMD Ryzen 9 7950X3D

[sLowEnd]

This is such a stupid take. What made you post this?

Probably copium overdose. It's silly to think there's only one kind of enthusiast.

It's like saying car enthusiasts only care about how fast a car can go.

 

[Xero717]

Can you show us some clock frequencies of the Vcache chiplet during gameplay? Would be interesting to see what it maxes out at and if it's basically just a 7800X3D with another chiplet on board and a little more cache.

 

[SL2]

Can you show us some clock frequencies of the Vcache chiplet during gameplay?

I've seen 5000 - 5250 MHz, it was on youtube but I can't find it now.

 

[jsven008]

One Word: Wow. Now this is the CPU AMD should of released to kick-off the Ryzen 7000 series in Sept. 2022. This CPU is everything the original 7950X should of been. Vastly improved efficiency, lower temperatures, and much improved gaming performance. Compared to the 13900k, it has 1/2(!) the power consumption in applications and a 1/3(!) the power consumption in games. Just Wow.

The new Ryzen 3D CPUs feature:
1) Vastly improved power consumption. Roughly 1/3 the power consumption of a 13900k in gaming. Impressive!
2) Noticeably improved temperatures. Lower power consumption results in lower temperatures and the cap at 89C ensures it never hits the "psychologically critical" 90C.
3) Much improved gaming performance. From the 7950X offering 86% of the performance of a 13900k to now beating it in 720P (while using 1/3 the power). Excellent!

The main reasons the initial Ryzen 7000 CPUs had slow sales in 2022 were their unimpressive power consumption, excessive temperatures, and mediocre gaming performance. The new Ryzen 3D chips fix all that. Kuddos to AMD for addressing the concerns with the initial 7000 launch CPUs (7900X and 7950X)!! Whether you like Intel or AMD, as consumers we all win with intense competition like this.

Great improvements, a great new CPU!

On a final note, another great review from W1zzard!! (I'm truly impressed)

 

[Xero717]

I've seen 5000 - 5250 MHz, it was on youtube but I can't find it now.

Thanks, so it could literally operate at the same Frequencies as the 7800X3D just with more vcache.

 

[SL2]

Thanks, so it could literally operate at the same Frequencies as the 7800X3D just with more vcache.

More Vcache? They both have 64 MB.

 

[ratirt]

I thought to myself not bad and then moved to power section. WTF, I sincerely thought @W1zzard made a mistake since the power consumption is so low. I read through the post and review and it hit me, that's what the chip is using. I have to say great chip. I'm really curious about the 7800x3d. That might be a great development to be fair especially for games.

 

[iDont]

Thanks W1zzard for the extensive review!

The reduction in power required for AMD to put out this performance vs. Intel, means to me that AMD has the edge currently. I do think the X3D chips are priced way too high at release date, but the upgrade promise of AM5 relieves this a bit. Seems like a great strategy to invest in AM5 and upgrade whenever the prices drop a bit, like the non-X3D Zen4 chips recently did over here .

I do wonder what Intel will put out whenever they manage to create 5nm or smaller chips. I think the difference in power draw could be (largely?) explained with the lithography technology. This is future talk however.

 

[W1zzard]

Video summary of the review

Would it be correct to assume that the 7800X3D will not require 'Game Mode' and the yucky Xbox Game Bar? That stuff is only required because of the CCD assignment, right?

Correct

@W1zzard Why did you switch to Mario Kart 8 Deluxe from Mario Odyssey with Ryujinx?

Mario Odyssey was difficult to get consistent results with and I had to retest several times on many CPUs, so I switched to Mario Kart 8, which is much easier in that regard. I was hoping to play Zelda far enough to go to places that have meaningful load, but not even time to start playing yet :/ Maybe over the summer for the Fall 2023 CPU Rebench

Can you show us some clock frequencies of the Vcache chiplet during gameplay? Would be interesting to see what it maxes out at and if it's basically just a 7800X3D with another chiplet on board and a little more vcache.

It hits 5.25 max briefly with heavy loads but will often go below 5 or even 4 with games because they dont fully load the cores

I've seen 5000 - 5250 MHz, it was on youtube but I can't find it now.

In apps like Cinebench, yes

In the game tests the relative performance is not consistent with the average of average fps. The order of cpus even change.

Good observation, this is an artifact of the way the two charts are calculated.

For "Relative Performance" each number in each chart is multiplied so that "stock green bar" has a score of 100 in that chart, then everything is averaged
For "Average FPS", all the FPS are averaged, without scaling, so higher FPS numbers have a higher effect on the average

Neither approach is wrong. I guess I could use geomean for Average FPS, but that would give different rankings, too, and would be based on a more complex method than what I currently have, making it harder to understand for the average reader. If someone can make good arguments for either or any other change I'm totally open to improving things

 

[Xero717]

More Vcache? They both have 64 MB.

7800X3D will have 104 L2+L3 Cache and 7950X3D has 144 L2+L3 Cache

 

[SL2]

7800X3D will have 104 L2+L3 Cache and 7950X3D has 144 L2+L3 Cache

True, but you said vcache, specifically.

Can you show us some clock frequencies of the Vcache chiplet during gameplay? Would be interesting to see what it maxes out at and if it's basically just a 7800X3D with another chiplet on board and a little more vcache.

 

[Xero717]

True, but you said vcache, specifically.

There, I specified entire cache for you.

 

[SL2]

There, I specified entire cache for you.

So? You're still wrong about the two models having different Vcache size.

 

[btk2k2]

Neither approach is wrong. I guess I could use geomean for Average FPS, but that would give different rankings, too, and would be based on a more complex method than what I currently have, making it harder to understand for the average reader. If someone can make good arguments for either or any other change I'm totally open to improving things

Geomean for avg FPS would at least help smooth out outliers.

Really though the ideal method for both is probably weighted relative performance. A 20% increase from 50fps to 60fps is far more valuable than a 20% increase from 500fps to 600fps when it comes to improving the gameplay experience but the latter in a simple average will skew the result in favour of the 100fps increase.

 

[Assimilator]

If you care about power you aren’t an enthusiast.

If you make ignorant blanket statements that are completely wrong.

 

[Bwaze]

The main reasons the initial Ryzen 7000 CPUs had slow sales in 2022 were their unimpressive power consumption, excessive temperatures, and mediocre gaming performance. The new Ryzen 3D chips fix all that. Kuddos to AMD for addressing the concerns with the initial 7000 launch CPUs (7900X and 7950X)!! Whether you like Intel or AMD, as consumers we all win with intense competition like this.

Great improvements, a great new CPU!

I'd say increase in platform price was also a huge factor in people still buying Ryzen 5000 instead of current generation.

And I don't know if these 3D CPUs will change the momentum, there is still huge price difference in buying a new gen, even when 7800X3D arrives, or slapping together a 5800X3D, cheap motherboard, cheap RAM and achieving almost the same result, especially if you don't game in 720p or 1080p... In 4K you're basically inside 5%, for perhaps half the price!

 

[Selaya]

its both combined tbh.
ddr5 tax + useless features tax (looking at you, pcie5.0) + lackluster performance

 

[FreezingPC]

Was expecting the difference to be bigger then just 0.3% (applications), good thing for the ones who asked for suck a ridicule CPU to happen.
Tks for the read W1zzard, i will be waiting for that 50+ games comparaison.

 

[Mussels]

The most impressive part of this is that it shows that older gen CPU's still hold up quite well for 60FPS gameplay

a 3300x or 10400F can keep those lows above 100FPS? Damned impressive
(Outlier games might encourage those upgrades faster, but it's still impressive)

Edit: Oh on the long boot times, the entire one (1) AM5 system i've worked on had long ass boot times until we re-seated the cooler - having it off-center or too tight resulted in really long boots but no stability issues when it did boot.

 

[TheinsanegamerN]

4 hrs? If I'm not actively using the PC to do something intensive then it's running AV1 encoding almost all the time, so we're talking more like 16 hrs/day. That's basically less time than between generations AND I'm on a superior platform (AM5) AND am not cooler-constrained and can use my fav air cooler AND have a better CPU all along.

If you're working heavy workloads like that for such long periods of time, you shouldnt be buying zen 5 or raptor lake. At that level you're looking at professional workstation level CPUs.

And also - Ferraris? When talking about $132 dollars? My guy stop using these brain dead analogies because you have no clue what you're saying.

Pot meet kettle.

 

[Adonisds]

Video summary of the review

Correct


Mario Odyssey was difficult to get consistent results with and I had to retest several times on many CPUs, so I switched to Mario Kart 8, which is much easier in that regard. I was hoping to play Zelda far enough to go to places that have meaningful load, but not even time to start playing yet :/ Maybe over the summer for the Fall 2023 CPU Rebench


It hits 5.25 max briefly with heavy loads but will often go below 5 or even 4 with games because they dont fully load the cores


In apps like Cinebench, yes


Good observation, this is an artifact of the way the two charts are calculated.

For "Relative Performance" each number in each chart is multiplied so that "stock green bar" has a score of 100 in that chart, then everything is averaged
For "Average FPS", all the FPS are averaged, without scaling, so higher FPS numbers have a higher effect on the average

Neither approach is wrong. I guess I could use geomean for Average FPS, but that would give different rankings, too, and would be based on a more complex method than what I currently have, making it harder to understand for the average reader. If someone can make good arguments for either or any other change I'm totally open to improving things

It all makes sense now, thanks. I will always use the relative performance because it seems weird and arbitrary to make, for instance, CS GO have a much higher weight than Cyberpunk just because it is able to reach higher FPS

 

[Kaleid]

Looks great tbh. Looks really good for the 7800X3D.

Yes, although I'd rather see that running at 5.5Ghz. Maybe 8800x3d

 

[Godrilla]

Interesting no 7900x3d reviews out. FYI my local microcenter has 50% more 7900x3d vs 7950X3D in stock. 24 vs 16 respectively.

 

[Gica]

The AMD Empire Strikes Back.

And attack the wallets. As we anticipated, they are desperately looking for an additional 1-2 percent strictly for advertising and the well-being of their personal wallet. Once this advance...colossal (+/- 2%, LOL) is obtained, the money migrates from the fans' wallets to the shareholders' wallets.
It is not for nothing that it is the worst in terms of efficiency/dollar, even in gaming. For less than 2% difference between 7950X3D and 7700X or 13600K (4K gaming), you pay ... I think double, but who else is like you?
The truth is that gaming at 200 FPS is much better than gaming at 195 FPS. The difference can be seen even from the Moon. Or not?

 

[Kaleid]

And attack the wallets. As we anticipated, they are desperately looking for an additional 1-2 percent strictly for advertising and the well-being of their personal wallet. Once this advance...colossal (+/- 2%, LOL) is obtained, the money migrates from the fans' wallets to the shareholders' wallets.
It is not for nothing that it is the worst in terms of efficiency/dollar, even in gaming. For less than 2% difference between 7950X3D and 7700X or 13600K (4K gaming), you pay ... I think double, but who else is like you?

Obviously some do a bit more than gaming and if helps secure AMDs future than I'm quite for it. We can't go back to intel monopoly.
Most obviously do not need these CPUs though, doubt here at an enthusiast website either. I sure don't