AMD Ryzen 9 7950X3D

[thewan]

I wonder why they removed the "diffused in xxx". Is the entire thing made in my country now? Thats probably the only good thing my country is doing, albeit mostly on foreign labor, so its not really thanks to my people. So maybe we're good at nothing after all.

Besides that, excellent review, we need all the less heat output we can get.

 

[Gameslove]

No comments... Ryzen R7 5800X3d is platinum legendary king of CPU.

 

[Space Lynx]

I wonder why they removed the "diffused in xxx". Is the entire thing made in my country now? Thats probably the only good thing my country is doing, albeit mostly on foreign labor, so its not really thanks to my people. So maybe we're good at nothing after all.

Besides that, excellent review, we need all the less heat output we can get.

foreign labor is still mostly good for your economy - they still buy things and pay taxes, keeping the local and national economies churning.

Where does this silly notion that you can't use your favorite cooler on intel comes from? Im using a u12a, happily runs ycruncher at 330 watts on a 13900k.

u12a is one of best coolers ever made imo, I always regretted not spending a little extra and getting one. I am not super impressed with my FC140, it does a good job overall in gaming, but it can't handle synthetic tests as well Noctua for some reason, but gaming temp wise it matches Noctua toe to toe.

Not sure why it saturates more quickly than Noctua though, maybe my case just has bad airflow, that is probably it really.

One question lingers in my mind about these new X3D CPU's. How does the performance compare in gaming when using the on-chip video between the X3D vs. non-3D chips?

check other reviews, I am sure Anandtech, gamersnexus, guru3d, linustechtips, jayz2twocents, someone somewhere out there has that review for you.

 

[JustBenching]

u12a is one of best coolers ever made imo, I always regretted not spending a little extra and getting one. I am not super impressed with my FC140, it does a good job overall in gaming, but it can't handle synthetic tests as well Noctua for some reason, but gaming temp wise it matches Noctua toe to toe.

Not sure why it saturates more quickly than Noctua though, maybe my case just has bad airflow, that is probably it really.

And I regret not getting the fc 140

 

[efikkan]

It seems to me that the biggest problem with this setup is all the software hoops you have to jump in order to get the CPU working properly.

And how many Windows updates will it take before this backfires and it becomes a performance penalty instead?
Tweaking scheduling to squeeze out a minuscule amount of performance can just as easily go the other way, and this will be yet another thing the QA teams need to test for regressions.

 

[5 o'clock Charlie]

and whether upcoming A320 chipset motherboards can help bring the platform cost down.

I assume A320 was a typo for A620 on the conclusion page? Thank you for the thorough review @W1zzard ! It was a great read

 

[MxPhenom 216]

For everyone arguing with the weird power doesn’t matter posters, these guys have already purchased an Intel processor and are trying to justify their purchasing decision and company allegience.



For everyone else that use these reviews to help them build or upgrade and are not holding out hope that a company will reward them based on their loyalty, these X3D CPUs are exactly what many of us want. Lowest power, among highest gaming response at many resolutions, great efficiency, etc. Price becomes the final deciding factor which is based on your disposable income or the needs of your job if this is a work PC as well.

They don't need to justify it at all. Majority of morons who complain about power consumption seen in review graphs don't have experience with the CPUs anyways. You can control Intel power consumption pretty damn easily with some BIOS tweaks to cut down the voltage the boards like to push on these chips at default. It way overshoots it, and its not necessary at all. You can pull the voltage back quite a bit with a few tweaks, chop power consumption almost in half with virtually zero performance loss.

And i thought this was a PC enthusiast forum where people might take some of this into account before commenting...

 

[Daven]

The 12600k was 4% slower @1080p in games with DDR4 3600 CL14 than it was with DDR5 6000 CL36 in Hardware Unboxed Memory Scaling review. I'd expect the 13900k to suffer a similar performance reduction. That would put the 13900k into around the 13700k's numbers in this review, and make it pretty obviously slower than the 7950X3D. DDR4 3600 CL14 isn't even all that cheap. For example, Gskill Flare X5 6000 DDR5 CL36 2x16 kit is $110 right now on pcpartspicker. The cheapest DDR4 3600 CL14 2x16 kit is $100. Cost savings for RAM have become negligible.

Besides, I don't see the point in 'saving money' when you're buying $500+ flagships, and if money is no issue, there's not much separating the 13900k vs 7950X3D. However, if you're just gaming and want value you're better off with something like a 13700k/13600k+DDR4.

In addition, if you decide to use some DDR4 lying around, you have to buy a DDR4 socket 1700 motherboard. That means upgrading everything to go any further the next time.

 

[A Computer Guy]

check other reviews, I am sure Anandtech, gamersnexus, guru3d, linustechtips, jayz2twocents, someone somewhere out there has that review for you.

I know but I like to hang out here.

 

[evernessince]

The point of high end Intel is you can pair memory 2000 MT faster than on AMD, so why wouldn't you.

13900K/7950X3D 6000 MT matching performance.
Is not the same as
13900K at 7800/8000 MT vs 7950X3D 6000 MT.

The problem is that not all retail 13900Ks are guaranteed to hit 8000 and you almost certainly want one of the top end motherboards with that to get the higher layer count and improved signal integrity.

At the end of the day you are talking $700 extra on the RAM and Mobo plus an extra $160 on the CPU when you could just get a 7800X3D and get way better power consumption, platform longevity, and similar if not better performance depending on your game selection. 1% lows on the 3D cache chips are insanely good. There's really no need for this many cores for the average consumer anyways.

The bigger question you'd have to ask yourself after all that is if the 13900K would even be the top performer vs a stock or tuned AMD comparative system. The 13900K gains less than 1% at 1080p with DDR5 7400, I'm not really sure it could beat the 7950X3D.

Outright gaming performance is slightly disappointing that it's not further ahead of Intel, basically tied.

But what kills Intel's 13900K and 13900KS is this:

View attachment 285683

Ryzen 9 7950X3D im Test gegen 5800X3D und 13900K(S)

Zen 4 3D ist da. Im Test erweist sich AMDs Ryzen 9 7950X3D gegen 7950X, 5800X3D und Core i9-13900K(S) als schnell und effizient.

www.computerbase.de

I mean the same performance but literally double (!!!) the power consumption to achieve it. Intel needs a total re-design at some point soon.

Game selection is everything with the 3D cache chips. Many titles that massively benefit like factorio and MS Flight sim were not included here. 1% lows aren't included here either. VR sees a very large benefit.

It doesn't matter in terms of cooling, which is the context.

Heat load absolutely matters for cooling. Not just that it makes the CPU harder to cool but also that said heat is being dumped into the room.

13900KF on Amazon.com: 568$
7950X3D: 700$ + mandatory DDR5

I can use a 13900KF at full, 4 hours at day, for 5 years, to reach the 132$ difference. I've to use 13900KF for more than 10 years to justify the price difference between them.

You can buy whatever you want, but the power consumption on a high end computer it's almost nothing to his retail pricing.

Both are excelents processors, but Intel it's cheapest and better on applications, almost a tie on games. You choose.

To be fair, you are going to reach the price difference the instant you want to upgrade that Intel system as it's a dead end platform.

Mind you, if you are fine with getting 2nd best like the 13900KF, you'd be fine with something like a 7950X, which is currently $589.

I don't see the point of only using a 13900K or 7950X for only 4 hours a day. If you aren't putting all those extra cores to use you are wasting your money. Might as well get a 7800X3D at that point and spend even less money as compared to the 13900KF.

And was it really necessary to use the 13900KF for comparison when the 13900K is a single dollar more?

At $.40 per kWh, the price difference would cover 330 kWh of electricity. That's 330,000 watts.

In the multithreading test, AMD hit 140 watts, intel 276. That's a difference of 136 watts. That is 2426 hours of operation. At 4 hours a day, 365 days a year, it would take you 606 days of operation, or 1.6 years, to break even.

Now, fi you are going off of the application tests, AMD pulls 79 watts, intel 169. Difference of 90 watts. 3667 hours of operation. At 4 hours per day, 917 days, 2..5 YEARS, to break even.

If breaking even after two and a half YEARS on $132 in electricity is a major concern for you, you are not in the market for $500+ CPUs in the first place. The "muh power bill" arguments simply put, do NOT work out. You're fretting over the cost of gas while browsing ferraris.

The 7950X3D and 13900K are targeted at professionals and prosumers that are intending on using those cores and they may often including running multi-day workloads letting the system run nearly 24/7. Numbers also don't take into account cooling costs during the summer. You are probably looking at $600+ for each 13900K system over the 7950X system per year if you are running these in a professional capacity.

It's very easy to see why this processor would be appealing to everyone but those that just buy the best as epeen.

They don't need to justify it at all. Majority of morons who complain about power consumption seen in review graphs don't have experience with the CPUs anyways. You can control Intel power consumption pretty damn easily with some BIOS tweaks to cut down the voltage the boards like to push on these chips at default. It way overshoots it, and its not necessary at all. You can pull the voltage back quite a bit with a few tweaks, chop power consumption almost in half with virtually zero performance loss.

Intel's 13th gen CPUs do not power scale that well:

You can limit Ryzen 7000 series and loose little performance. With Intel you are going to loose a very significant chunk.

 

[zlobby]

With one simple trick and the entire latest generation of intel CPU are obsolete.

 

[JustBenching]

The problem is that not all retail 13900Ks are guaranteed to hit 8000 and you almost certainly want one of the top end motherboards with that to get the higher layer count and improved signal integrity.

At the end of the day you are talking $700 extra on the RAM and Mobo plus an extra $160 on the CPU when you could just get a 7800X3D and get way better power consumption, platform longevity, and similar if not better performance depending on your game selection. 1% lows on the 3D cache chips are insanely good. There's really no need for this many cores for the average consumer anyways.

The bigger question you'd have to ask yourself after all that is if the 13900K would even be the top performer vs a stock or tuned AMD comparative system. The 13900K gains less than 1% at 1080p with DDR5 7400, I'm not really sure it could beat the 7950X3D.

View attachment 285725



Game selection is everything with the 3D cache chips. Many titles that massively benefit like factorio and MS Flight sim were not included here. 1% lows aren't included here either. VR sees a very large benefit.



Heat load absolutely matters for cooling. Not just that it makes the CPU harder to cool but also that said heat is being dumped into the room.



To be fair, you are going to reach the price difference the instant you want to upgrade that Intel system as it's a dead end platform.

Mind you, if you are fine with getting 2nd best like the 13900KF, you'd be fine with something like a 7950X, which is currently $589.

I don't see the point of only using a 13900K or 7950X for only 4 hours a day. If you aren't putting all those extra cores to use you are wasting your money. Might as well get a 7800X3D at that point and spend even less money as compared to the 13900KF.

And was it really necessary to use the 13900KF for comparison when the 13900K is a single dollar more?



The 7950X3D and 13900K are targeted at professionals and prosumers that are intending on using those cores and they may often including running multi-day workloads letting the system run nearly 24/7. Numbers also don't take into account cooling costs during the summer. You are probably looking at $600+ for each 13900K system over the 7950X system per year if you are running these in a professional capacity.

It's very easy to see why this processor would be appealing to everyone but those that just buy the best as epeen.



Intel's 13th gen CPUs do not power scale that well:

View attachment 285721

You can limit Ryzen 7000 series and loose little performance. With Intel you are going to loose a very significant chunk.

That last graph doesn't tell you the whole story. I would even call it misleading. He is using tdp limits on zen 4, and tdp is not power consumption. A 7950x at 35w consumes 50w+, so he is basically comparing 35 vs 50w. The actual efficiency difference is way too small, around 10%

 

[W1zzard]

One question lingers in my mind about these new X3D CPU's. How does the performance compare in gaming when using the on-chip video between the X3D vs. non-3D chips?

No changes to the IGP, should be the same performance

I assume A320 was a typo for A620 on the conclusion page? Thank you for the thorough review @W1zzard ! It was a great read

Yup lol, fixed

1% lows aren't included here either.

Wait, what? Look again

 

[evernessince]

That last graph doesn't tell you the whole story. I would even call it misleading. He is using tdp limits on zen 4, and tdp is not power consumption. A 7950x at 35w consumes 50w+, so he is basically comparing 35 vs 50w. The actual efficiency difference is way too small, around 10%

Incorrect, the article from which the last graph was pulled from used total package power: https://www.anandtech.com/show/17641/lighter-touch-cpu-power-scaling-13900k-7950x

Wait, what? Look again

Whoops, my bad. Somehow missed the minimum FPS page.

 

[Space Lynx]

5800x3d is still sweet spot for me if I can get one used, i play soooo many games that benefit from x3d cache, like starcraft 2, FFXIV, MS Flight Sim, Factorio, and the list goes on and on.

 

[W1zzard]

Whoops, my bad. Somehow missed the minimum FPS page.

It's a new addition with this review

 

[A Computer Guy]

5800x3d is still sweet spot for me if I can get one used, i play soooo many games that benefit from x3d cache, like starcraft 2, FFXIV, MS Flight Sim, Factorio, and the list goes on and on.

If buying new it's up for an amazing deal today I recently discovered.

 

[AnotherReader]

I can continue but I think the point it's clear. You need to update your biases about DDR4 and DDR5 (and I'm writing this from a computer with DDR5).

DDR4 vs DDR5 does matter for new games especially the ones that use raytracing.

 

[HD64G]

I can understand that one can select a performance superior CPU to slower one even if it consumes double and needs much better and expensive cooling. But to say that you would buy an equal -at best- performing CPU that does so, only fanboyism can reason that decision, especially when AM5 will have more future as a platform.

 

[JustBenching]

Incorrect, the article from which the last graph was pulled from used total package power: https://www.anandtech.com/show/17641/lighter-touch-cpu-power-scaling-13900k-7950x



Whoops, my bad. Somehow missed the minimum FPS page.

Incorrect yourself. Maybe actually read the review the next time. This is from the 3rd page.

https://images.anandtech.com/graphs/graph17641/130462.png

At 65w tdp the zen 4 part uses 90w. That is 40% more power

I can understand that one can select a performance superior CPU to slower one even if it consumes double and needs much better and expensive cooling. But to say that you would buy an equal -at best- performing CPU that does so, only fanboyism can reason that decision, especially when AM5 will have more future as a platform.

Again with the cooling... When will that game die?

Only fanboyism can explain someone repeating the same nonsensical argument

 

[Jism]

Why is this test performed with only a Aircooler and not a beefy watercooling for such a higher end CPU?

 

[Gasaraki]

13900KF on Amazon.com: 568$
7950X3D: 700$ + mandatory DDR5

I can use a 13900KF at full, 4 hours at day, for 5 years, to reach the 132$ difference. I've to use 13900KF for more than 10 years to justify the price difference between them.

You can buy whatever you want, but the power consumption on a high end computer it's almost nothing to his retail pricing.

Both are excelents processors, but Intel it's cheapest and better on applications, almost a tie on games. You choose.

Also the hoops you have to jump through. Page 5, jeez...

 

[Kaleid]

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

That's a silly claim. Obviously Switch players aren't gamers either because they game with 30 watts or so max.

I wouldn't want my stereo or TV to waste a lot of energy either for no good reason either. But obviously Intel also knows this and they will create something more efficient in the future..

 

[FLEXOBENDER]

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

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

 

[Panther_Seraphin]

That last graph doesn't tell you the whole story. I would even call it misleading. He is using tdp limits on zen 4, and tdp is not power consumption. A 7950x at 35w consumes 50w+, so he is basically comparing 35 vs 50w. The actual efficiency difference is way too small, around 10%

And Intel is just as misleading as well, but this is what happens when marketing overules engineering. At "stock" Intel is deciding to use over 75 watts more than their already absurdley set 253 watt TDP limit.

https://images.anandtech.com/graphs/graph17641/130462.png

What this is showing to me is that Intel needs a complete architecture refresh vs the constant refreshes from the intial Core architecture back in 2011. We have hit the end of the Pentium 4 era again with intel unfortunately and I honestly look forward to seeing if they manage to pull off Core 2.0. The Original E/Q6600s were God Sends and then the evolution in the i7 920 and 2600/2700k onwards were amazing.