News AMD Ryzen 9 5950X and 5900X Review: Zen 3 Breaks the 5 GHz Barrier

[jeremyj_83]

With the Anandtech review, we know that Zen 3 has a higher IPC than Tiger Lake and therefore Ice Lake/Rocket Lake as well. Using the estimated score per GHz we see that in Spec2017 Zen3 has a 13.2% higher INT IPC and a 7.77% higher FP IPC. That means that to be competitive Intel will need to keep the same boost clocks on Rocket Lake as it does for Comet Lake.

 

[Red_Viper_Martell]

With thermal performance apparently key to hitting that magic 5.0 number, does anyone happen to know if the IHS will be soldered for these as was done previously?

 

[ingtar33]

There is something wrong with THG's AMD benchmarking rig. Their AMD gaming FPS numbers at both 1080p and 1440p are way down from what others are finding, indicating a driver or hardware fault, reducing the AMD results in comparison to intel; considering one benchmarker was able to achieve parody between these new AMD chips and the top of the line intels with the AMD chip locked to 4ghz, only underlines how badly THG's rig underperformed what most others were benching.

 

[InvalidError]

considering one benchmarker was able to achieve parody

Achieving parody is what every good benchmark should strive for!

 

[a.henriquedsj]

What happened at CS Go? LOL

 

[San Pedro]

The 1440p results here show a much bigger difference between a 2700x and the rest of the CPUs compared to other sites. I wonder what the deal is. I'm trying to decide if I should upgrade or not, and 1440p gaming data will be the determining factor.

 

[San Pedro]

Thank you. I do not understand why Anandtech was seeing such drops for AMD in the charts at 1440 and 2160 resolutions, sometimes even the Ryzen 2 beat it. Tom's did not have that issue.

But the 5600X is the go to CPU for 95% of us!

Anandtech tested 1440 and 4k at low settings, which I'm sure will lead to different results.

 

[seanzeu]

How are these sold out everywhere? Didn't they say we wouldn't have these issues? Wow

 

[mhanxgarox]

According to GamersNexus, in most gaming benchmarks not even 5.2 Ghz OCed 10900K could defeat stock 5950X.
Probably because GN used GTX 3080 in their review which utilized more raw power of 5950X.
For information, Toms used the fallen GPU king : 2080 Ti.

 

[InvalidError]

How are these sold out everywhere? Didn't they say we wouldn't have these issues? Wow

AMD is making a bazillion different things on 7nm and is likely maxed out on all 7nm wafer starts it has available. On-going availability issues for the first many months from launch was inevitable IMO.

 

[Phaaze88]

@mhanxgarox
To add to that, Steve at Hardware Unboxed used a 3090, but he also removed the power limits, which other Steve didn't do.

None of them(reviewers) are wrong though.

 

[NightHawkRMX]

The 5950x is on par with a 10900k in games. Honestly, neither one of those CPUs are a good choice for games.

A 5600x or 10600k will be so close, especially when the latter is overclocked, that you will probably not notice any difference in most games. What you will notice, however, is much less damage done to your wallet.

If Z490 motherboards were priced well I could reccomend the 10600k. I can't with current pricing. Sure you can pair it with a lower tier board but then you loose performance with 2666mhz ram cap and no oc potential.

5600x+included cooler + cheapest b550 vs 10600k with cheapest decent cooler plus cheapest z490.
PCPartPicker Part List

Type|Item|Price
:----|:----|:----
CPU | AMD Ryzen 5 5600X 3.7 GHz 6-Core Processor | $299.00 @ B&H
Motherboard | Gigabyte B550M DS3H Micro ATX AM4 Motherboard | $94.99 @ B&H
| Prices include shipping, taxes, rebates, and discounts |
| Total | $393.99
| Generated by PCPartPicker 2020-11-05 21:30 EST-0500 |

PCPartPicker Part List

Type|Item|Price
:----|:----|:----
CPU | Intel Core i5-10600K 4.1 GHz 6-Core Processor | $264.99 @ Newegg
CPU Cooler | Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler | $24.99 @ Newegg
Motherboard | Gigabyte Z490M GAMING X Micro ATX LGA1200 Motherboard | $139.94 @ Amazon
| Prices include shipping, taxes, rebates, and discounts |
| Total (before mail-in rebates) | $439.92
| Mail-in rebates | -$10.00
| Total | $429.92
| Generated by PCPartPicker 2020-11-05 21:31 EST-0500 |

the 10600k setup is similar performing, but is more expensive, is less efficient, and doesn't support pcie 4.

You can't go wrong with either setup, however.

 

[okbud]

could we re-run this four rank vs dual rank test, cos zen 2 is doing really better with 4 sticks

https://www.tomshardware.com/reviews/amd-ryzen-3000-best-memory-timings,6310-2.html

 

[PaulAlcorn]

According to GamersNexus, in most gaming benchmarks not even 5.2 Ghz OCed 10900K could defeat stock 5950X.
Probably because GN used GTX 3080 in their review which utilized more raw power of 5950X.
For information, Toms used the fallen GPU king : 2080 Ti.

As listed in the article, we used an RTX 3090.

 

[Phaaze88]

CPU | Intel Core i5-10600K 4.1 GHz 6-Core Processor | $264.99 @ Newegg
CPU Cooler | Cooler Master Hyper 212 EVO 82.9 CFM Sleeve Bearing CPU Cooler | $24.99 @ Newegg

That pairing is a bit of a stretch, isn't it?

 

[PaulAlcorn]

There is something wrong with THG's AMD benchmarking rig. Their AMD gaming FPS numbers at both 1080p and 1440p are way down from what others are finding, indicating a driver or hardware fault, reducing the AMD results in comparison to intel; considering one benchmarker was able to achieve parody between these new AMD chips and the top of the line intels with the AMD chip locked to 4ghz, only underlines how badly THG's rig underperformed what most others were benching.

Which benchmarks, specifically? The AMD processors won by a pretty handy margin in our tests.

 

[Cableaddict]

I think the reviewer made a mistake.
The article says, "The Ryzen chips continue to expose 20 lanes of PCIe 4.0 to the user and stick with DDR4-3200 memory as the base spec."

I thought the last Ryzen series (As well as current Intel cpus) support only 16 direct-to-cpu PCIe lanes.
Is that not so?

 

[InvalidError]

I thought the last Ryzen series (As well as current Intel cpus) support only 16 direct-to-cpu PCIe lanes.
Is that not so?

Nope, AM4 has always had x16 for GPU/other + x4 for NVMe/other.

LGA1200 has 16+4 too but Intel ran into issues, again, and had to scrap 4.0 and +4 on Comet Lake.

 

[Sleepy_Hollowed]

Amazing. Can't really even recommend Intel for anything but for the oddity use case of anyone wanting to build a small Home Theater PC or on a regular PC due to the intel SGX requirement for UHD. That's the oddest use case ever though so for anything else, there's no question about using AMD, especially since you can disable the equivalent of the management engine for security purposes, and you can't even do that on Intel CPUs.

 

[NightHawkRMX]

That pairing is a bit of a stretch, isn't it?

Maybe, though 10th gen has actually decent thermals due to the tricks they did to make them easier to cool.

 

[JarredWaltonGPU]

According to GamersNexus, in most gaming benchmarks not even 5.2 Ghz OCed 10900K could defeat stock 5950X.
Probably because GN used GTX 3080 in their review which utilized more raw power of 5950X.
For information, Toms used the fallen GPU king : 2080 Ti.

Sorry, this is wrong. We used the RTX 3090 for the gaming and proviz testing. 2080 Ti was only used on application tests where the GPU isn't much of a factor (like Cinebench, POV-Ray, LAME, etc.).

As for others wondering about different levels of performance from other sites, unless they're using the same game at the same settings with the same CPU and GPU (and at least close on RAM and SSD and mobo), there will be variations in performance.

 

[Phaaze88]

Maybe, though 10th gen has actually decent thermals due to the tricks they did to make them easier to cool.

They HAD to be conservative with those power limits.
But you and I know that they're being held back by them.

Most people who are getting the 10th gen K SKUs, at some point, are going to raise/remove the power limits for max performance... then get slapped in the face by how much hotter they run when doing so.

 

[NightHawkRMX]

They HAD to be conservative with those power limits.
But you and I know that they're being held back by them.

Most people who are getting the 10th gen K SKUs, at some point, are going to raise/remove the power limits for max performance... then get slapped in the face by how much hotter they run when doing so.

This is true, however I was referring more to the whole thinner die thicker IHS aspect of the trickery to make them run cooler.

I know a 3600 can run easily on a 212 evo from personal experience with another PC. Around 70c which isn't bad.

10600k is a little harder to cool but it should still be not an issue on this cooler, unless of course you remove those limits.

 

[Phaaze88]

As someone who doesn't change platforms that often, one thing I have picked up on is that it is more beneficial for me to upgrade on a platform at the end of the line, rather than the beginning or the middle.
My 7820X is still plenty capable, so I'll be waiting to see what the fully mature AM5 and LGA 1700 will do - bar any knock on wood accidents that force me to change, at least...


@NightHawkRMX
~Ohh... gotcha.

 

[NightHawkRMX]

I never think its a good idea to upgrade unless you need to.

The 5600x beats the 3600 EASILY in games, and a nicer cpu would always be nice since I play at 1080p 144hz, but my 3600 does all I need.