AMD Ryzen 7 1800X 3.6 GHz

[L|NK|N]

This is the best and most informative Ryzen review on the web. Thank you for the intense amount of thought and effort that you put into this article W1zzard.

 

[Dethroy]

The Ryzen 7 1800X is by far the least interesting product of its whole family.
The R7 1700 is the most interesting 8/16 SKU imho, the R5 1600X will probably be the SKU of choice for most gamers that value perf/$, whereas the 1500X will probably deliver impressive dual core perf/$ for gamers on a tight budget that don't want to gamble on game devs stepping up their game any time soon to utilize more than 2-4 cores.

Here is a spreadsheet I made that compares most of the Ryzen lineup (1800X all the way down to the 1500X) on a perf/$ basis.

Note: my math assumes 1:1 scaling for both frequency and cores and is therefor not necessarily indicative of real world performance but should still give you a good guesstimate

 

[Fluffmeister]

The 1700 non-x does indeed appear to be the best of bunch (at least when it comes to the 8c16t variety). The upcoming 1600(X) chips might turn into the even sweeter sweet spot.

Shame about the flaky platform, but no doubt the kinks will be worked out eventually, I thank all you beta testers out there in the meantime.

 

[toxzl2]

Wizard, why did you use 2666mhz memory at 1080p

Look at this

https://community.amd.com/community...4/tips-for-building-a-better-amd-ryzen-system

 

[happy medium]

gtx 1080, i7 7700 @ 4.8 = 123.5fps.

gtx1080 with a i7 7700 @ 4.2 = 123.5 fps.

That's a gpu bottleneck.
This review has flaws and Ryzen damage control written all over it.
If it were a Titan P or 1080ti it would have made Ryzen look pathetic.

I see you used ddr4 2600 (the fastest memory for the chip) with the Ryzen.
Why didn't you use ddr4 3400 memory with the 7700?

I see you overclocked the Ryzen chip.
Mabe I missed it, did you overclock the 7700?
Well I guess it would not have made a difference, some of the games were tested with gpu bottlenecks even at 1080p.
Just look at the gtx1080ti scores way higher.
Look at the gtx1080 scores with the 7700k @ 4.2 and 4.8. they are the same!

1440p is worse! Should have used crossfire/sli and watched the Ryzen cpu get crushed with its low memory bandwidth.
Why do these reviewers go out of there way to skew the results?

Hey look my i7 920 @ 4.2 games the same as the 7700k @ 4,2 with a gtx1060 3gb @ 4k.
Really! come on!

Grab 2 gtx1080ti's, a few games that sli well, use medium details, and a 1440p monitor.
Overclock a i7 7700k with ddr4 3000+ to 4.8, and a Ryzen cpu @ 4.0
Watch the Ryzen cpu get clobbered.

You need to eliminate the bottleneck!



I can't believe you guys don't see through this BS.

 

[King Banakon]

nice unbiased review W1zzard .. keep up

 

[birdie]

AMD doesn't need to overturn Intel, if the can just take 5-10% market share they'll make billions

I'm sure they'll make billions even with Zen 1.0, it's just Zen 1.0 is not that attractive for existing Intel CPUs owners (aside from those who use their PCs for very specific tasks like rendering/compiling/encoding using x264).

 

[GreiverBlade]

good show jolly good show AMD....

now gaming test prove that it's a worthy "next" in my buy list but in the form of the R7 1700 instead of trying to stay "blue" and go 7700K as i will upgrade to 1440p before i check on CPU mobo upgrade (sidegrade for my 6600K tho but i don't care ... i like to jump ship ) or also waiting on R5 1600 review ... a 6C/12T for the price of a 6600/7600K ? ... i can't say no ...

as i not only do gaming but sometime encoding and other task where Ryzen can shine ... there is no better option for the price asked.

 

[Divide Overflow]

I hope it doesn't take long for the platform to mature and support much higher speed DDR4 memory. Ryzen performance appears to scale extremely well with faster RAM.

 

[OneMoar]

well that went pretty much as expected
so we need to buy a 120 dollar kit of ram to compete with a cpu thats 160 dollars cheaper
deal of the century

once again amd hedged there bets on having more cores and not a single developer cares

 

[Iceni]

Really interesting review.

I'm dumbfounded by the poor overclock performance. And the fact the chip actually looses performance in some tests. I'm pretty sure the platform needs more development time because of that. By the time we see R5 hopefully they have the bios and drivers functioning a little better.

Also the results against the 2500K. Thank you for those. Performance wise once R5 hits I'm pretty sure it will make for a very hard decision for the older Intel users as to what upgrade path to take. Price is going to be a major factor, but the single thread performance isn't as bad as I was lead to believe on other reviews.

The gaming results as well are not as bad as some would have you think. Intel are still fastest, and have 1080p covered, but for anyone that has more resolution the decision making just got a whole lot tougher. The higher res results just show that AMD have made a great chip, and on their platform you will game and have the horse power to have a productive machine for other tasks.

I'm seriously looking forwards to that Ryzen R5 1600X hitting the reviewers!

 

[mastrdrver]

@Wizzard

Thanks for the review.

Do you think that the bad launch may have been because of the downsizing that has happened over the years? That AMD just didn't have enough people to launch this right? Just a thought I've been having.

Also do you think there will soon (less than a year) board revisions to help solve some problems inherent in the motherboard designs?

Thanks

 

[MrMilli]

You forgot to mention if you're testing these games in DX11 or DX12 (at least the ones that support it).
It's specially the DX12 mode of games that are not running too well on Ryzen. Being a low level API, these games will need optimization to run well on Ryzen.
I already posed about this on TR forum. ComputerBase tested all games in DX11 and DX12. The difference is noticeable.

https://techreport.com/forums/viewtopic.php?f=2&t=119280

Ryzen easily looses 20-30% of performance in DX12 in many games.

7700K vs 1800X:

• 
  Battlefield 1 DX11: 116 vs 122
  Battlefield 1 DX12: 127 vs 90
  Deus Ex: MD DX11: 87 vs 80
  Deus Ex: MD DX12: 83 vs 63
  RotTR DX11: 152 vs 135
  RotTR DX12: 168 vs 117
  Total War: W DX11: 43 vs 40
  Total War: W DX12: 42 vs 30

 

[efikkan]

What is really interesting about this cpu is that if you disable one CCX and do a lil OC, it will perform better in most of these buggy games (everyone is considering it as a legit performance measure ... sadly).
In plus, it will still beat the i7 7700k in some of the other tests.

Once again, AMD performance sucks and of course it's the software's fault. Unlike with GPUs, there is actually very little to optimize for specific CPUs in software, especially since Ryzen is more superscalar.

The issue with gaming is that the CPU workload is not optimized, it's actually riddled with cache misses and branch predictions, which makes the CPU rely mainly on the prefetcher. Even though Ryzen has more theoretical throughput, it has a vastly inferior prefetcher, and is unable to compete even with Sandy Bridge in this manner. Overclocking wouldn't solve issues caused by cache misses, since the penalty is a constant in ns, not in clock cycles. The only resolution is to try to reduce the number of them.

If you're a gamer or a power user which games, Ryzen is simply not good enough. Even though there are a few programs where Ryzen really shines, that doesn't make up for all the cases where it's not. When a product is not universally performing great, the users needs to carefully watch if the product's benefits even apply to their workload. If there are three benchmarks which has nothing to do with your usage where the product really shines, then that's irrelevant.
Let's say a hypothetical user does some gaming, some Photoshop, and of course web browsing, then a quad core from Intel will actually outperform an octa core of Ryzen. As always real performance matters, not theoretical figures. Very few Ryzen buyers ever do enough Blender to ever experience the advantage at all. CPUs such as i7-6800K will also give much more value for a number of use cases.

What ?
this is basically the same performance of the future 1600x (same chip, same clocks, only two cores disabled) for $249, not good enough

Look at all the graphs; for a number of them it's 10% loss, some even 15-20%. If you're buing a GTX 1070 or better you'll be then wasting money.
If they had compared the selection of Intel CPUs, ranging from i5-6600K, to i7-6800K, i7-6900K, etc. You'll see that there would be no significant gaming difference between them, yet Ryzen is far behind. This is not good enough.

It is the last time I respond to you, but... get your facts straight.

PhysX is on github:

https://developer.nvidia.com/physx-source-github

AMD can go and optimize it for their GPUs (or CPUs) if they wanted...

You know, facts don't matter to fanboys, only the narrative to make Nvidia look evil…

 

[Ravenas]

Thanks for the review Wizzard.

I don't agree with you regarding integrated graphics being a con. A con because Intel offers it for nothing more than marketing? Anyone buying a 7700k to use it with the integrated graphics is completely throwing their money in the wrong direction.

Board revisions... Has anyone followed the state of MOBO manufactures regarding AMD? Pre Ryzen, has anyone noted a revision for a MOBO for AM3+ 990 chipset in the last year and a half? How much market share does AMD have to claim before MOBO manufactures pay attention.

I'm just stating I don't agree with the above cons, that is my opinion.

I would have rather of seen the 1800x compared to the 6900k. AMD touted the 1800x as a processor that directly competes with the 6900k, and yet this review is strictly aligned against the 7700k. I'm not really sure how this helps me as a reader?

Not what I expect from AMD regarding overclocking performance. Yes I realize they are aiming for power efficiency and this is not the 1900x, but I honestly am in a small group of consumers who don't care about 65W vs 125W... To say the least, I am let down by the overclocking... I will attribute some of this to your test setup, both the motherboard selection and cooling.

Overall great processor and competition for a company whose market value hovered 5 billion pre launch. (intel 165 billion)

 

[Nkd]

gtx 1080, i7 7700 @ 4.8 = 123.5fps.

gtx1080 with a i7 7700 @ 4.2 = 123.5 fps.

That's a gpu bottleneck.
This review has flaws and Ryzen damage control written all over it.
If it were a Titan P or 1080ti it would have made Ryzen look pathetic.

I see you used ddr4 2600 (the fastest memory for the chip) with the Ryzen.
Why didn't you use ddr4 3400 memory with the 7700?

I see you overclocked the Ryzen chip.
Mabe I missed it, did you overclock the 7700?
Well I guess it would not have made a difference, some of the games were tested with gpu bottlenecks even at 1080p.
Just look at the gtx1080ti scores way higher.
Look at the gtx1080 scores with the 7700k @ 4.2 and 4.8. they are the same!

1440p is worse! Should have used crossfire/sli and watched the Ryzen cpu get crushed with its low memory bandwidth.
Why do these reviewers go out of there way to skew the results?

Hey look my i7 920 @ 4.2 games the same as the 7700k @ 4,2 with a gtx1060 3gb @ 4k.
Really! come on!

Grab 2 gtx1080ti's, a few games that sli well, use medium details, and a 1440p monitor.
Overclock a i7 7700k with ddr4 3000+ to 4.8, and a Ryzen cpu @ 4.0
Watch the Ryzen cpu get clobbered.

You need to eliminate the bottleneck!



I can't believe you guys don't see through this BS.

WTF lol. If anything I would say why didn't wizard test those games with SMT off that seem to benefit at 4.0ghz instead of 3.6, and the integrated graphics CON was not really fair. No one buys this system for intergrated graphics that is coming with zen APU. This review was mostly balanced.

You sound like you are trolling and I shouldn't even respond to you. Who gets a frickin titan x to game at 1080p? Shit gtx 1080 is damn good enough for anything 1440p.

 

[buggalugs]

I think AMD did an incredible job all things considered. If I wanted to build a workstation I would go AMD, and even for mid-range gaming, which is most of the market.

When they sort out the motherboard/bios issues, and memory compatibility, performance will only get better

 

[sweet]

I think AMD did an incredible job all things considered. If I wanted to build a workstation I would go AMD, and even for mid-range gaming, which is most of the market.

When they sort out the motherboard/bios issues, and memory compatibility, performance will only get better

I think the problem is we are not pushing Ryzen in the right way at this moment.

Just have a quick look at ~35% gains after tweaking, very impressive indeed.

 

[Lionheart]

8.6? At least a 9 & I don't understand how the motherboards are horrible?

 

[Basard]

Great review, with and equally great variety of different benches.... I see in the conclusion that the new platform was a pleasure to work with... Sad that everybody dropped the ball with that crap, a lot of people need to get fired, that's BS. I mean, call security and have them kick them all in the asshole on the way out. So many things could have been done better. And there's going to be a lot of overtime worked to fix this shit storm.

 

[MrGenius]

So nobody saw the 7770K typo yet...besides me?

EDIT: And I see it a bunch more times(21 more by my count) throughout the review. Now that I'm looking for it.

Googling i7-7770K to make sure it's not a thing. Nope. Didn't think so...

I mean it's not right?

It's not listed as one of the test setups either.
https://www.techpowerup.com/reviews/AMD/Ryzen_7_1800X/4.html

 

[FordGT90Concept]

I tried searching but couldn't find an answer: does this benchmark reflect High Performance profile enabled? Could explain why Ryzen's power consumption is so low during gaming (or could be the fact it's only using 25-50% of the CPU so the rest just shuts down).

 

[GoldenX]

...CPUs such as i7-6800K will also give much more value for a number of use cases.

How can a more expensive processor + motherboard offer more value?

About the git of physx, read the license, you are not allowed to use it, only compile it. That github is only for the SDK, it's not for implementing the API in any way.

 

[jigar2speed]

What a shame, maybe another time AMD.

You were reading AMD FX review and commented now ?

 

[Naito]

Awesome review as always. It seems it may be best to wait for some of the platform bugs to be ironed out, before I look into it. AMD have definitely sparked my interest this round, though. Haven't touched an AMD CPU since the PII days.