AMD Ryzen 7 1800X CPU Review

[cryoburner]

This is a $500 chip that from all the tests above, lags behind a $350 chip (i7 7700)
NOT ONLY in games, but in most desktop-office uses, and some workstation uses as well (like autocad, solidworks, etc.).
It's $150 more expensive than the i7 7700 and it only surpasses it, in rendering uses and handbrake. That is neither a win, nor what was suggested by amd before release. For all our sake i hope this can improve down the road with better support.

Intel's own $1,050 i7 6900k also "lags behind" their $350 i7 7700k when it comes to lightly-threaded performance. You simply can't expect a processor with 8 cores to compete with another that has 4 higher-clocked cores when those additional cores aren't even being used. Most existing games don't yet really benefit from having more than four cores, since Intel's high pricing of their 6 and 8 core chips has kept them as a niche product that most game developers haven't bothered optimizing for. So sure, if you primarily use lightly-threaded applications, a higher-clocked quad-core chip might be better for you. This initial batch of 8 core Ryzen CPUs is not explicitly targeting that use case though, but rather that of Intel's 8 core chips, and in that scenario, they offer comparable performance at less than half the cost.

And I would say that it's at least a "win" that AMD's CPUs are once again roughly on-par with Intel's in terms of efficiency, and use an architecture that should benefit a bit more from compiler optimizations aimed at Intel CPUs. They don't necessarily have to beat Intel's chips in terms of raw performance when they can offer comparable products at more reasonable prices. Ryzen should at least make AMD's chips competitive again, and that can be considered a win for everyone.

 

[elbert]

Now that Ryzen is out game developers should be able to optimize for the new cores. Gamers should not have to turn off SMT to gain fps once the games are optimized. I look for optimized games to pick up performance on Ryzen 7 before R5 and R3's hit the market.

 

[Lucky_SLS]

u guys could have swapped in bf1 for bf4 to give us a taste of the dx12 optimized performance...
but good job. a solid initial review.

 

[gio2vanni86]

Well i got one but im more a gamer so to see it not do well or even equal to my 3570k is just a let down. So exchange and waiting for Skylake extreme. My AMD venice core was something back in 05. I'll still cherish that chip. Ryzen can go Rise itself from its own grave.

 

[anbello262]

"(...) if you specifically need an eight-core processor for content creation, you can go with the 1800X and give up some gaming alacrity, or, if high frame rates are top priority, buy the Core i7-7700K knowing it won't be as fast elsewhere."

And people are calling bias, just because they don't like the facts they read.

I find these new cpus quite attractive, but I certainly don't want to feel like I'm on a beta and tinker/troubleshoot too much, so I'll personally wait about 6 months to a year before buying one.
But again, I find this to be really interesting. If I get serious with video editing I will look for one of these.

 

[Crashman]

Basically, this is how I did things in the PREVIOUS decade:
1.) Pick the CPU that's best for your needs.
2.) If it's nearly a match between an Intel and AMD processor, remember "who needs your money the most". 😀

 

[Jpwner]

https://imgflip.com/i/1kop0o

¯\_(?)_/¯

 

[InvalidError]

Now that Ryzen is out game developers should be able to optimize for the new cores. Gamers should not have to turn off SMT to gain fps once the games are optimized.

In principle, enabling SMT while the workload shouldn't be sufficient for it to be necessary shouldn't produce any significant performance degradation. Hopefully, this is only an OS scheduling issue (Windows issuing work to a core's two threads instead of prioritizing splitting it between cores) and can be worked out with a software patch. If the issues are in the instruction scheduler or elsewhere in the CPU's pipeline, then that could be problematic.

 

[cryoburner]

Benchmark testing is done this way:

- On CPUs, they use the highest end GPUs to prevent the CPU from being a bottleneck. This produces the highest numbers possible.

- On GPU testing, they use the highest end CPUs to prevent the GPU from being a bottleneck. This produces the highest numbers possible.

With that logic, pairing a 1800X with a Titan X is ideal because this means that the GPU cannot be a bottlneck. Pairing a high end CPU with a mid range GPU like an RX-480 or GTX 1060 would throw the numbers way, way, way off. You can quote driver stats and BIOS updates all you want, but it doesn't work that way in real life product testing.

While that may be mostly true, I can also see the reasoning behind wanting to see test results from both brands of cards as well. Nvidia are most likely to optimize their drivers to run best on Intel's CPUs, simply because those have the largest market share, and because those are what reviewers will undoubtedly be benchmarking their cards with. There's undoubtedly going to be less incentive for them to optimize their drivers for AMD's CPUs. AMD, on the other hand, will not only want their graphics drivers to be optimized for Intel's CPUs, but also for their own. They still need good Intel performance to place well in benchmarks, but it's also in their best interest to make their own CPUs look good. As a result, their driver optimizations could potentially be more fair to both brands. Even if the differences are minor, it would at least be interesting to see tested, to verify whether any such discrepancies existed.

And sure, you don't want the graphics card to be a bottleneck in a CPU benchmark, and AMD hasn't released any cards at the GTX 1080 level yet, but that could be accounted for simply by dropping resolution further. An RX 480 at 720p might not create any more of a bottleneck than a GTX 1080 at 1080p.

 

[bignastyid]

Lets keep it civil. Otherwise bans will be handed out.

 

[albundyhere]

i'm leaning heavily on the handbrake encoding so the ryzen is something i would consider, but not at twice th cost of a 6700k. encoding takes the most amount of time to do out of any task. but i need to find a better review. many of the benchmarks are skewed. the 9590 and the 8350 are used interchangeably, and the subset of processor comparison is very slim. also on the handbrake encoding, what is the benchmark based off of? x264 or HEVC and at what quality setting and resolution (4k?) and at what duration? what are metrics details?

 

[ttt_2017]

I think that AMD made a mistake comparing their 8 cores chips against Intel i7 7700K chip ... bad strategy ...

They should compare those chips against 6 and 8 cores intel chips only . and release a 4/8threads cores or 6 cores Ryzen with very high clock speed ( 4.5 Ghz) and Attack the i7 7700k chips with it ... and I think they can clock their 4 or 6 CPU Higher than Intel out of the box.

 

[Crashman]

Benchmark testing is done this way:

- On CPUs, they use the highest end GPUs to prevent the CPU from being a bottleneck. This produces the highest numbers possible.

- On GPU testing, they use the highest end CPUs to prevent the GPU from being a bottleneck. This produces the highest numbers possible.

With that logic, pairing a 1800X with a Titan X is ideal because this means that the GPU cannot be a bottlneck. Pairing a high end CPU with a mid range GPU like an RX-480 or GTX 1060 would throw the numbers way, way, way off. You can quote driver stats and BIOS updates all you want, but it doesn't work that way in real life product testing.

While that may be mostly true, I can also see the reasoning behind wanting to see test results from both brands of cards as well. Nvidia are most likely to optimize their drivers to run best on Intel's CPUs, simply because those have the largest market share, and because those are what reviewers will undoubtedly be benchmarking their cards with. There's undoubtedly going to be less incentive for them to optimize their drivers for AMD's CPUs. AMD, on the other hand, will not only want their graphics drivers to be optimized for Intel's CPUs, but also for their own. They still need good Intel performance to place well in benchmarks, but it's also in their best interest to make their own CPUs look good. As a result, their driver optimizations could potentially be more fair to both brands. Even if the differences are minor, it would at least be interesting to see tested, to verify whether any such discrepancies existed.

And sure, you don't want the graphics card to be a bottleneck in a CPU benchmark, and AMD hasn't released any cards at the GTX 1080 level yet, but that could be accounted for simply by dropping resolution further. An RX 480 at 720p might not create any more of a bottleneck than a GTX 1080 at 1080p.

All of our testing on the previous CPU generation showed that AMD's graphics cards work best on Intel CPUs We did full Nvidia to Intel/AMD and Radeon to Intel/AMD to see which card lost the most performance on the slower AMD CPU, and it was consistently the Radeon.

Historic trends do not necessarily apply to new architecture, so we should be asking for another one of those articles 😀

 

[FormatC]

Same with workstations. I especially used a Quadro P6000 (it's faster than Titan X Pascal) to see, how this apps were running into the CPU-bottleneck. With a smaller card it might be difficult to mention this big differences. But to be honest: nobody is pairing a 1000 Euro CPU with a 300 Euro VGA card. But what I mentioned very often, is the optimization of all this pro apps for Intel CPUs. You can try what you wan't - software from A-brands like Dassault, Solidworks or Audodesk or simply the fastest Freeware. If its already compiled, it is optimized for Intel. The reason is easy to find: in the last years nobody had a reason to spend a lot of hours to make a lame Bulldozer a tick faster. I hope, the programmers will change their mind to see more AMD optimized libraries on the market.

If some charts graphics have small issues (bar colors etc.), I say sorry. We have now the time to correct it. But if AMD gave me the sample only two days (!) before launch, it was simply impossible to proof all this things in this short time twice. Over 60 benchmarks in this small time windows were also for me simply too much. Ask AMD, why they sent the CPUs so late to Europe 😀

Someone asked, which board died. It was an Asus Crosshair Hero VI and it will be investigated now, what happened with this board. Even a flashback wasn't successful, it's simply dead. This may happen, I had also died Intel boards from various brands. This is called destiny, random, bad luck, whatever.

 

[Ishan_bansal]

I'm looking at 1700/7600K/7700K for 1080p gaming. The Processor Mobo combo for 1700/B350 and 7600K/Z270/Cooler costs nearly the same, but for 7700K it's around 150$ more expensive than the other two. This has made it quite difficult for me to arrive on a decision.Any suggestions?

 

[FormatC]

I'm looking at 1700/7600K/7700K for 1080p gaming. The Processor Mobo combo for 1700/B350 and 7600K/Z270/Cooler costs nearly the same, but for 7700K it's around 150$ more expensive than the other two. This has made it quite difficult for me to arrive on a decision.Any suggestions?

Simply wait a few days/weeks if you can. I see the biggest disadvantage of Ryzen in the not optimized microcode. I've talked with a few BIOS developers and they all were not satisfied with the current status quo. It was simply too less time to optimize all. Wait for new firmware, read mainboard tests and OC results. Take your time to wait for a stabilized price and decide it a tick later

 

[Supporter]

I lol @ everyone who bought one for gaming.

 

[DIrtyMic]

Considering the resources that Intel has versus AMD, this is an impressive showing. Intel's architecture is beyond mature, so there is little I think they can squeeze out in optimizations alone. I am sure that with firmware updates and working with software and hardware developers further performance can be had. This is a win for all consumers and will force Intel's hand both on pricing and engineering. Bravo AMD.

 

[FormatC]

I lol @ everyone who bought one for gaming.

Together with a RX480 it might be a good combo. And especially in MMO's with big maps, a lot of ppl and if you wish to stream your game, this may work perfect.

 

[DIrtyMic]

And if the lower and mid market APUs perform, Intel has something to worry about.

 

[schaft]

Can't wait to see it on market. Its just not a good idea for intel to monopolize the industry. Hope this will fire up the competition. I just not able to agree with "AMD's Ryzen 7 1700 has a 65W TDP, making it the lowest-power eight-core desktop CPU available." Intel have xeon that was 65 watt with 12 core 24 thread (the one I am using now), although it has very low clock speed and not many people know it.

Hope amd can light up the cpu competition which has been lump ever since FX-8xxx series (seem like ages to me)

 

[FormatC]

And if the lower and mid market APUs perform, Intel has something to worry about.

I'm still waiting for this new APUs. If they can combine the VGA power with a bit more IPC in the CPU part it is enough for the most simple jobs. I wrote in the past so many APU reviews, made a lot of entry-level builds... The biggest disadvantage was always the IPC and a small Intel Pentium with a cheap entry level VGA card was in combination often cheaper and not slower (mostly faster) in games. If AMD can solve this issue, it might be a good start to complete the line-up in the right way.

Intel made a big fault to cancel the Iris Pro project. Good performance, but in the wrong CPUs and combined with horrible drivers. Luck (and a big chance) for AMD.

I have no idea, what the fanboys expected. Ryzen is a server CPU (Naples), modified and simplified for the consumer market, nothing else. AMD had not the ressources to develop both things at the same time and separately. For this conditions the result is even better as expected (from my side). If AMD can fix the memory issues/bottlenecks and optimize the microcode, we will see a big progress. But the current state wasn't able to satisfy the market 100%.

The people expected simply too much. AMD can not run over water or divide the sea, but they made a very interesting server/workstation CPU that can be also used for the the consumer market. Not more, but also not less. This is, from my very private point of view, the best sign for a good re-start.

 

[nader21007]

https://img.purch.com/o/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9YLzMvNjU2OTY3L29yaWdpbmFsLzAxLVRvbTJELURpcmVjdC1Uby1EZXZpdmUucG5n

Are you Idiot Tom? or Well-paid by Intel ? What is this 3.8 Ghz invention? Since when HW Review sites use sth like this nonsense?
6900K Score is: 2456
1800X Score is : 2884
How come a 6900K @ 3.8Ghz could score: 3718 ??
more than 50% increase in the score for 18% increase in frequency?
is this a new phenomenon?
Of course Tom has called intel before painting this plot to destroy Ryzen, they paid Tom $$$ to manipulate benchmarks.
in this chart 1800X is better than the 6900K, yet Tom is showing the intel inferior CPU on top, and you ignore the reality. Look down the chart.

 

[ivanarj]

For Naples, please consider some SQL Server benchmarks.

 

[FormatC]

https://img.purch.com/o/aHR0cDovL21lZGlhLmJlc3RvZm1pY3JvLmNvbS9YLzMvNjU2OTY3L29yaWdpbmFsLzAxLVRvbTJELURpcmVjdC1Uby1EZXZpdmUucG5n

Are you Idiot Tom? or Well-paid by Intel ? What is this 3.8 Ghz invention? Since when HW Review sites use sth like this nonsense?
6900K Score is: 2456
1800X Score is : 2884
How come a 6900K @ 3.8Ghz could score: 3718 ??
more than 50% increase in the score for 18% increase in frequency?
is this a new phenomenon?
Of course Tom has called intel before painting this plot to destroy Ryzen, they paid Tom $$$ to manipulate benchmarks.
in this chart 1800X is better than the 6900K, yet Tom is showing the intel inferior CPU on top, and you ignore the reality. Look down the chart.

Your comment shows, that you haven't realized the problem behind this benchmark. Please think logical and try to understand, what happened in real. This is a "simple" 2D job and the stock i7 6900K will not reach his highest possible clock rate in this low-priority multi-threaded workloads. And BTW: you linked the wrong picture.

more than 50% increase in the score for 18% increase in frequency?

If it comes to multi-threaded workloads of 20% per core or less, the clock rate of the 6900K is simply too low (nearly as in idle!). The 7700K can use only the half of threads and is getting higher load levels per core. As follow the clock rate is significant higher and the result better. The same with the 6900K @3.8 Ghz, this CPU can run even faster now (over 50% more clock speed as stock!). If you read it right, it also shows a big disadvantage of the 6900K. 😉

This is also one of the rare situations, when a quad-channel memory makes sense. This benchmark is drawing billions of curves, lines, dots, polygons etc, is filling it, but all direct into a big, allocated memory-block. Simply troughput.

The stock Ryzen 7 1800X was using mostly the full clock speed, the stock Intel Core i7 6900K was in "sleep mode". It also shows, that AMD was able to change the states faster than Intel and also more fine-tuned for lower load levels. This benchmarks makes totally sense, but only if you are able to understand it. Your bashing is totally senseless.

If AMD had given me the necessary time, it might be better explained with frequency charts and so on. Not my fault. I'm happy that we were able to finish it as is.