Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

Page 130 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

agouraki

Member
Feb 18, 2017
26
15
51
im dissapointed with gamer nexus....

taking into consideration the raw power of ryzen,the inconsistensy of benchmarks across games and the SMT "bug" that they themselves noticed it...

you cant just put a video paiting them as I5 of gaming... heck they even pass i5 already on games....

i bet even intel fanbois thought that title was bullshit...

i agree with linus the potential is there and we will see it in the next months hopefully.
 
  • Like
Reactions: clocks11

unseenmorbidity

Golden Member
Nov 27, 2016
1,395
967
96
im dissapointed with gamer nexus....

taking into consideration the raw power of ryzen,the inconsistensy of benchmarks across games and the SMT "bug" that they themselves noticed it...

you cant just put a video paiting them as I5 of gaming... heck they even pass i5 already on games....

i bet even intel fanbois thought that title was bullshit...

i agree with linus the potential is there and we will see it in the next months hopefully.
He seemed to clearly be pissed off by amd. You can tell by his review and especially that follow up video.
 

looncraz

Senior member
Sep 12, 2011
722
1,651
136
im dissapointed from gamer nexus taking into consideration the raw power of ryzen,the inconsistensy of benchmarks across games and the SMT "bug"

you cant just put a video paiting them as I5 of gaming... heck they even pass i5 already on games....

i bet even intel fanbois thought that title was bullshit...

I think they are just relaying what they're seeing. And they aren't alone, many reviewers are seeing the same. They are, however, painting AMD's responses in a very much more aggressive light than what AMD is saying.

I mean, seriously, if you're trying to compare a $500 CPU to a $1,000 CPU you usually give the $500 some benefit of the doubt by leveling non-CPU factors, such as memory bandwidth. You disclose this, of course, but you at least test with the field being level.

And when you're testing games you don't ONLY test at low resolutions - you make a series of tests at various resolutions so people can get a feel for where the bottleneck resides.

AND, critically, when you have one area where the results are just flat-out confusing and out of line with the rest of your tests... you make a point that your results may be as a result of something not inherent with the CPU that may be resolved in due time.

We've already seen numerous aspects of Ryzen that cause game performance, in particular, to plummet (to the levels of a high end i5 or middle i7... but not in all cases). SMT, CCX bandwidth, BIOS issues, software confusion... I mean, seriously, we are seeing nVidia drivers filling up VRAM on Ryzen... that's not the CPU doing that, that's nVdia's drivers. Sabotage? Maybe... but it could just be that the driver is confused by the topology and is assuming it doesn't have the bandwidth to stream data into the VRAM...

What happened to the true investigative reviews of old?
 

agouraki

Member
Feb 18, 2017
26
15
51
And when you're testing games you don't ONLY test at low resolutions - you make a series of tests at various resolutions so people can get a feel for where the bottleneck resides.

AND, critically, when you have one area where the results are just flat-out confusing and out of line with the rest of your tests... you make a point that your results may be as a result of something not inherent with the CPU that may be resolved in due time.


What happened to the true investigative reviews of old?
i couldnt say it better myself... the evidence all point out that any reviews atm are to be taken with a grain of salt if you expect to buy a cpu that will last you 4 years+.

also where are the reviews? people kept telling us wait for reviews wait for reviews but i see more french/german in depth reviews than english.
 

AtenRa

Lifer
Feb 2, 2009
14,000
3,357
136
TDP ain't power usage. We know you like AMD but please keep it real.

Please read what i said and dont make it personal , I know very well what TDP means.
I have explained earlier that lower TDP have a big impact in TCO due to cheaper cooling infrastructure and lower energy consumption of your server room cooling solution (you dont need lower ambient temps etc etc).

DataCenterCooling.jpg


TraditionalDCenv.jpg




Yes, Ryzen at lower clocks seems to have great performance/dollar. But Intel server chips, eg xeons also run at lower clocks, at least some models. Intels TDP is so high due to AVX2 workloads. However as we have seen actually power use without AVX2 is pretty similar between the 2 chips.

Power use is not equal between the two without AVX, Ryzen has significantly lower power than 6900K in many workloads. Example bellow, in HandBrake the R7 1800X almost has the same performance vs Core i7 6900K but at lower power.

R7 1800X ----> 56,10fps using 174W and perf/watt = 0,3224
Core i7 6900K ---> 57,68fps using 204W and perf/watt = 0,2827


http://www.legitreviews.com/amd-ryzen-7-1800x-1700x-and-1700-processor-review_191753/12

handbrake-fast-ryzen-1800x.jpg


ryzen-power-chart.jpg




Still, yes I agree it looks great for Naples in this regard, but Naples isn't out yet and Skylake-X will be soonish. 14nm+ from Intel isn't bad at all for server/mobile use. Yes 7700k semi-rebrand wasn't great but laptop chips had pretty huge increases due to the new process.

Naples is coming Q2 2017, Skylake-X is coming Q3 2017.

Price of CPU however is somewhat irrelevant for TCO-cost of a server. Matters for us yes, but not much for server. I rather have AMD charge more for server chips and make more profit for future R&D. Also don't forget to count software cost to TCO. If you run software that costs you several 100k a year, even 5% better performance matters more even if the chips cost $1000 more.

For big datacenters and Cloud Server where your CPU count is in the thousands, the cost of the CPUs and cost of cooling infrastructure takes a significant portion of the CapEx.
 

french toast

Senior member
Feb 22, 2017
988
825
136
He seemed to clearly be pissed off by amd. You can tell by his review and especially that follow up video.
This is what happens when you try to influence reviewers= bad sentiment.
Amd should have either got the bios sorted sooner or most likely delayed the review.

Not do what they did, idiots.
 

piesquared

Golden Member
Oct 16, 2006
1,651
473
136

That reddit thread seems quite credible.
This quote is very interesting.
Default is actually: Balanced. Windows manages Ryzen's power/clocks, not Ryzen's SenseMI

It all makes sense given how well Ryzen performs. It cant just start underperforming when gaming. There is no specialized part of the die specifically for the GPU it uses the core like any other application. If Ryzen is strong everywhere else it should be in games too. Perhaps the link to the GPU may not be running at full speed with balanced power profile.
 
Last edited:

dfk7677

Member
Sep 6, 2007
64
21
81
I'm seeing the 8C16T Ryzen 1800X on par with 4C4T i5 in games. For 6600K/7600K vs 1800X we are talking similar clocks but 4x the threads for the Ryzen. How multithreaded are teh games? I point to the gaming benchmarks and let you decide.

Will an i5 be obsolete for gaming purposes in 2018? Again, I point to the gaming benchmarks that compare w/ i5 and let you decide.

http://www.gamersnexus.net/hwreview...review-premiere-blender-fps-benchmarks/page-7
http://www.pcgamer.com/the-amd-ryzen-7-review/5/

It's my opinion that I'll game with perfectly enjoyable settings, even in 2018, on my lil lowly 4T i5's. And I'm perfectly happy if people disagree with me. I'm not stopping anyone from buying CPUs. I love CPUs too and I understanding wanting the "BEST" performance. Most users tho, most gamers, will probably still survive in 2018 on i5's. Not all users are die-hard enthusiasts that get pissed and drop $$$ if they get an FPS spike or have to dial a quality setting down. That's the .1% here at AT. We are an odd bunch lol but I love it here ;).

But would you recommend an i5-7600K or an 1600X (3.6/4.0) to someone that wants to upgrade without spending a lot of money on a processor?
 

Zucker2k

Golden Member
Feb 15, 2006
1,810
1,159
136
I am afraid those are system-wide wattage with a GPU and we cannot tell how much of it is GPU and how much of it is CPU. :)/QUOTE]

Yes, system wide, as is all the other tests. Believe me, A GTX 1080, working 50% harder at 1080p isn't doing work for free. You can estimate how much power the gpu is consuming by looking at the other tests. The R7 1700 is consuming about 14% less power but losing by 50%.
 

moonbogg

Lifer
Jan 8, 2011
10,635
3,095
136
Curious to see some OC results with SMT disabled. I wonder if it will clock any higher without SMT and what the combined performance benefits may be for gaming. Some reviewers showed OC results, and results without SMT, but not the two combined. It was one or the other.
 

tamz_msc

Diamond Member
Jan 5, 2017
3,710
3,554
136
I think they are just relaying what they're seeing. And they aren't alone, many reviewers are seeing the same. They are, however, painting AMD's responses in a very much more aggressive light than what AMD is saying.

I mean, seriously, if you're trying to compare a $500 CPU to a $1,000 CPU you usually give the $500 some benefit of the doubt by leveling non-CPU factors, such as memory bandwidth. You disclose this, of course, but you at least test with the field being level.

And when you're testing games you don't ONLY test at low resolutions - you make a series of tests at various resolutions so people can get a feel for where the bottleneck resides.

AND, critically, when you have one area where the results are just flat-out confusing and out of line with the rest of your tests... you make a point that your results may be as a result of something not inherent with the CPU that may be resolved in due time.

We've already seen numerous aspects of Ryzen that cause game performance, in particular, to plummet (to the levels of a high end i5 or middle i7... but not in all cases). SMT, CCX bandwidth, BIOS issues, software confusion... I mean, seriously, we are seeing nVidia drivers filling up VRAM on Ryzen... that's not the CPU doing that, that's nVdia's drivers. Sabotage? Maybe... but it could just be that the driver is confused by the topology and is assuming it doesn't have the bandwidth to stream data into the VRAM...

What happened to the true investigative reviews of old?
I actually don't have any issues with their data - it's where I expected it to be in relation to other reviews which were done on the ASUS board.

However, it's their conclusions that they wrote in their article on the website that I have a problem with. It's pretty egregious, to say the least.

I mean entire paragraphs dedicated to how the Sniper Elite 4 demo was 'looking at the sky', skirting the need to include higher resolutions because they are GPU bound, how Cinebench is well 'meh' in real world etc. - all in relation to the AMD press event stuff. You got your own data - how about talking about that?

What was really bad was when they said that you'd probably want to use GPU-acceleration for workloads that lend themselves to that route - something like how a GTX 1080 is three times faster in rendering. Are they being serious? What is the point in bringing that up in a CPU review? Of course, everybody knows, or ought to know that. Why bring this up?

They condensed all the non-gaming performance metrics in a sentence or two by saying that the 1800X is competitive with the 6900K, and then finally went on to conclude that it is an i5 in gaming. That title was the second most clickbait headline I've seen, the first one being 'good, but not for gaming' from ArsTechnica.

On top of that, they say that they won't change their conclusions, even after there is a growing knowledge of the kind of issues that people are facing.
 

looncraz

Senior member
Sep 12, 2011
722
1,651
136
This is what happens when you try to influence reviewers= bad sentiment.
Amd should have either got the bios sorted sooner or most likely delayed the review.

Not do what they did, idiots.

They didn't really try to influence them, they made suggestions about certain configurations they should test. Intel does/did the same thing - same as nVidia.

You can actually hear the AMD guy make his suggestions and then see the way it is wrongly interpreted.
 

looncraz

Senior member
Sep 12, 2011
722
1,651
136
But would you recommend an i5-7600K or an 1600X (3.6/4.0) to someone that wants to upgrade without spending a lot of money on a processor?

What you'd recommend would depend on the use case.

Even for gaming the answer, at this time, isn't particularly clear.

The 1600X will run every game just fine - some better than others... some much better than the i5... some a little worse... some way worse.

What games are being played, whats games are intended to be played in the future, how often video cards are upgraded - and at what level... is the monitor going to be upgraded as well?

Someone who is buying a mid-range CPU isn't likely to be buying high end graphics cards and monitors.
 

looncraz

Senior member
Sep 12, 2011
722
1,651
136
I actually don't have any issues with their data - it's where I expected it to be in relation to other reviews which were done on the ASUS board.

However, it's their conclusions that they wrote in their article on the website that I have a problem with. It's pretty egregious, to say the least.

I mean entire paragraphs dedicated to how the Sniper Elite 4 demo was 'looking at the sky', skirting the need to include higher resolutions because they are GPU bound, how Cinebench is well 'meh' in real world etc. - all in relation to the AMD press event stuff. You got your own data - how about talking about that?

What was really bad was when they said that you'd probably want to use GPU-acceleration for workloads that lend themselves to that route - something like how a GTX 1080 is three times faster in rendering. Are they being serious? What is the point in bringing that up in a CPU review? Of course, everybody knows, or ought to know that. Why bring this up?

They condensed all the non-gaming performance metrics in a sentence or two by saying that the 1800X is competitive with the 6900K, and then finally went on to conclude that it is an i5 in gaming. That title was the second most clickbait headline I've seen, the first one being 'good, but not for gaming' from ArsTechnica.

On top of that, they say that they won't change their conclusions, even after there is a growing knowledge of the kind of issues that people are facing.

Yes, there is no doubt that they went heavy on the negative side and failed to be even-handed.

But, they are called Gamer's Nexus... Ryzen 7 just isn't a hardcore gaming CPU. If you look at the world through that narrow prism (and only ~20% of the HEDT market...) then a CPU like the 6900k or 1800X will not be appealing.

Still, we have numerous reviews showing gaming results actually being in-line with the rest of the chip's performance. They all have a few things in common: they are running updated BIOS, fully updated Windows 10, focusing on averages more than maximums (who really cares if you can get 30% higher maximum when your average is only 5% different?), are configuring Windows power settings properly, and aren't using Asus motherboards (not to dig at Asus, I have full faith that they will work diligently to resolve the outstanding issues and deliver a fantastic product - which is why my C6H order still stands :p).
 

Joric

Junior Member
Mar 4, 2017
14
6
16
The Ryzen 1700 looks like a great value chip.
The only result I've seen that bothers me is the Watchdogs 2 results.
I wouldn't expect Ryzen to be able to make up the IPC and clock speed deficit to the 7700k on minimally multithreaded games/applications.
Watchdogs 2 however is multithreaded to the point where the 6950X shows a significant performance lead if the GPU isn't bottlenecking things.
Yet the Ryzen processors (while still out performing the 7700K) are losing out significantly to even the 6800K, let alone the 6900K.
This makes me wonder if the SMT implementation has some serious flaws.
The idea of having to check individual applications to determine whether they are helped, hindered or unaffected by SMT is very unappealing.
 
  • Like
Reactions: moonbogg

Puffnstuff

Lifer
Mar 9, 2005
16,015
4,785
136
I just read a review over at PC Gamer on the Ryzen and they do a great job of explaining their mixed results. They also used the Asus Crosshair motherboard but noted that with a bios update performance jumped 10%, however they also noted that Gigabyte boards seemed to be returning better results. They are planning to post their findings using other boards soon so we can get an objective comparison of multiple boards on the same benchmarks with the same cpu's which will remove some of the variables present in other peoples testing.

They do note that application performance is very good on Ryzen which would bode well for productivity work. This is in line with some of the other application benchmarks we've seen and I feel like AMD would've been better served by leveraging this to promote sales rather than using gaming which hasn't worked out too well thus far.

AMD's marketing should've done a better job of positioning the chips prior to launch because too much emphasis was placed on gaming which has promoted high expectations in this area that were not actually achieved leading to a loss of confidence. There's no doubt that if a company were looking for new pc's for productivity specific work that the Ryzen would do a great job, especially crunching spreadsheets, so there's a market to be had and they should focus on its true strengths.
 

R0H1T

Platinum Member
Jan 12, 2013
2,582
162
106
The Ryzen 1700 looks like a great value chip.
The only result I've seen that bothers me is the Watchdogs 2 results.
I wouldn't expect Ryzen to be able to make up the IPC and clock speed deficit to the 7700k on minimally multithreaded games/applications.
Watchdogs 2 however is multithreaded to the point where the 6950X shows a significant performance lead if the GPU isn't bottlenecking things.
Yet the Ryzen processors (while still out performing the 7700K) are losing out significantly to even the 6800K, let alone the 6900K.
This makes me wonder if the SMT implementation has some serious flaws.
The idea of having to check individual applications to determine whether they are helped, hindered or unaffected by SMT is very unappealing.
Only certain games have shown negative bias so far as dealing with SMT is concerned, nearly every other application, whether lightly threaded or not, has shown gains from AMD's version of SMT better than even Intel's HT which btw is over five gen old.
The games themselves could be better equipped, with patches, to handle SMT & I believe there'd be a dedicated patch for Windows (OS) as well.
 
Last edited:
  • Like
Reactions: looncraz

french toast

Senior member
Feb 22, 2017
988
825
136
If was to list the best gaming cpus right now, focusing on perf/$ it would look something like this;
1, Intel pentium G4560. 70$
2, Amd R7 1700 329$
3, Intel core i7 6700k 260$
4, Intel i5 7600k 240$
5, Intel i7 7700k 349 $
6, Intel i7 6800k 420$?
7, Amd R7 1700x 399$

The fastest out of all of them for 2017/18 probably the 6800k, with the R7 1700x further out.
Of course this asumes bios updates and windows patch.

In 2 months with the introduction of R5 1600x and 1400x as well as game updates i could envisage that order being shook up, 6 months out from that with raven ridge on 14 nm LPU + vega igpu, well intel is going to have a problem.
 

Dygaza

Member
Oct 16, 2015
176
34
101
What does it actually mean in real use when windows sees caches wrongly (as seen from coreinfo)? Someone with more technical knowledge could open this up a bit.
 

MongGrel

Lifer
Dec 3, 2013
38,751
3,068
121
I'll still probably be using my cheap X5650/X5680 rigs for myself I imagine.

I used to use ThunderBirds about two decades ago, last time I had tried an AMD chip out to be honest.