• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

Page 177 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

CatMerc

Golden Member
Jul 16, 2016
1,114
1,146
136
  • Like
Reactions: Malogeek

Topweasel

Diamond Member
Oct 19, 2000
5,339
1,528
136
I would argue that most people who buy 7700k/1700 level CPUs do in fact build them primarily for gaming. People who do actually need powerful workstations, and are budget constrained, would probably be better off buying a couple generations old xeon than any high end mainstream cpu.

I don't know if you're trying to make some sort of "check your privilege" point, but you'd have to think I'm a complete moron to not know that poor people exist. Desktop PCs are a small niche and I would bet the vast majority (of the constantly dropping number) being sold today are < $500 complete systems like the one I saw wheeling by me in a cart at Walmart on Saturday. That has nothing to do with pairing a 8 core high end CPU with a mid/low end GPU, which would still be out of reach for the vast majority of humans who are not upper class in developed nations.
It depends maybe it applies to the 7700 crowd and that's why there is a huge gap here between users. Some people look at Ryzen and see a CPU the price of an 7700 and may look at what a 7700 would generally be used for. I think you would be surprised at the amount of users that look at Ryzen not as a 7700 competitor but a 6900k competitor. Some of these people got i5's or i7's in the past because the barrier for more was prohibitive. Some people like myself are on aging E series setups who want to modernize. The $300+ CPU user tends to be a person with money to burn either a 20 something without a worry in the world or a much bigger portion that is 30+ stable job, maybe like me we still play games. But our system still has to do more. Hell the number one reason I am buying a ryzen setup is because I was planning on upgrading my 3930 to a 6900, now for the cost of just upgrading my system, I can build a whole new computer. Now I have 14c28t to use at home instead of just 8c16t.

Trading 10% frames in a game for double the cores or being a third the price either way you want to look at it. Though that isn't really fair. I have always valued more cores. Early adopter of the Athlon64x2, I got a Phenom x4 near launch, and jumped on to the 3930 pretty quickly.
 

MajinCry

Platinum Member
Jul 28, 2015
2,488
557
136
Where in game to test is a fair point, you do want a worst case scenario.
What i was saying is that with diff cores, number of cores, memory controllers, res scaling is not identical between the tested CPUs so testing at low res is not representative or sufficient.
Data is hard to find as nobody tests properly the 6900k and Ryzen in games but plotting the res scaling SKU vs SKU is what reviewers should do.
Testing at low res is to eliminate as much of the GPU bottleneck as possible. Chuck in crossfire 480s (or Vega) to get a clear idea of single threaded draw call perf, and we got ourselves an optimum setup.

If you want to test the GPU, turn the resolution way up. To test the CPU, turn the resolution way down, so the GPU has as little effect as possible. We want to see how much grunt the CPU has, not how much the GPU has.
 

imported_jjj

Senior member
Feb 14, 2009
660
430
136
Testing at low res is to eliminate as much of the GPU bottleneck as possible. Chuck in crossfire 480s (or Vega) to get a clear idea of single threaded draw call perf, and we got ourselves an optimum setup.

If you want to test the GPU, turn the resolution way up. To test the CPU, turn the resolution way down, so the GPU has as little effect as possible. We want to see how much grunt the CPU has, not how much the GPU has.
You are missing my point, the CPU load profile changes with FPS (so with changes in res when using the same GPU) and when the SKUs tested are very different, that matters.
It's like the difference between acceleration and speed in cars. At low res you test acceleration , at normal res you test speed.
 
  • Like
Reactions: sirmo

hotstocks

Member
Jun 20, 2008
81
26
91
When I build a pc, I build it for gaming and to last, because gaming is where you go obsolete the fastest. If you start with a top of the line processor, you can get 5 years out of it, top of the line GPU about 2-3 years. Now I only may game 10% of the time on my PC, but when I do, I want a solid 60+ fps with all the eye candy. I don't need 120/144hz competitive play. But while I am gaming, there are a ton of other apps, programs, and threads in the background, all my daily stuff, so my guess is that Ryzen while loses to 7700 in pure gaming with nothing in the background, Ryzen can probably still keep 60+ fps while gaming and all your other programs/tasks are opened. But NOBODY is testing this real life scenario! That is what should be tested, and I would, but I am NOT building a Ryzen till motherboards are rock solid with 32gb 3200mhz ddr. I need my machine for work and gaming and can't have random crashes, reboots, lost data, or incompatibilities. So if AMD/MS and mobo/ram manufacturers get their sh!t together before Intel releases it's next 6 or 8 core cpu, they have my business, if not, oh well. This half baked early Ryzen launch was a HUGE mistake, should have waited and fixed mobo/ram/windows issues and actually had mobos that work and don't brick themselves or bootloop. So everyone for the last two weeks has been beta testers, that is not how a launch should be. Everyone excited, then pulling out their hair and returning parts because of issues and inconsistencies.
 

innociv

Member
Jun 7, 2011
54
20
76
Ok.. the forum won't let me do a quote. But in regards to Mafia III and it seemingly already being Ryzen optimized:
It's probably one of the very few that does its own thread scheduling instead of relying on the OS scheduler.

And as for why Ryzen suddenly does better in games it seemed CPU bound in like Total Warhammer and Rise of the Tombraider:
The Nvidia driver is multithreaded. It almost certainly does not understand the cache latency timings from crossing CCX yet thus Nvidia GPUs may very well be running worse on Ryzen no matter what game is used. Radeon drivers may be better optimized for Ryzen already.
Still, those are GPU bottlenecked tests... would be better to use R9 Fury in crossfire.

It's pretty insane that no one thought to test Ryzen on an RX480 or R9 Fury at 480p/720p low settings to remove/lessen the GPU bottleneck.
But "professional hardware reviewers". lol...
 
Last edited:

MajinCry

Platinum Member
Jul 28, 2015
2,488
557
136
Ok.. the forum won't let me do a quote. But in regards to Mafia III and it seemingly already being Ryzen optimized:
It's probably one of the very few that does its own thread scheduling instead of relying on the OS scheduler.

And as for why Ryzen suddenly does better in games it seemed CPU bound in like Total Warhammer and Rise of the Tombraider:
The Nvidia driver is multithreaded. It almost certainly does not understand the cache latency timings from crossing CCX yet thus Nvidia GPUs may very well be running worse on Ryzen no matter what game is used. Radeon drivers may be better optimized for Ryzen already.
Still, those are GPU bottlenecked tests... would be better to use R9 Fury in crossfire.
But with the draw call benchmark results posted by roybotnik (the youtube videos), we saw that AMD's graphics driver loses a large amount of performance, when the driver thread is on a different CCX from the main thread.
 

bystander36

Diamond Member
Apr 1, 2013
5,154
132
106
The vast majority of 8 cores Ryzen buyers would spend little on a GPU as the PC is not a gaming console and a 200$ GPU is sufficient for 1080p and can be stretched to 1440p with lower settings in the most demanding games.
You seem to think that "people in their right mind" build PCs just for gaming so i am guessing you are part of the smartphone generation that even does video editing on their phones.Not sure who's in their right mind and who isn't but lets not try to decide that one now.
If you build a PC just for gaming, get a quad core Ryzen with HT, OC it and spend more on the GPU as it's the best way to get the best gaming perf for the money. And in a few years upgrade the CPU without spending on a mobo.

The CPU shouldn't even be a factor in gaming (at all) outside super budget gaming boxes. It got this bad because Intel has refused to offer more cores but this has nothing to do with normality. Hopefully AMD goes 12 cores in 2019 and you'll start to understand that.

To be clear, the PC can do more than just gaming. Very few folks build PCs for just gaming and they are usually upper class in the most developed nations but the world is much bigger than that.
These CPU reviews are doing people a disservice. I'm sorry to be the one to inform you, but lots of games are CPU limited in spots, and quite a few more are CPU limited a lot. Single player games tend to be GPU limited most the time, but with quite a few spots that are CPU bound. That's why you see fair low minimums a lot, and the worst part, is they rarely test the bad spots of a game, where a fast CPU matters.

If you are after a minimum of 30 FPS, you are good to go on anything, but I imagine most people jumping into CPU gaming want at least 60 FPS and that's when a fast CPU does matter, and it matters even more past that point.
 
Last edited:

DuronBurgerMan

Junior Member
Mar 13, 2017
21
19
41
It's fascinating going through all these benchmarks. Ryzen occasionally dominates, or at least makes a good showing, depending on the game... yet it remains slow on the majority of games. Either this CPU has a split personality, or we're seeing a good architecture getting hamstrung by either scheduling, software optimization, or just early release issues (or any combination of the three).

At least it's not Bulldozer, which just sucked at almost everything.
 

Malogeek

Golden Member
Mar 5, 2017
1,390
778
106
yaktribe.org
Ok.. the forum won't let me do a quote.
The multi-quote system is broken however you can click Reply on a post and it will add that entire post as quote in the reply. You can also highlight a specific portion of the text and you get an option to reply to that particular portion, which is added to the reply.
 

CatMerc

Golden Member
Jul 16, 2016
1,114
1,146
136
Ok.. the forum won't let me do a quote. But in regards to Mafia III and it seemingly already being Ryzen optimized:
It's probably one of the very few that does its own thread scheduling instead of relying on the OS scheduler.

And as for why Ryzen suddenly does better in games it seemed CPU bound in like Total Warhammer and Rise of the Tombraider:
The Nvidia driver is multithreaded. It almost certainly does not understand the cache latency timings from crossing CCX yet thus Nvidia GPUs may very well be running worse on Ryzen no matter what game is used. Radeon drivers may be better optimized for Ryzen already.
Still, those are GPU bottlenecked tests... would be better to use R9 Fury in crossfire.

It's pretty insane that no one thought to test Ryzen on an RX480 or R9 Fury at 480p/720p low settings to remove/lessen the GPU bottleneck.
But "professional hardware reviewers". lol...
;)

My next article is going to be interesting if the next few tests match up with my theory.
So far it looks like I've stumbled on something very very interesting.
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
13,984
3,682
136
;)

My next article is going to be interesting if the next few tests match up with my theory.
So far it looks like I've stumbled on something very very interesting.
I actually have both R9 Fury and RX 480 cards + Ryzen 1800X. On latest non-insider Windows 10 build. Anything you want me to test? I can do up to triple Crossfire RX 480.
 

CatMerc

Golden Member
Jul 16, 2016
1,114
1,146
136
I actually have both R9 Fury and RX 480 cards + Ryzen 1800X. On latest non-insider Windows 10 build. Anything you want me to test? I can do up to triple Crossfire RX 480.
Unless you have a 6900K too, not really no.
 

Crumpet

Senior member
Jan 15, 2017
745
538
96
So.. spent 4 hours doing some video editing today.

i5 6600 ground to an absolute stop 3 times. Glad i'm a chronic saver. Really wish my motherboard would arrive.......
 

innociv

Member
Jun 7, 2011
54
20
76
Windows7 scheduler does not really work any better. Performance increases in Windows7 over 10 are likely for the same reason some games run better in Win7 for Kabylake than they do in win10.

Windows7 does not understand the architecture any better. It does not better attempt to keep an application running on one CCX. I'm not sure how people get this idea that a game's performance on an OS is 100% down to the scheduler and that must be why Windows7 gives better performance in some games...
 

Rifter

Lifer
Oct 9, 1999
11,518
745
126
https://community.amd.com/community/gaming/blog/2017/03/13/amd-ryzen-community-update?sf62107357=1

Robert Hallock claims AMD does not find any issues with W10 scheduler? I have yet to see core usage % in W7 task manager but going by the above (if W10 has no scheduling issues) then both OS versions should be the same for which cores are loaded in apps?
Thats wierd, ive seen many pictures that clearly show the scheduler is a problem.

This is probably MS refusing to fix it and AMD trying to keep MS happy and going along with it.

Which is really dumb IMO.
 
  • Like
Reactions: beginner99

DuronBurgerMan

Junior Member
Mar 13, 2017
21
19
41
Thats wierd, ive seen many pictures that clearly show the scheduler is a problem.

This is probably MS refusing to fix it and AMD trying to keep MS happy and going along with it.

Which is really dumb IMO.
Not necessarily. They may simply mean that Windows is properly differentiating physical cores from logical cores, which is true. The Windows 10 scheduler IS doing that right. There's nothing broken with it. That being said, it isn't optimized for Ryzen, because of the two CCXs acting, in some ways, almost like separate chips. So AMD comes out and says, basically, "calm down, people, Microsoft didn't do anything wrong." Which is true, and builds some goodwill for AMD... so that hopefully they can convince MS to optimize more for Ryzen. In other words there's a difference between being broken and being unoptimized.

After all, you don't build goodwill by blaming Microsoft for everything -- it was AMD that decided this was the way they wanted to do it knowing full well that Windows wasn't currently optimized for this scenario. They had to know this would be an issue. Give 'em some time.
 
  • Like
Reactions: DeeJayBump

Dresdenboy

Golden Member
Jul 28, 2003
1,730
554
136
citavia.blog.de
So while Zen should easily be able to hit 1000, why doesnt it? The horsepower is surely there.. what is holding back 5%?
I'll give it a shot: Timer resolution.

For some new results: HT4U did some benchmark runs at different mem speeds and also with 2 cores disabled (simulating the 1600X).
https://www.ht4u.net/reviews/2017/amd_ryzen_7_1800x_im_test/

1800X DDR4-2133 vs. DDR4-3200
Many games run ~15% faster! Now imagine switching off SMT and fixing any thread/CCX ping pong in games.

https://www.ht4u.net/reviews/2017/amd_ryzen_7_1800x_im_test/
 
Last edited:

innociv

Member
Jun 7, 2011
54
20
76
Some of those games like Witcher 3 are known to run much faster with faster RAM. Fallout 4 is another one.
It's just bad programming that leaves them hitting memory a lot.

No VIDEO GAME should be RAM speed limited like that. It take poor programming to make it happen.
 

ASK THE COMMUNITY