Question Should i upgrade? or is my current cpu not set up right?

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
I currently have a ryzen 2600 @ 4.2 ghz with 3600mhz ram in dual channel. It bottlenecks my 980ti quite a bit. i recently played darksiders 3 and my gpu was running @ 70% max during gameplay @1080p ultra no vsync.
Does anyone have experience with this issue? according to all the benchmarks out there my cpu should barely bottleneck this gpu and when it does it should be between 5-10%. if i cant find a way to resolve this then i'll probably buy an 8600k and a z390 mobo soon. since switching to AMD i always seem to be waiting for that cpu with better ipc and higher clocks.
 
4.2 is a decent overclock for any ryzen.

Some games are cpu limited.
Not so much with a lack of threads, but with the performance of the single master thread.

You might run a couple of experiments.

1. REDUCE your overclock to about 36, a 15% reduction in performance.
Run your games.
That will give you a clue as to how sensitive your particular game is to cpu core speed.
Or, conversely what a 9600K would do running 15% faster at near 5.0.

2.
Experiment with removing one or more cores/threads. You can do this in the windows msconfig boot advanced options option.
You will need to reboot for the change to take effect. Set the number of threads to less than you have.
This will tell you how sensitive your games are to the benefits of many threads.
If you see little difference, your game does not need all the threads you have.
That will tell you how important 6/8/16 threads are to your games.


If you buy a Z390 motherboard, buy one of the 9th gen K processors.
The i5-9600K is a better buy than a i5-8600K. They both have 6 cores, but the 9th gen K processors have soldered heat spreaders.
They run cooler and overclock better than the 8th gen counterparts.
Likely to the 5.0 level.
 

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
https://cpu.userbenchmark.com/Compare/Intel-Core-i5-9600K-vs-Intel-Core-i5-8600K/4031vs3941
4.2 is a decent overclock for any ryzen.

Some games are cpu limited.
Not so much with a lack of threads, but with the performance of the single master thread.

You might run a couple of experiments.

1. REDUCE your overclock to about 36, a 15% reduction in performance.
Run your games.
That will give you a clue as to how sensitive your particular game is to cpu core speed.
Or, conversely what a 9600K would do running 15% faster at near 5.0.

2.
Experiment with removing one or more cores/threads. You can do this in the windows msconfig boot advanced options option.
You will need to reboot for the change to take effect. Set the number of threads to less than you have.
This will tell you how sensitive your games are to the benefits of many threads.
If you see little difference, your game does not need all the threads you have.
That will tell you how important 6/8/16 threads are to your games.


If you buy a Z390 motherboard, buy one of the 9th gen K processors.
The i5-9600K is a better buy than a i5-8600K. They both have 6 cores, but the 9th gen K processors have soldered heat spreaders.
They run cooler and overclock better than the 8th gen counterparts.
Likely to the 5.0 level.

i tried all that and the difference was negligable. using re 2 on high settings with vsync off both test showed gpu usage at 80%. to clarify, i game at 1080p on a 144hz monitor AAA games so high single thread performance is key. I know the cpu ceases to be a bottleneck at 4k but a monitor and gpu upgrade is out of the question atm as im saving hard for a new car.

i did a quick google search on cpu test methodology and then tested it. I dropped the resolution to 720p all lowest settings no vsync and my gpu usage dropped to 60% on RE2. thats a cpu bottleneck right? Is there a cpu heavy game that is well optimised i could use for testing?

i took you're advice and checked out the 9600k but the performance increase compared to the 8600k is tiny. the 9600k oc'z to 5.1, the 8600k oc'z to 5.0. the single core score difference is 5. theres a benchamark below illustrating how close these cpu's are in performance.

https://cpu.userbenchmark.com/Compare/Intel-Core-i5-9600K-vs-Intel-Core-i5-8600K/4031vs3941

what would be the best option for high single core performance and be good enough for about 3 years? i've got roughly £300-£400 that i could use for a platform change.

i'm leaning towards the 8600k because the mobo's are cheaper and there's currently a sale on this cpu near me. are ther any options i should consider? maybe an older used cpu? thanks in advance
 
Last edited:
as of 3/22/2018
What % of I5-8600k chips can oc
at a aggressive vcore of 1.4 or so and delidded
4.9 96%
5.0 85%
5.1 66%
5.2 38%
5.3 13%

Note the delidded caveat.
That is perhaps a $50 service.

I do not have similar statistics for the 9600K but I assume it would be similar with the exception that delidding is not necessary to get those clock rates.

Intel ram prefetch is very good.
Intel runs reasonably well on even single channel if you are using a discrete graphics card.
This report shows very little difference between single channel and dual channel when using a discrete graphics card.
I doubt that 4 channel is a meaningful help.
http://www.silentpcreview.com/article1478-page6.html
 
  • Like
Reactions: daylightriot

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
as of 3/22/2018
What % of I5-8600k chips can oc
at a aggressive vcore of 1.4 or so and delidded
4.9 96%
5.0 85%
5.1 66%
5.2 38%
5.3 13%

Note the delidded caveat.
That is perhaps a $50 service.

I do not have similar statistics for the 9600K but I assume it would be similar with the exception that delidding is not necessary to get those clock rates.

Intel ram prefetch is very good.
Intel runs reasonably well on even single channel if you are using a discrete graphics card.
This report shows very little difference between single channel and dual channel when using a discrete graphics card.
I doubt that 4 channel is a meaningful help.
http://www.silentpcreview.com/article1478-page6.html

Do you have similar numbers for air cooling. no delid? i'd be happy if 4.8ghz is within reach. thanks in advance. Also, have you seen any gpu bottlenecks with your 1080ti?
 
I currently have a ryzen 2600 @ 4.2 ghz with 3600mhz ram in dual channel. It bottlenecks my 980ti quite a bit. i recently played darksiders 3 and my gpu was running @ 70% max during gameplay @1080p ultra no vsync.
Does anyone have experience with this issue? according to all the benchmarks out there my cpu should barely bottleneck this gpu and when it does it should be between 5-10%. if i cant find a way to resolve this then i'll probably buy an 8600k and a z390 mobo soon. since switching to AMD i always seem to be waiting for that cpu with better ipc and higher clocks.

The easiest way to check for a CPU bottleneck is to just use Ryzen Master or in your Bios, change the CPU clock speed. Change it down to say 3.6ghz instead of the 4.2ghz you have now. You'll see a linear corresponding drop in FPS if it's cpu bottlenecked. If it's not cpu bottlenecked, you won't see much of an fps drop.

Perhaps you should hold off for a couple months, Ryzen 3000 series will be here around June. It's only 2-3 months away and a simple plop in upgrade.

But I think that Darksiders game is poorly optimized to begin with, even for Intel CPU's. Here's a guy with an i7 6700k and 1080ti which only gets it to maintain 60fps, which is pathetic for a game that looks that dated. Your Ryzen chip will probably be very comparable to Intel's 6th gen chips. https://steamcommunity.com/app/606280/discussions/0/1742227264193433282/?l=hungarian
 
Last edited:
  • Like
Reactions: daylightriot
Do you have similar numbers for air cooling. no delid? i'd be happy if 4.8ghz is within reach. thanks in advance. Also, have you seen any gpu bottlenecks with your 1080ti?
The limit to any overclock will be either the Vcore you can tolerate or the temperature that approaches throttling.
Normally, the maximum vcore is the limiter.
You just do not want to push your chip past a safe voltage.
1.4 seems to be the max.
A 8600K with a overclock may reach the vcore limit before the thermal limit.
A well binned chip will get to a higher multiplier at the vcore max.
In my case, I paid a premium to silicon lottery for a well binned 8600K.
As part of the deal, the chip was professionally delidded with a warranty.
I do not run at the max capability, and I never see the temperatures go above 65c unless stress testing.

I have been monitoring my activity using GPU-Z.
I see no indication that my GTX1080ti is having any troubles.
Of course, my monitor is only capable of 60HZ and my games are primarily of the strategy type.
Civilization 4/5/6 mostly.
Whenever I run fraps, I see 60fps steady.

If your games are fast action, your results will differ.
Also, I never see a high processor utilization.
Perhaps 35% on all cores.
That suggests to me that I have two active threads and windows is spreading out their activity across all available threads.
Only once have I seen all 6 threads at 100%
That is when running FSX.
 
  • Like
Reactions: daylightriot

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
The easiest way to check for a CPU bottleneck is to just use Ryzen Master or in your Bios, change the CPU clock speed. Change it down to say 3.6ghz instead of the 4.2ghz you have now. You'll see a linear corresponding drop in FPS if it's cpu bottlenecked. If it's not cpu bottlenecked, you won't see much of an fps drop.

Perhaps you should hold off for a couple months, Ryzen 3000 series will be here around June. It's only 2-3 months away and a simple plop in upgrade.

But I think that Darksiders game is poorly optimized to begin with, even for Intel CPU's. Here's a guy with an i7 6700k and 1080ti which only gets it to maintain 60fps, which is pathetic for a game that looks that dated. Your Ryzen chip will probably be very comparable to Intel's 6th gen chips. https://steamcommunity.com/app/606280/discussions/0/1742227264193433282/?l=hungarian

I know you're talking sense when you advise to wait but i've been doing that a long time.

I've tried limiting the cpu speed as well as a battery of other tests on different games and it all points to a cpu bottleneck. The only thing i haven't looked into is thermal throttling. my cpu runs at about 50c while gaming so I assume this isnt the problem. my 980ti is liquid metal cooled and runs at similar temps.

I originally switched from a 6700k to ryzen 1st gen. huge downgrade on IPC. 2nd gen ryzen saw a 10% IPC increase and was still not comparable to my 6700k. bear in mind the 6700k is about 15-20% better in ipc oc'ed vs an oc'ed r5 2600,

https://cpu.userbenchmark.com/Compare/Intel-Core-i7-6700K-vs-AMD-Ryzen-5-2600/3502vs3955

theres an ever growing list of games in my steam collection that are being gimped by this cpu. RE2, The witcher 3, Assassins creed oddyssey to name a few.
I feel that unless AMD can gain 20-30% increase in IPC with their new chip's then i'll end up moving back over to intel.

It looks to me like amd is more focused on power efficiency and core count than single thread performance. At a guess I'd say this is because all their CPU's come from the same die as their server cpu's. This is good for them from a cost perspective but not great for gamers.

I'll take your advice and wait a few months but i doubt zen 7nm is going to break their current pattern of targeting the server market. It would be a shame to start paying intel's bullshit prices again for a cpu that can keep up with current hardware.
 
I just got done play Resident Evil 2 with my Vega 64 and Ryzen 1600. I wasn't cpu bottlenecked in that game. I was certainly GPU bottlenecked because my Vega 64 was screaming like a banshee the whole time, maxed at the whole time (Except for cut scenes, which are artificially limited to 30fps). Good thing I wear headphones. I monitor cpu usage with NZXT CAM software on my other screen while playing the game. I do have some of the settings turned down based on how I like the look, but then I render at 130% higher than 1440p, then downscale to 1440p to make everything look rich and better anti-aliased. I don't like the TAA or SMAA implementation of anti-aliasing in that game because they look more like blur filters. So I think I used FXAA and render higher then downscale to make it look better. So I'm definitely pushing my GPU to it's knees, but not CPU bottlenecked at all. Perhaps it's different if you're trying to get ultra high frame rates at 1080p in a game. For single player games, I try to max out the quality and smoothness of the game and keep it around 60-80fps for best immersion experience possible.
 

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
I just got done play Resident Evil 2 with my Vega 64 and Ryzen 1600. I wasn't cpu bottlenecked in that game. I was certainly GPU bottlenecked because my Vega 64 was screaming like a banshee the whole time, maxed at the whole time (Except for cut scenes, which are artificially limited to 30fps). Good thing I wear headphones. I monitor cpu usage with NZXT CAM software on my other screen while playing the game. I do have some of the settings turned down based on how I like the look, but then I render at 130% higher than 1440p, then downscale to 1440p to make everything look rich and better anti-aliased. I don't like the TAA or SMAA implementation of anti-aliasing in that game because they look more like blur filters. So I think I used FXAA and render higher then downscale to make it look better. So I'm definitely pushing my GPU to it's knees, but not CPU bottlenecked at all. Perhaps it's different if you're trying to get ultra high frame rates at 1080p in a game. For single player games, I try to max out the quality and smoothness of the game and keep it around 60-80fps for best immersion experience possible.

LMAO I used similar settings on re2. fxaa, 130% on the upscaler, all the blury options set to off or toned down :D have you compared max settings to lowest yet? spoiler, low settings look pretty damn good. havent finished hunks run through yet, keep gettin pwned by mr x :/ or the new extra scenarios. Gonna come back to it after DMC 5 :p

Back on topic though, you saw no CPU bottlenecking at all? I've been told that the best way to check is to turn down the resolution to 720p, set all graphics settings to their lowest and turn off vsync. Does NZXT Cam show gpu utilisation? if not, afterburner does. Could you do this test for me? just so i know one or the other that the cpu in my setup is not the issue?
 

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
I chose windows ltsb as my operating system because i wanted a windows 7 experience on windows 10 and so far it hasnt disappointed. I update when i want, disable stupid things like cortana, its great.

However, I have an issue. I use an r5 2600 and a 980ti with 3600mhz ram in dual channel. I game at 1080p on a 144hz monitor and i get quite a bit of cpu bottlenecking in cpu heavy games. my gpu utilisation in these titles is 70-80%.

Could the problem be my OS?
 

Rogue Leader

It's a trap!
Moderator
Windows LTSB is part of the Enterprise branch, have you installed all the latest drivers? Also gaming wise Microsoft makes constant improvements, some of which you may not have and may be holding you back. Where did you get this copy of Windows, running Enterprise for home use is prohibitively expensive.
 

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
Windows LTSB is part of the Enterprise branch, have you installed all the latest drivers? Also gaming wise Microsoft makes constant improvements, some of which you may not have and may be holding you back. Where did you get this copy of Windows, running Enterprise for home use is prohibitively expensive.

Reply all drivers are up to date. im more concerned with not having ryzen specific updates on this OS. A quick check has told me this seems to be the case. so my question is, will these updates improve my cpu bottleneck problem? or is my issue hardware related, weak single core strength or something.

does anyone have a ryzen 2600 and a powerful gpu on standard windows that could give me some info on this issue? maybe do a couple of tests for me? I'm seriously considering a switch back to intel and dont want to do it unless theres no other option.
 
So I ran the test and RE2 is a very well optimized game. It uses 6-7 threads very well, but is very much GPU bottlenecked on my Vega 64. I'm pegged at 99-100% the whole time, and it might dip down to 95% in some scenes, it's basically maxed and certainly GPU bottlenecked for me.

In RE2 at least, your problem is actually the 6GB of GDDR5 memory. My RE2 is using 6.5GB of actual HBM2 memory.

Here's the graph, the beginning was just me in the menu system getting ready to enter the game, and exiting the game at the end:
uc
 
Be careful how you interpret cpu usage.

Windows will spread the activity of a single thread over all available threads.
So, if you had a game that was single threaded and cpu bound, it would show up on a quad core processor as 25%
utilization across all 4 threads.
leading you to think your bottleneck was elsewhere.
It turns our that few games can usefully use more than 2-3 threads.
How can you tell how well threaded your games or apps are?
One way is to disable one thread and see how you do.

You can do this in the windows msconfig boot advanced options option.
You will need to reboot for the change to take effect. Set the number of processors to less than you have.
This will tell you how sensitive your games are to the benefits of many threads.
If you see little difference, it tells you that you will not benefit from more cores.
Likely, a better clock rate will be more important.

My interpretation of your graph is that the likely situation is that the single master thread is running full out and that there is the equivalent of a second or third thread doing much less work.
 
Be careful how you interpret cpu usage.

Windows will spread the activity of a single thread over all available threads.
So, if you had a game that was single threaded and cpu bound, it would show up on a quad core processor as 25%
utilization across all 4 threads.
leading you to think your bottleneck was elsewhere.
It turns our that few games can usefully use more than 2-3 threads.
How can you tell how well threaded your games or apps are?
One way is to disable one thread and see how you do.

You can do this in the windows msconfig boot advanced options option.
You will need to reboot for the change to take effect. Set the number of processors to less than you have.
This will tell you how sensitive your games are to the benefits of many threads.
If you see little difference, it tells you that you will not benefit from more cores.
Likely, a better clock rate will be more important.

My interpretation of your graph is that the likely situation is that the single master thread is running full out and that there is the equivalent of a second or third thread doing much less work.

No, when actually looking at all cores, 1 core is doing about 75%, and 5-6 threads are all doing about 40%. The rest are doing 0-10%. It does spread well across 6-7 cores, I just can't fit all the graphs on a single image. But as you can see from the very top graph, it's GPU limtied at 100% gpu usage.
 

Rogue Leader

It's a trap!
Moderator
Reply all drivers are up to date. im more concerned with not having ryzen specific updates on this OS. A quick check has told me this seems to be the case. so my question is, will these updates improve my cpu bottleneck problem? or is my issue hardware related, weak single core strength or something.

does anyone have a ryzen 2600 and a powerful gpu on standard windows that could give me some info on this issue? maybe do a couple of tests for me? I'm seriously considering a switch back to intel and dont want to do it unless theres no other option.

I have an 1800X with a Vega 64 and I will tell you the Ryzen related windows updates are important. There are plenty of games that will 100% that CPU, but I can also get 100% out of the GPU so I'm not really concerned, they are a decent match. There have been significant changes in the coding in Windows to allow Ryzen to fully work properly.

I understand your desire to hold back Windows updates, but there are ways on Home or Pro to prevent updates as well without locking yourself out from things that will help you.
 

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
So I ran the test and RE2 is a very well optimized game. It uses 6-7 threads very well, but is very much GPU bottlenecked on my Vega 64. I'm pegged at 99-100% the whole time, and it might dip down to 95% in some scenes, it's basically maxed and certainly GPU bottlenecked for me.

In RE2 at least, your problem is actually the 6GB of GDDR5 memory. My RE2 is using 6.5GB of actual HBM2 memory.

Here's the graph, the beginning was just me in the menu system getting ready to enter the game, and exiting the game at the end:
uc

in RE2 i use settings that keeps my vram at 5GBish. so it cant be a vram issue. Did you do this test in 720p low settings? most of my current research online is leading me to believe that the 2600 will bottleneck anything above a gtx 1060 @1080p. This seems to be true for a lot of intels lower and mid range cpu's too.

I'm going to try dual booting with windows 10 pro and see if i can get a performance uplift this way.
 

daylightriot

Commendable
Mar 4, 2018
89
5
1,545
Update

After work today I set up dual boot and installed windows 10 pro. I updated W10 pro installing all updates. A quick check showed a number of ryzen updates installed. I then used the same settings in RE2, the same gpu driver, and recorded a gpu utilisation video with shadow play on both OS's.

Video 1 windows 10 pro
View: https://youtu.be/aKfhFmay-jY


Video 2 windows ltsb
View: https://youtu.be/sFe23Ddl9yc


as you can see, the so called ryzen and gaming updates do next to nothing.

I believe i have done everything possible at this point to try and remove this bottleneck at a software level. this leaves a hardware upgrade the only feasible option.

I would like to thank all of you for your contributions in helping with this issue. I'll wait for next gen ryzen, watch every hardware review benchmark of it i can as well as check single thread scores, if amd still lags behind intel I will change platform.
 

Rogue Leader

It's a trap!
Moderator
Thanks for posting your findings.

I do want to point out if you look at basically any gaming test Ryzen processors always lag behind Intel in terms of pure FPS performance. If you want the maximum FPS from whatever system you are building, Ryzen is not your first choice.

The advantage with Ryzen is that the performance drop off is minimal, and still more than enough to saturate todays gaming GPUs giving you well over 60 fps at 1080p in any game, and not a massive percentage behind Intel, all the while being in many cases significantly cheaper for the processor and beating Intel at certain other tasks (whether or not those are important to you is personal preference). At higher resolutions the Ryzen processors are side by side with intel as the GPU becomes the limiting factor.

This is one of those times that yes there is a "bottleneck" in the purest sense of the word. But the reality is, is it affecting your gaming? Not really, no. Do your games play smoothly and at a high fps, without performance issues? Yes they do. Bottlenecks aren't going to hurt your equipment, and in many cases unless you are playing competitively, does it really matter if you get only 134 fps vs an Intel processor at 144 fps? It doesn't. Is it worth $100-$300 more for the processor? Certainly not.
 
  • Like
Reactions: RobCrezz