Ryzen 3600 upgrade very underwhelming

Swish · Jan 17, 2020

No man... dont tell me the 3600 is a downgrade from a 4790K.... I'm busy doing the same "upgrade"

Sent from my SM-N970F using Tapatalk

reactor_sa · Jan 18, 2020

ash515 said:
Changed to this and this is current:

Memory 3200
FCLK 1600
CPU core ratio 42
Precision boost overdrive enabled
Dram timing 16-19-19-19-38

Either the voltage crashed or timings, I suspect timings.
From setting the CPU to 42 and doing timings, I gained 27 fps on max and 2 on average. Still quite a bit behind the 4790K

going to keep tweaking!

CPU vtage 1.25 (offset +0.15)
SOC vtage 1.025 (offset - 0.075)
RAM vtage 1.35

I think gamers nexus did a thing on timings and zen 2 performance tuning. Check it on YouTube or their site. There are some sub timings that play a big part.

ash515 · Jan 18, 2020

Swish said:
No man... dont tell me the 3600 is a downgrade from a 4790K.... I'm busy doing the same "upgrade"

Sent from my SM-N970F using Tapatalk

I don't think it is I think my setup is not performing at full speed.
Other games I noticed a big fps and frametime improvements like in CS:GO

ash515 · Jan 19, 2020

Hi All

Reloaded windows and started from scratch.

Did a few tests with setting my ram to:

3400Mhz CL17
3200Mhz CL16
3000Mhz CL15

CPU Core ratio to 42 (even with fan at full speed and messing with voltage temps were too high)

In Odyssey I got very high settings:

Average 62
Min 38
Max 115

This was 1080p Very high preset (I know it isn't ideal with a few settings crushing performance) was using this as a direct comparison to other websites / reviews etc

Other games tested against other tests and older runs I did on my previous setup still showing I am off the mark by a 10-20 FPS difference.

Overclocking the CPU and adjusting RAM is not really showing any significant improvement or change.

Using afterburner I did not see anything irregular, CPU boosted to 4.1 and GPU was hitting the same clocks as before.

Any further ideas? Software driver issue?

baasgene · Jan 20, 2020

ash515 said:
Hi All

Reloaded windows and started from scratch.

Did a few tests with setting my ram to:

3400Mhz CL17
3200Mhz CL16
3000Mhz CL15

CPU Core ratio to 42 (even with fan at full speed and messing with voltage temps were too high)

In Odyssey I got very high settings:

Average 62
Min 38
Max 115

This was 1080p Very high preset (I know it isn't ideal with a few settings crushing performance) was using this as a direct comparison to other websites / reviews etc

Other games tested against other tests and older runs I did on my previous setup still showing I am off the mark by a 10-20 FPS difference.

Overclocking the CPU and adjusting RAM is not really showing any significant improvement or change.

Using afterburner I did not see anything irregular, CPU boosted to 4.1 and GPU was hitting the same clocks as before.

Any further ideas? Software driver issue?

Could you perhaps run 3dmark Timespy and post the results for us? That is the best indication we can get. Also just keep HWMonitor running in the background so that we can confirm that the GPU and CPU is running at 100% whilst running TimeSpy. Afterburner is fine too.

I presume all your temperatures are fine too?

One of my family friends bought a Ryzen 3400g combo from Evetech and he was running a GTX 960 on that, but it was completely useless it scored like 800 physics score in Time spy. Three days later the pc just switched off and couldn't switch on again, had to send it back to Evetech in the end. I tested my PSU on it and it still didn't boot and tried my RAM, I even reseated that CPU. Evetech barely had thermal paste on that thing sheesh, it was completely see through to the heatsink lol luckily I still had some thermal paste left from my previous build.

So I presume his motherboard was the culprit. I'll whatsapp him today and ask what happened I actually forgot about it, it was during the holidays.

Anyway it's unrelated, run a benchmark so we can compare results

ScottulusMaximus · Jan 20, 2020

ash515 said:
Hi All

Reloaded windows and started from scratch.

Did a few tests with setting my ram to:

3400Mhz CL17
3200Mhz CL16
3000Mhz CL15

CPU Core ratio to 42 (even with fan at full speed and messing with voltage temps were too high)

In Odyssey I got very high settings:

Average 62
Min 38
Max 115

This was 1080p Very high preset (I know it isn't ideal with a few settings crushing performance) was using this as a direct comparison to other websites / reviews etc

Other games tested against other tests and older runs I did on my previous setup still showing I am off the mark by a 10-20 FPS difference.

Overclocking the CPU and adjusting RAM is not really showing any significant improvement or change.

Using afterburner I did not see anything irregular, CPU boosted to 4.1 and GPU was hitting the same clocks as before.

Any further ideas? Software driver issue?

You're not going to see a huge improvement in fps in most games, that'll still be mostly limited by the gfx card. So testing by benchmarking games isn't going to show anything.

Qui_Illustrati · Jan 20, 2020

Run a Cinebench 15 / 20. Compare results of the CPU with other results from the same CPU and comparative CPU's.

Sola Scriptura · Jan 20, 2020

ash515 said:
Hi All

Reloaded windows and started from scratch.

Did a few tests with setting my ram to:

3400Mhz CL17
3200Mhz CL16
3000Mhz CL15

CPU Core ratio to 42 (even with fan at full speed and messing with voltage temps were too high)

In Odyssey I got very high settings:

Average 62
Min 38
Max 115

This was 1080p Very high preset (I know it isn't ideal with a few settings crushing performance) was using this as a direct comparison to other websites / reviews etc

Other games tested against other tests and older runs I did on my previous setup still showing I am off the mark by a 10-20 FPS difference.

Overclocking the CPU and adjusting RAM is not really showing any significant improvement or change.

Using afterburner I did not see anything irregular, CPU boosted to 4.1 and GPU was hitting the same clocks as before.

Any further ideas? Software driver issue?

download and run userbenchmark then post the results here. you can compare your cpu and ram score with others on the website.

Home - UserBenchmark

Top Market Research & Genuine Product Insight from our Users. Quickly find category leading products with UserBenchmark.

www.userbenchmark.com

Dollos · Jan 20, 2020

why dont you borrow someone elses ram just to check wha happens? i reckon one component is of..ryzen really is fussy about ram...

JollyJamma · Jan 20, 2020

Blood_Bought said:
download and run userbenchmark then post the results here. you can compare your cpu and ram score with others on the website.

Home - UserBenchmark

Top Market Research & Genuine Product Insight from our Users. Quickly find category leading products with UserBenchmark.

www.userbenchmark.com

UserBenchmark is useless, especially for AMD CPUs.

ash515 · Jan 20, 2020

ScottulusMaximus said:
You're not going to see a huge improvement in fps in most games, that'll still be mostly limited by the gfx card. So testing by benchmarking games isn't going to show anything.

The thing is how have my fps gone backwards that much?

Surely I would see linear results if it was my GPU bottlenecking

ash515 · Jan 20, 2020

Qui_Illustrati said:
Run a Cinebench 15 / 20. Compare results of the CPU with other results from the same CPU and comparative CPU's.

Will do this again and post results.

Going to include hwinfo printout, msi afterburner while in game.

baasgene · Jan 20, 2020

ash515 said:
Will do this again and post results.

Going to include hwinfo printout, msi afterburner while in game.

To rather reflect overall performance, a TimeSpy run would really be cool. Maybe for all we know your GPU is not firing on all cylinders for some reason. So if your physics score on Time Spy is fine, yet your graphics score is not then we can narrow the problem down.

virus786 · Jan 20, 2020

AMD is like Evetech. You never know what you're getting after you've spent your money.
Prove me wrong.

baasgene · Jan 20, 2020

a 1070 G1 should pull around 182 watts standard, and the Ryzen at stock should be around 140 watts during multi threaded loads. Then given the rest of your goods you should be in the range of 400 watt.

Any overclocking will greatly increase that, given that your psu is Bronze certified, we are cutting it close. It should be fine though, how old is the psu for interest sake? Now correct me if I'm wrong, I'm not a technical person, but is a 520watt Bronze certified power supply not basically 520w minus 15%? Because it's up to 85% efficient? Meaning we are left with around 450 watts give or take?

Reason why I am asking, your Cinebench score of 475 and 3510 is perfectly in line. It seems the problem comes in when the whole system is under stress, which is why I want to see the Time Spy results.

ash515 · Jan 20, 2020

baasgene said:
a 1070 G1 should pull around 182 watts standard, and the Ryzen at stock should be around 140 watts during multi threaded loads. Then given the rest of your goods you should be in the range of 400 watt.

Any overclocking will greatly increase that, given that your psu is Bronze certified, we are cutting it close. It should be fine though, how old is the psu for interest sake?

Reason why I am asking, your Cinebench score of 475 and 3510 is perfectly in line. It seems the problem comes in when the whole system is under stress, which is why I want to see the Time Spy results.

I will run Time Spy tonight when I am home and leave afterburner running to catch results.

When I had my 4790K my GPU would boost to 1949Mhz core and 4100Mhz memory, unfortunately my GTX 1070 overclocks like ass.

This is identical with the 3600.

Max temp under full load was 68 degrees.

The memory used has some issue from what I read online and even a 50mhz bump on the core crashed most games.

Why I left it on stock OC mode.

Well it is a really good powersupply hasn't given me any issues since day one, it is 4 years old though.
I do have a evga 520w I can potentially wrestle out of my brothers PC to test but I don't see this being a issue as my pc never restarted once under load.

baasgene · Jan 20, 2020

ash515 said:
I will run Time Spy tonight when I am home and leave afterburner running to catch results.

When I had my 4790K my GPU would boost to 1949Mhz core and 4100Mhz memory, unfortunately my GTX 1070 overclocks like ass.

This is identical with the 3600.

Max temp under full load was 68 degrees.

The memory used has some issue from what I read online and even a 50mhz bump on the core crashed most games.

Why I left it on stock OC mode.

Well it is a really good powersupply hasn't given me any issues since day one, it is 4 years old though.
I do have a evga 520w I can potentially wrestle out of my brothers PC to test but I don't see this being a issue as my pc never restarted once under load.

Mmm first let's just check out that Time Spy score tonight.

Eon · Jan 20, 2020

virus786 said:
AMD is like Evetech. You never know what you're getting after you've spent you money.
Prove me wrong.

This is a lot more complicated than trying to throw in an appeal to ignorance. Your are stating a false dichotomy.

My money is also on the graphics card, If you are getting an improvement in cs:go which is known as CPU intensive then the relationship between the gfx card and the cpu could be the drop in frames in other games, and also not seeing any improvement in some.

Qui_Illustrati · Jan 20, 2020

virus786 said:
AMD is like Evetech. You never know what you're getting after you've spent your money.
Prove me wrong.

or ... you could just prove yourself right.

You're sadly mistaken if you think Intel processors don't experience issues.
Besides, 90% of people that actually know what they're talking about in the industry right now agree that AMD is kicking Intels ass in every way that matters

The 3600 is one of the best bang for buck CPU's around. I'm sure the OP will get to the bottom of the issue.

ash515 · Jan 20, 2020

Recho said:
This is a lot more complicated than trying to throw in an appeal to ignorance. Your are stating a false dichotomy.

My money is also on the graphics card, If you are getting an improvement in cs:go which is known as CPU intensive then the relationship between the gfx card and the cpu could be the drop in frames in other games, and also not seeing any improvement in some.

You have a valid point, however I am using latest drivers fresh installed and used DDU to remove older ones.
GPU is behaving the same was as before from checking core clocks and memory clocks, power etc.

I do not have another GPU to test, only other GPU in the house is a 1060 3gb evga

ash515 · Jan 20, 2020

Qui_Illustrati said:
or ... you could just prove yourself right.

You're sadly mistaken if you think Intel processors don't experience issues.
Besides, 90% of people that actually know what they're talking about in the industry right now agree that AMD is kicking Intels ass in every way that matters

The 3600 is one of the best bang for buck CPU's around. I'm sure the OP will get to the bottom of the issue.

Every where I read the 3600 is a beast and I did extensive research before I upgraded.

There must be a setting, driver, overlock etc I am missing that is eating that last bit of performance.

Qui_Illustrati · Jan 20, 2020

ash515 said:
Every where I read the 3600 is a beast and I did extensive research before I upgraded.

There must be a setting, driver, overlock etc I am missing that is eating that last bit of performance.

I was running a 3600 quite happily with a 1080ti. Bloody amazing value/price for the performance.

What does Cinebench show?

baasgene · Jan 20, 2020

ash515 said:
You have a valid point, however I am using latest drivers fresh installed and used DDU to remove older ones.
GPU is behaving the same was as before from checking core clocks and memory clocks, power etc.

I do not have another GPU to test, only other GPU in the house is a 1060 3gb evga

If your Time Spy score is in the region of 7000 physics and 5700-6000 graphics, then both the GPU and CPU is fine. Time Spy doesn't stress the whole system really.

You could also run some stress test loops with 3dmark, FPS must stay above a 97% stable margin (meaning FPS shouldn't differ more than 3% between loops)

ash515 · Jan 20, 2020

Qui_Illustrati said:
I was running a 3600 quite happily with a 1080ti. Bloody amazing value/price for the performance.

What does Cinebench show?

I would love a 1080ti!

Multithreaded - 3510
Single core - 475

That was with:

CPU Ratio to auto
Voltages on auto
RAM XMP Applied - 3200mhz CL16-19-19-19-38, Voltage 1.35V

Hwinfo showed CPU Boosted to 4.1GHZ during the cinebench run.

Qui_Illustrati · Jan 20, 2020

Also, try the Unigine benchmark - i find it stresses the GPU better than 3DMark IMO. Just that and Cinebench should give you numbers you can compare online.

UNIGINE: real-time 3D engine

Real-time 3D engine / VR platform for Simulation & Training, 3D Digital Twins, CAD, BIM, GIS. 3D graphics and physics library for С++ and C# developers.

unigine.com

ash515 · Jan 20, 2020

baasgene said:
If your Time Spy score is in the region of 7000 physics and 5700-6000 graphics, then both the GPU and CPU is fine. Time Spy doesn't stress the whole system really. That's why I would then pinpoint your psu not having enough juice to keep them all running at full speed 24/7.

You could also run some stress test loops with 3dmark, FPS must stay above a 97% stable margin (meaning FPS shouldn't differ more than 3% between loops)

Okay thank you, let me try this as well tonight.

baasgene · Jan 20, 2020

ash515 said:
Okay thank you, let me try this as well tonight.

Scratch the PSU, it should be fine, I just did some research and not having enough power cannot negatively affect performance, it would rather shut down instead.

So it's not the PSU.

Qui_Illustrati · Jan 20, 2020

An article mentioning similar numbers to your Cinebench

Ryzen 5 3600 low 3DMark scores

My CPU is getting 6,299 on Time Spy and 17,752 on Firestrike, but my Cinebench scores are normal (479 single, 3,411 CPU).

linustechtips.com

How do you compare to these :

3DMark.com search

www.3dmark.com

baasgene · Jan 20, 2020

Qui_Illustrati said:
An article mentioning similar numbers to your Cinebench

Ryzen 5 3600 low 3DMark scores

My CPU is getting 6,299 on Time Spy and 17,752 on Firestrike, but my Cinebench scores are normal (479 single, 3,411 CPU).

linustechtips.com

How do you compare to these :

3DMark.com search

3DMark.com search

www.3dmark.com

Yeah add a Fire Strike run just to be safe, then we cover DX11 and DX12. And Fire Strike is cpu limited these days.

ash515 · Jan 20, 2020

Qui_Illustrati said:
An article mentioning similar numbers to your Cinebench

Ryzen 5 3600 low 3DMark scores

My CPU is getting 6,299 on Time Spy and 17,752 on Firestrike, but my Cinebench scores are normal (479 single, 3,411 CPU).

linustechtips.com

How do you compare to these :

3DMark.com search

3DMark.com search

www.3dmark.com

Bios settings

CPU Ratio - Auto
XMP applied - 3200mhz CL16-20-20-20-38 @ 1.35V
PBO enabled

Run 1:

Cinebench R15

Single Core - 189

Multithread - 1505

Cinebench R20

Single core - 471
Multithreaded - 3376 (this went down)

Bios settings

CPU Ratio - 42
Manual - 3200mhz CL16-19-19-19-38 @ 1.35V
PBO enabled
CPU Vcore - 1.13750

Run 2:

Cinebench R15

Single Core - 190

Multithread - 1512

Cinebench R20

Single core - 473
Multithreaded - 3449

Ryzen 3600 upgrade very underwhelming

Legendary Member

Forum Addict

Senior Member

Senior Member

VIP

Legendary Member

Proud Dad!

VIP

cHeAp ChArO

What's my age again?

Senior Member

Senior Member

VIP

The iNFEKt0R

VIP

Senior Member

VIP

Google

Proud Dad!

Senior Member

Senior Member

Proud Dad!

VIP

Senior Member

Proud Dad!

Senior Member

VIP

Proud Dad!

VIP

Senior Member

Users who are viewing this thread

Similar threads

Top Donors

Share this page