Jump to content

Bad performance on brand new 3900x?

Go to solution Solved by Leanora,

Seems like you are using gaming mode.

Please go to AMD Ryzen Master and disable it, then try again.

Hey guys,

I just built a new rig and I'm getting really crappy performance when running Cinebench r20 and other benchmarks. I'm getting around 3k-3.5k in cinebench. That's half of what I should be getting. The funny thing is, the first time I ran it (a few days ago) I got almost 7k in Cinebench. But I messed with some bios settings trying to get the voltage to drop below 1.4 when idle. The only thing I did was turn off Precision Boost Override (PBO) in BIOS. Since than I turned it back on, reset bios to optimized defaults, messed with power settings in windows... I'm out of ideas. This is also my first AMD build.

I am running a custom loop and temps are acceptable. I'm getting mid 40s when idle and the highest temp I saw after many passes in Cinebench was 78. Not the greatest temp, but far below the 95 throttle limit. From what I read, temps on this CPU are pretty crazy for everyone so I haven't worried about that too much. 

Motherboard: ROG Strix X570-E Gaming

CPU: Ryzen 9 3900x

Ram: G.SKILL TridentZ RGB Series 32GB
Cooling: 1 cpu + 1 gpu on 2 360mm rads running with 6 corsair high pressure fans

PSU: 1300 Watt Seasonic Platinum

 

Anyone have any ideas? 

Link to comment
Share on other sites

Link to post
Share on other sites

Just clear CMOS

CPU: i7-2600K 4751MHz 1.44V (software) --> 1.47V at the back of the socket Motherboard: Asrock Z77 Extreme4 (BCLK: 103.3MHz) CPU Cooler: Noctua NH-D15 RAM: Adata XPG 2x8GB DDR3 (XMP: 2133MHz 10-11-11-30 CR2, custom: 2203MHz 10-11-10-26 CR1 tRFC:230 tREFI:14000) GPU: Asus GTX 1070 Dual (Super Jetstream vbios, +70(2025-2088MHz)/+400(8.8Gbps)) SSD: Samsung 840 Pro 256GB (main boot drive), Transcend SSD370 128GB PSU: Seasonic X-660 80+ Gold Case: Antec P110 Silent, 5 intakes 1 exhaust Monitor: AOC G2460PF 1080p 144Hz (150Hz max w/ DP, 121Hz max w/ HDMI) TN panel Keyboard: Logitech G610 Orion (Cherry MX Blue) with SteelSeries Apex M260 keycaps Mouse: BenQ Zowie FK1

 

Model: HP Omen 17 17-an110ca CPU: i7-8750H (0.125V core & cache, 50mV SA undervolt) GPU: GTX 1060 6GB Mobile (+80/+450, 1650MHz~1750MHz 0.78V~0.85V) RAM: 8+8GB DDR4-2400 18-17-17-39 2T Storage: HP EX920 1TB PCIe x4 M.2 SSD + Crucial MX500 1TB 2.5" SATA SSD, 128GB Toshiba PCIe x2 M.2 SSD (KBG30ZMV128G) gone cooking externally, 1TB Seagate 7200RPM 2.5" HDD (ST1000LM049-2GH172) left outside Monitor: 1080p 126Hz IPS G-sync

 

Desktop benching:

Cinebench R15 Single thread:168 Multi-thread: 833 

SuperPi (v1.5 from Techpowerup, PI value output) 16K: 0.100s 1M: 8.255s 32M: 7m 45.93s

Link to comment
Share on other sites

Link to post
Share on other sites

I've done that. I've done the whole jumper reset procedure in the motherboard manual and all my settings reverted to default. Still no difference in scores.

Link to comment
Share on other sites

Link to post
Share on other sites

What speed is it running when you do the test?  

 Use the default settings in bios, don't forget to set xmp on whatever it was when you ran it with 7k score.

In windows set the power config on amd balanced settings 

 

mid 40 on idle is okay 

78 has more to do with your fan profile or fans themselves I think ? 

78 is still acceptable I think 

 

I have 2x 360 with noctua industrial 2000 rpm fans 

on idle is fluctuates between 39 and 52, it crawls up and jumps down when fans go a bit higher.

On full load it doesn't go above 66

 

 

You might also want to try reflashing the bios.  Might be an issue with not properly clearing the bios when going back to defaults. I do not know if that's a problem or not with ryzen 3/500 series though

AMD 5900x, 2x16GB ( G.skill F4-3600C16 ), Radeon VII, Radeon 570,  Corsair 1200i , SSD 240/500GB, HDD 640GB,1TB,5TB, 1TB NVME. Lian li 011Dyanmic XL, big ass LTT mouse-pad & stealth water bottle

Link to comment
Share on other sites

Link to post
Share on other sites

try setting windows power mode to higher performance another thing you can try is to check task manager while running cinebench to see if any other programs are hogging resources so open resource monitor and monitor resource usage something is hogging cpu usage big time hence the low scores also a bios reflash might help

 

as for temps i have to ask what thermal paste are you using and also are your noctua fans designed for high static pressure fans can make a difference in temps for a rad for example a standard noctua case fan running at 2000 rpm will cool a rad worse then a high static pressure fan running at 800 rpm because the static pressure fan is designed to move more air through tighter spaces

Link to comment
Share on other sites

Link to post
Share on other sites

Gonna ask did you use the MOBO software suite to make any changes to the settings if so ditch that crap and remove it from your system they can and do have persistent settings that will drive you nuts if thats the case

My daily driver: The Wrath of Red: OS Windows 10 home edition / CPU Ryzen TR4 1950x 3.85GHz / Cooler Master MasterAir MA621P Twin-Tower RGB CPU Air Cooler / PSU Thermaltake Toughpower 750watt / ASRock x399 Taichi / Gskill Flare X 32GB DDR4 3200Mhz / HP 10GB Single Port Mellanox Connectx-2 PCI-E 10GBe NIC / Samsung 512GB 970 pro M.2 / ASUS GeForce GTX 1080 STRIX 8GB / Acer - H236HLbid 23.0" 1920x1080 60Hz Monitor x3

 

My technology Rig: The wizard: OS Windows 10 home edition / CPU Ryzen R7 1800x 3.95MHz / Corsair H110i / PSU Thermaltake Toughpower 750watt / ASUS CH 6 / Gskill Flare X 32GB DDR4 3200Mhz / HP 10GB Single Port Mellanox Connectx-2 PCI-E 10GBe NIC / 512GB 960 pro M.2 / ASUS GeForce GTX 1080 STRIX 8GB / Acer - H236HLbid 23.0" 1920x1080 60Hz Monitor HP Monitor

 

My I don't use RigOS Windows 10 home edition / CPU Ryzen 1600x 3.85GHz / Cooler Master MasterAir MA620P Twin-Tower RGB CPU Air Cooler / PSU Thermaltake Toughpower 750watt / MSI x370 Gaming Pro Carbon / Gskill Flare X 32GB DDR4 3200Mhz / Samsung PM961 256GB M.2 PCIe Internal SSDEVGA GeForce GTX 1050 Ti SSC GAMING / Acer - H236HLbid 23.0" 1920x1080 60Hz Monitor

 

My NAS: The storage miser: OS unRAID v. 6.9.0-beta25 / CPU Intel i7 6700 / Cooler Master MasterWatt Lite 500 Watt 80 Plus / ASUS Maximus viii Hero / 32GB Gskill RipJaw DDR4 3200Mhz / HP Mellanox ConnectX-2 10 GbE PCI-e G2 Dual SFP+ Ported Ethernet HCA NIC / 9 Drives total 29TB - 1 4TB seagate parity - 7 4TB WD Red data - 1 1TB laptop drive data - and 2 240GB Sandisk SSD's cache / Headless

 

Why did I buy this server: OS unRAID v. 6.9.0-beta25 / Dell R710 enterprise server with dual xeon E5530 / 48GB ecc ddr3 / Dell H310 6Gbps SAS HBA w/ LSI 9211-8i P20 IT / 4 450GB sas drives / headless

 

Just another server: OS Proxmox VE / Dell poweredge R410

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, dreamcast4599 said:

try setting windows power mode to higher performance another thing you can try is to check task manager while running cinebench to see if any other programs are hogging resources so open resource monitor and monitor resource usage something is hogging cpu usage big time hence the low scores also a bios reflash might help

 

as for temps i have to ask what thermal paste are you using and also are your noctua fans designed for high static pressure fans can make a difference in temps for a rad for example a standard noctua case fan running at 2000 rpm will cool a rad worse then a high static pressure fan running at 800 rpm because the static pressure fan is designed to move more air through tighter spaces

 

I think you are mixing up my comment with that of the OP's

 

The OP has 6 corsair high pressure fans. 

 

As for my Noctua fans, yes they are high pressure and switch between around 900 to 1000 or mayhap I should say me not hearing them at all and me hearing them in the background. 

AMD 5900x, 2x16GB ( G.skill F4-3600C16 ), Radeon VII, Radeon 570,  Corsair 1200i , SSD 240/500GB, HDD 640GB,1TB,5TB, 1TB NVME. Lian li 011Dyanmic XL, big ass LTT mouse-pad & stealth water bottle

Link to comment
Share on other sites

Link to post
Share on other sites

Ok so I tried everything that was suggested. I even reinstalled the chipset drivers so that the AMD profiles would be reinstalled. I just flashed the mobo bios and ran cinebench again:
 image.png.c8ffb9a287663d1e09096cb8453745f3.png
As you can see, i'm still getting mid 3000s (the most recent one is the one in orange). I have half a mind to just reinstall windows and start from scratch.

On all core Cinebench my CPU sits at ~1.35v and 4.2 GHz. Single core it goes up to 4.7 GHz and sits at 1.4v. I watched Task Manager and nothing is taking significant CPU performance.
I have also considered that this might be a cinebench issue but my initial 3D mark run was also brought down by half.

Initial Run: image.png.0ddddb9108bab4d658db2cc0aa0ad375.png  

 

 

 

Current run: image.png.edd8f73f31e38b75feaa9730482f0a0c.png

 

 

I've tried windows power on every setting even the default windows ones.
I'm using thermal grizzly kryonaut thermal paste that came with my EK Velocity Water Block.


I did install the Asus AI suite and the Armoury Crate that was prompted when I first booted into windows. But I have since uninstalled it in my attempts to troubleshoot this.

Link to comment
Share on other sites

Link to post
Share on other sites

So I have found something while looking at Cinebench just now:

image.png.4c642e7501c7bbcc6b2c15a0f26fec3f.png
image.png.e2d0c8ae8c696458a6811a2e232e45d1.png
Cinebench and Task Manager only see half my threads and cores. Which explains why my score is cut in half. What would cause this? I haven't shut down any cores. And even if I did, I would imagine a full Bios re-flash would undo that.


Any Ideas?

Link to comment
Share on other sites

Link to post
Share on other sites

Seems like you are using gaming mode.

Please go to AMD Ryzen Master and disable it, then try again.

AMD 5900x, 2x16GB ( G.skill F4-3600C16 ), Radeon VII, Radeon 570,  Corsair 1200i , SSD 240/500GB, HDD 640GB,1TB,5TB, 1TB NVME. Lian li 011Dyanmic XL, big ass LTT mouse-pad & stealth water bottle

Link to comment
Share on other sites

Link to post
Share on other sites

I think there are more than you enable "game mode". The R20 score is just low. Some setting is off as well.

 

Even with Game mode, the score should be around 3800 and Creator mode close to 7400 with default(no PBO,auto OC).

Link to comment
Share on other sites

Link to post
Share on other sites

20 minutes ago, Deli said:

I think there are more than you enable "game mode". The R20 score is just low. Some setting is off as well.

 

Even with Game mode, the score should be around 3800 and Creator mode close to 7400 with default(no PBO,auto OC).

From a quick google and my own test it's more like upper 6900- low 7200.

IF you're at 7400 with stock settings then congrats, it seems you have a golden sample.

 

I do wonder if the type of memory matters and to what point.

AMD 5900x, 2x16GB ( G.skill F4-3600C16 ), Radeon VII, Radeon 570,  Corsair 1200i , SSD 240/500GB, HDD 640GB,1TB,5TB, 1TB NVME. Lian li 011Dyanmic XL, big ass LTT mouse-pad & stealth water bottle

Link to comment
Share on other sites

Link to post
Share on other sites

22 minutes ago, Leanora said:

From a quick google and my own test it's more like upper 6900- low 7200.

IF you're at 7400 with stock settings then congrats, it seems you have a golden sample.

 

I do wonder if the type of memory matters and to what point.

I actually have a little lemon. My chip need 1.368v for 4.3Ghz stable. My R20 score for 4.3Ghz is 7710. So 73xx is normal for default. At default, the CPU will boost to ~4.1 Ghz all core.

 

One thing to point out is temperature. If it's high, it'll boost lower. I've a 280mm AIO, nothing special. Temperature peaks at 72C when running default with AIDA64. Some of the lower scores you see on the net, probably using the stock cooler or some less effective cooling solutions?

Link to comment
Share on other sites

Link to post
Share on other sites

11 minutes ago, Deli said:

I actually have a little lemon. My chip need 1.368v for 4.3Ghz stable. My R20 score for 4.3Ghz is 7710. So 73xx is normal for default. At default, the CPU will boost to ~4.1 Ghz all core.

 

One thing to point out is temperature. If it's high, it'll boost lower. I've a 280mm AIO, nothing special. Temperature peaks at 72C when running default with AIDA64. Some of the lower scores you see on the net, probably using the stock cooler or some less effective cooling solutions?

 

https://www.cgdirector.com/cinebench-r20-scores-updated-results/

Control +F 3900x

 

With stock settings xmp disabled

I'm at 7040-7100 

idle temp 33~, load 68~

2x 360 rads

peak speed 4140 on all cores

4385 single core

Numbers are from AMD Ryzen master

Latest bios ( 10F? )

 

When I use actual programs I get spikes hitting 46xx on some cores. ( with previous ABBA bios/ 7C? )

 

Anyhow I hope the OP can use this as reference

AMD 5900x, 2x16GB ( G.skill F4-3600C16 ), Radeon VII, Radeon 570,  Corsair 1200i , SSD 240/500GB, HDD 640GB,1TB,5TB, 1TB NVME. Lian li 011Dyanmic XL, big ass LTT mouse-pad & stealth water bottle

Link to comment
Share on other sites

Link to post
Share on other sites

17 minutes ago, Leanora said:

 

https://www.cgdirector.com/cinebench-r20-scores-updated-results/

Control +F 3900x

 

With stock settings xmp disabled

I'm at 7040-7100 

idle temp 33~, load 68~

2x 360 rads

peak speed 4140 on all cores

4385 single core

Numbers are from AMD Ryzen master

Latest bios ( 10F? )

 

When I use actual programs I get spikes hitting 46xx on some cores. ( with previous ABBA bios/ 7C? )

 

Anyhow I hope the OP can use this as reference

It seems there is quite a variant amount chips. I remember many 3900x reviews have R15 scores between 3100- 3200. However Gamer Nexus hits 327x with auto OC in its review. Mine can do 3279 also with auto OC. Or 3244 without. My Cinebench scores remain more or less the same with old and new AGESA.

 

I think RAM does nothing for Cinebench score. I have 3200MHz CL16 32GB.

 

Either way, OP only scores 6446 is way too low for whatever reasons.

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, Deli said:

It seems there is quite a variant amount chips. I remember many 3900x reviews have R15 scores between 3100- 3200. However Gamer Nexus hits 327x with auto OC in its review. Mine can do 3279 also with auto OC. Or 3244 without. My Cinebench scores remain more or less the same with old and new AGESA.

 

I think RAM does nothing for Cinebench score. I have 3200MHz CL16 32GB.

 

Either way, OP only scores 6446 is way too low for whatever reasons.

It was game mode. However, you're right. My temps are pretty high. Shouldn't be considering that I'm using a custom loop with 2 360 mm rads and 6 rad fans. At peak my temps hit 78 degrees. I used Kryonaut as my paste and I'm sure I applied it well ( done it plenty of times). Voltages are at stock though sitting at 1.4v pretty much constantly. I have to learn how to drop those down for my temps to drop down and hopefully my scores to improve. Do you have any resources I can use?

Not a big overclocker. Mostly done it on intel side which was stupid simple.

Link to comment
Share on other sites

Link to post
Share on other sites

Can you post a link to the exact fans you have, I want to take a look at the specs.

 

In BIOS, go to the fan page and make sure that the fans that are mounted on both rads cool according to the CPU sensor, not the system sensor or something else.

 

Check the fan curve.

 

image.png.e1de127a3636ade71ca9547f5a95731a.png

 

That's the setting I use for all fans ( 2x 3 way splitter ), the pump follows pretty much the same curve.

Try to find a balance that works well for you.

 

Edit:

On a side note, at 100% the fans go 2000 RPM, I don't know why the number is incorrect.

AMD 5900x, 2x16GB ( G.skill F4-3600C16 ), Radeon VII, Radeon 570,  Corsair 1200i , SSD 240/500GB, HDD 640GB,1TB,5TB, 1TB NVME. Lian li 011Dyanmic XL, big ass LTT mouse-pad & stealth water bottle

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, Clock_Work said:

It was game mode. However, you're right. My temps are pretty high. Shouldn't be considering that I'm using a custom loop with 2 360 mm rads and 6 rad fans. At peak my temps hit 78 degrees. I used Kryonaut as my paste and I'm sure I applied it well ( done it plenty of times). Voltages are at stock though sitting at 1.4v pretty much constantly. I have to learn how to drop those down for my temps to drop down and hopefully my scores to improve. Do you have any resources I can use?

Not a big overclocker. Mostly done it on intel side which was stupid simple.

I notice Ryzen Master and HWinfo report different Vcore voltages. The default voltage in the Gigabyte BIOS is 1.225v. Which HWinfo reports the same(almost) 1.226v. However Ryzen Master always shows 0.06-0.1v higher. It will hover around 1.32- 1.34v while doing Cinebench R20.

 

I play with undervolting and overvoltaging. Seem for my particular chip, -0.025v gives the best Cinebench score, improves like 15 points or so, nothing big, mind you. Trying to do massive 0.10v both under or overvolt will result big drop of Cinebench score.

 

I use a nothing-special 280mm AIO with two Noctua fans. With a maxmum of 1000rpm for quietness. It will only hit ~72C with default clock speed. Or 84C if manual overclock to 4.3GHz.

 

I think your mobo set the Vcore voltage too high for some reason. That explants both the low Cinebench score and high temperature.

Link to comment
Share on other sites

Link to post
Share on other sites

5 hours ago, Leanora said:

Can you post a link to the exact fans you have, I want to take a look at the specs.

 

In BIOS, go to the fan page and make sure that the fans that are mounted on both rads cool according to the CPU sensor, not the system sensor or something else.

 

Check the fan curve.

 

image.png.e1de127a3636ade71ca9547f5a95731a.png

 

That's the setting I use for all fans ( 2x 3 way splitter ), the pump follows pretty much the same curve.

Try to find a balance that works well for you.

 

Edit:

On a side note, at 100% the fans go 2000 RPM, I don't know why the number is incorrect.

These are my fans: Corsair LL120
I actually cool based on water temperature not on CPU temp. Having the fans ramp up every time I open an application is pointless and annoying as small bursts of heat don't actually do much to increase water temps in the long run. Its the water that needs to be cooled anyway. That's my reasoning at least, and it seems to be the sentiment in the water cooling community.
This is my Curve tho:
image.thumb.png.a8ee08bc14fded22cf726f7d3368067f.png
 

 

5 hours ago, Deli said:

I notice Ryzen Master and HWinfo report different Vcore voltages. The default voltage in the Gigabyte BIOS is 1.225v. Which HWinfo reports the same(almost) 1.226v. However Ryzen Master always shows 0.06-0.1v higher. It will hover around 1.32- 1.34v while doing Cinebench R20.

 

I play with undervolting and overvoltaging. Seem for my particular chip, -0.025v gives the best Cinebench score, improves like 15 points or so, nothing big, mind you. Trying to do massive 0.10v both under or overvolt will result big drop of Cinebench score.

 

I use a nothing-special 280mm AIO with two Noctua fans. With a maxmum of 1000rpm for quietness. It will only hit ~72C with default clock speed. Or 84C if manual overclock to 4.3GHz.

 

I think your mobo set the Vcore voltage too high for some reason. That explants both the low Cinebench score and high temperature.



My ram was actually defaulting to 1866 for some reason. I used the XMP profile for it and my Cinebench jumped up to 6999. I also lowered the cpu voltage to 1.3v and I seem to be maintaining a stable 41 degrees on idle and 57 on full usage now. I think I'd call the issue resolved haha. Ryzen is definitely more finicky than intel was. I mean when Skylake came out (my previous CPU), it had its quirks but nowhere near as puzzling as this.
Hitting 81 degrees at stock on a custom loop unless you bring voltages down is unacceptable no matter what AMD says. No wonder people with AIOs are hitting the thermal limit with even the most light overclocks.
I'm not gonna bother with overclocking since it doesn't seem like you can actually squeeze more than a few points out of the CPU regardless of how you approach it. So I just set a PBO overclock in Ryzen master and I'm calling it a day.

Thanks for all your help everyone.

Link to comment
Share on other sites

Link to post
Share on other sites

8 hours ago, Clock_Work said:

These are my fans: Corsair LL120
I actually cool based on water temperature not on CPU temp. Having the fans ramp up every time I open an application is pointless and annoying as small bursts of heat don't actually do much to increase water temps in the long run. Its the water that needs to be cooled anyway. That's my reasoning at least, and it seems to be the sentiment in the water cooling community.
This is my Curve tho:

My ram was actually defaulting to 1866 for some reason. I used the XMP profile for it and my Cinebench jumped up to 6999. I also lowered the cpu voltage to 1.3v and I seem to be maintaining a stable 41 degrees on idle and 57 on full usage now. I think I'd call the issue resolved haha. Ryzen is definitely more finicky than intel was. I mean when Skylake came out (my previous CPU), it had its quirks but nowhere near as puzzling as this.
Hitting 81 degrees at stock on a custom loop unless you bring voltages down is unacceptable no matter what AMD says. No wonder people with AIOs are hitting the thermal limit with even the most light overclocks.
I'm not gonna bother with overclocking since it doesn't seem like you can actually squeeze more than a few points out of the CPU regardless of how you approach it. So I just set a PBO overclock in Ryzen master and I'm calling it a day.

Thanks for all your help everyone.

I think it has more to do with mobo vendors setting the Vcore too high, rather than AMD's problem. At least since AGESA ABBA, the high spikes of voltage under low loads isn't a problem anymore.

Link to comment
Share on other sites

Link to post
Share on other sites

Sounds like the issue is solved.

As for cpu temp vs water temp;

It's responsiveness ( possible spikes ) vs stability. I don't think there is a right or wrong, it's more about a choice that you need to make.

Anyhow good luck with your system

AMD 5900x, 2x16GB ( G.skill F4-3600C16 ), Radeon VII, Radeon 570,  Corsair 1200i , SSD 240/500GB, HDD 640GB,1TB,5TB, 1TB NVME. Lian li 011Dyanmic XL, big ass LTT mouse-pad & stealth water bottle

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×