• 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."

AMD Ryzen 5000 Builders Thread

Page 52 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Noid

Platinum Member
Sep 20, 2000
2,283
103
106
Just realized I been here over 20 years ... uh
Anyway, had a complete shutdown yesterday " in game " again.
( no BSOD, no Logs of any kind this time ... )
So, yea .. a scary event....

I've never had this type of failure before, and my " gut feeling " was telling me my PSU was failing. ( Antec 750 - I'm pushing the limits on everything so ... )

Picked up a Corsair HX1000 - hopefully I got it right.

[ EDIT ]

So far all is Ok.
Antec is " giving " me a pseudo replacement.
( not fully modular as it should be .. no warranty )
Even though I was only 3 years into a 5 year warranty, they have no parts and it's discontinued.
 
Last edited:

Noid

Platinum Member
Sep 20, 2000
2,283
103
106
Does anyone have recommendations (or something already written up) on testing the stability of your build?
There were old Sticky threads about your Q.
Tons of reading material here, IF YOU LOOK.

P95 is best for finding CPU / memory issues.
( the GOLDEN standard is 12 hours error free in P95 I believe - yes you can use your computer at the same time - but I dont recommend gaming for obvious reasons )
Looping GRAPHICS tools will identify artifacts or driver failures on your GPU under load. ( you will need to use your eyes, and ears find these problems - typically loop longer than any gaming sessions you would have - 8 hours is good )
 
Last edited:
  • Like
Reactions: LikeLinus

LikeLinus

Lifer
Jul 25, 2001
11,213
512
126
The absolute 110% 'is it stable' I feel is prime95, the thread about core cycler is a good reference for testing individual cores in a consistent and orderly manner.

Is there a certain part you feel is not stable?
No, nothing wrong, but I don't want to put it into video production without testing it first. Especially when you're talking about several thousand dollars worth of parts, I'd rather test everything to ensure stability in case something pops-up and it's past the point of returning items. Warranty is a pain sometimes.

I'll search at the Core Cycler thread. Thanks for the feedback.
 

LikeLinus

Lifer
Jul 25, 2001
11,213
512
126
There were old Sticky threads about your Q.
Tons of reading material here, IF YOU LOOK.
I realize that, but as you said "old" and "tons of reading material". This is something I'm trying to get done ASAP to deploy in our production setup. It's very time consuming to comb through tons of threads. That's why I was hoping there was something pre-written with a good step-by-step to follow.

I'll stick with Prime95 w/ the recommended 12hr stability goal. Thank you!
 

Noid

Platinum Member
Sep 20, 2000
2,283
103
106
I'll stick with Prime95 w/ the recommended 12hr stability goal.
Sounds like you haven't used it in awhile .. you need to keep an eye on it.
I use the " Large FFT " test. ( recommended for x570 IIRC )
P95 does not show a " red " indicator on failure now.
You need to open the main window for visual on a thread failure.
( if the CPU utilization is not 100% , one or more threads have failed )
 
Last edited:

LikeLinus

Lifer
Jul 25, 2001
11,213
512
126
Sounds like you haven't used it in awhile .. you need to keep an eye on it.
It does not show a " red " indicator on failure now.
You need to open the main window for visual on a thread failure.
( if the CPU utilization is not 100% , one or more threads have failed )
No, I actually haven't. Appreciate the note and heads-up. I'll install it later this afternoon and see how it goes.
 

DrMrLordX

Lifer
Apr 27, 2000
17,652
6,646
136
Prime95, LinPACK, and other fully AVX-optimized benchmarks have to be used in proper context nowadays. The last CPU I owned where those could be used promiscuously was my 1800x since AVX didn't put it in a seperate stability/power use category. Now if you're running stock then use, Prime95 smallFFTs for CPU, Prime95 blend with tuned FFT sizes for overall (or y-cruncher 2-3 times in row), and maybe one of the RAM stability testers for the RAM alone. If you are using a static overclock then depending on your CPU, you need to set stability targets based on whether you plan on being stable during AVX workloads. Like I have a static OC for my 3900x that is game-stable vs a slightly different one for CineBench R20/R23 or Blender Benchmark, and then another one for Prime95 SmallFFTs or y-cruncher.
 
  • Like
Reactions: LikeLinus and Noid

cytg111

Lifer
Mar 17, 2008
16,677
6,490
136
So here I am assembling some of the parts and to my surprise, this

"Noctua NH-U12S redux"

Seems to come with pre-applied thermal paste.

I had already digged out some old arctic silver from the archives...

My gut instinct is to use what is already applied and nothing else. Right?
Do you "twist it a little" when you put it on?
 

Gideon

Golden Member
Nov 27, 2007
1,429
2,875
136
Memory Info (Check rank)
○ MemTest86
○ AIDA64
○ Thaiphoon Burner
○ TestMem5
○ ZenTiming
I'd alsoo add Karhu memtest to the list:

You have to buy it. But it runs well in windows with other stuff active, finds errors very very quickly compared to most other software (I've had "stable" memory pass a night of Memtest86 and fail in the first Karhu run in 5 minutes).

While this is highly anecdotal I've personally had 0 memory related issues with custom timings, once it passed even 1 Karhu run to 100%
 

Leeea

Golden Member
Apr 3, 2020
1,044
1,240
96
Jfc, I got the plus, cause who needs the wifi right? Anyway, the plus cant flash without a cpu and is NOT 5000 ready out of the box… grrrrrr
Literally hit the wall 15 mins ago.
ouch! seriously painful turn of events!

No shame in RMAing that thing if you do not have a CPU around to flash it with. The difference in price between the plus and the pro is $25, which is cheaper then trying to find a cpu to flash it with.

I went to the pro for the ability to flash it without having a CPU in it. That and the internal usb C header for my case.
 
Last edited:

LikeLinus

Lifer
Jul 25, 2001
11,213
512
126
I ran Cinebench R23 and my score was 27K (multi). Is that a good score or should I try to overclock my 5950x? I have everything stock right now. 5950x w/ 64GB ram and a 3080 Asus GPU.
 
Last edited:

Dave3000

Golden Member
Jan 10, 2011
1,078
41
91
Ever since July my 5800x has been hitting 90C within 4-5 seconds while running CB20. While it stayed at 90C, core clock speeds eventually throttled down to 4.4-4.45 GHz lately. I'm getting around 5930-5980 on the multicore run as the score. When I first got this CPU back in November, CPU core temperatures were between 84-87C and CB20 score was about 100 points higher on the multicore run. Also lately OCCT within 4 seconds of running the Linepack test, the core package hit 90C on my 5800x. Back in the winter and spring of this year it hit 85-86C in the same test and too much longer to hit those temperatures than it takes to hit 90C now. Room temperature was around 26C. Also after removing the front panel from my Meshify C case, and running CB20 again, my 5800x still hit 90C within the first run, and that was with two CoolerMaster Silent 120mm case fans as intake and one as rear exhaust. I went back to the original fans and case fan configuration that came with the case and that was one intake around the middle of the front and one rear exhaust at that, no improvement that I could see in CPU temperatures. I'm using a Noctua U12A CPU cooler with the stock NH1 thermal paste and used the pea-size blob in the center method.
 

Makaveli

Diamond Member
Feb 8, 2002
4,335
522
126
Ever since July my 5800x has been hitting 90C within 4-5 seconds while running CB20. While it stayed at 90C, core clock speeds eventually throttled down to 4.4-4.45 GHz lately. I'm getting around 5930-5980 on the multicore run as the score. When I first got this CPU back in November, CPU core temperatures were between 84-87C and CB20 score was about 100 points higher on the multicore run. Also lately OCCT within 4 seconds of running the Linepack test, the core package hit 90C on my 5800x. Back in the winter and spring of this year it hit 85-86C in the same test and too much longer to hit those temperatures than it takes to hit 90C now. Room temperature was around 26C. Also after removing the front panel from my Meshify C case, and running CB20 again, my 5800x still hit 90C within the first run, and that was with two CoolerMaster Silent 120mm case fans as intake and one as rear exhaust. I went back to the original fans and case fan configuration that came with the case and that was one intake around the middle of the front and one rear exhaust at that, no improvement that I could see in CPU temperatures. I'm using a Noctua U12A CPU cooler with the stock NH1 thermal paste and used the pea-size blob in the center method.
Is PBO on?

if so are you set to motherboard limits or using other settings for

PPT
TDC
EDC
 

Dave3000

Golden Member
Jan 10, 2011
1,078
41
91
PBO is off. I even manually disabled it in the BIOS instead of leaving it on Auto, which made no difference. I have an Asus B550-F ROG Strix (Wi-Fi) motherboard updated with the latest official BIOS for this board. Also based on what I've been reading online, I should be getting around 6050-6100 at stock without PBO anyways in CB20's multicore on a 5800x.
 

Sick Willie

Member
Apr 8, 2010
166
13
81
So here I am assembling some of the parts and to my surprise, this

"Noctua NH-U12S redux"

Seems to come with pre-applied thermal paste.

I had already digged out some old arctic silver from the archives...

My gut instinct is to use what is already applied and nothing else. Right?
Do you "twist it a little" when you put it on?
Yes, the stuff they use is pretty good. As far as twisting, whatever you want to do. I just make sure it's seated at the right angle (perpendicular, lol).
 
  • Like
Reactions: cytg111

cytg111

Lifer
Mar 17, 2008
16,677
6,490
136
Jfc, I got the plus, cause who needs the wifi right? Anyway, the plus cant flash without a cpu and is NOT 5000 ready out of the box… grrrrrr
Literally hit the wall 15 mins ago.
Nope, already upflashed, there is a sticker next to the cpu socket that says what it is flashed with @production.. and im good ... full steam ahead :):)
 
  • Like
Reactions: scineram and Leeea

Noid

Platinum Member
Sep 20, 2000
2,283
103
106
used the pea-size blob in the center method.
Look to see what the top of the chip and the bottom of your cooler look
like after a removal ( or re-application )
The paste should have covered the entire surface evenly.

I prep my chip and HS base with Isopropyl alcohol.
Compressed air to blow dust.
I have been putting 4 little blobs near center for each corner and one in center.
( these chips have alot of surface )
Then, use a tooth pick to smear out flat to the edges.
( use something that will not leave debris - smear it like your plastering )
I visually look for no metal showing, then place it in the mb.
Just be careful not to add to much paste.
( just remove excess, and smooth out again )

My Noctua's are very sensitive to dust.
( max fan rotation is 1500rpm and my idle is 1450 )
Cleaning is a monthly occurrence for me.
( removing the fans to clean the blades also )

Some have reported that Noctua's have bent at the base.
They are flexible, and need to be handled carefully.
 
Last edited:

Leeea

Golden Member
Apr 3, 2020
1,044
1,240
96
used the pea-size blob in the center method
I always dump it all on! Noctua supplied a tube with my heatsink, and I dumped the entire tube on.

Ever since I realized the difference in temperature between perfect and way to much was about 1C, thanks to Gamers Nexus, I just liberally apply way to much on.

I think it lasts longer to that way, but I have no proof of that.
 

Leeea

Golden Member
Apr 3, 2020
1,044
1,240
96
Where's the GPU?
Lucky me, I already had one :)

Here is a picture of it stuffed together with the GPU. As you can see, it is a bit close to the CPU heatsink. Next time I am buying a smaller CPU heatsink, this thing is way to big. I may have also installed it upside down ...

bitClose.jpg

By far and away, my biggest problem so far is I have not been able to sync the RGB on the GPU with the RGB on the mainboard. It is a total clash of colors! AMD uses coolermasters software to control the RGB, and it does not see the mainboard. On the other hand, ASUS controls with the crate software, and it does not see the GPU RGB. It is a "disaster" :).
 
Last edited:
  • Like
Reactions: Kenmitch

bigboxes

Lifer
Apr 6, 2002
33,134
11,389
146
Lucky me, I already had one :)

Here is a picture of it stuffed together with the GPU. As you can see, it is a bit close to the CPU heatsink. Next time I am buying a smaller CPU heatsink, this thing is way to big. I may have also installed it upside down ...

View attachment 47502

By far and away, my biggest problem so far is I have not been able to sync the RGB on the GPU with the RGB on the mainboard. It is a total clash of colors! AMD uses coolermasters software to control the RGB, and it does not see the mainboard. On the other hand, ASUS controls with the crate software, and it does not see the GPU RGB. It is a "disaster" :).
Do what I did. Get a windowless case.
 

Dave3000

Golden Member
Jan 10, 2011
1,078
41
91
Look to see what the top of the chip and the bottom of your cooler look
like after a removal ( or re-application )
The paste should have covered the entire surface evenly.

I prep my chip and HS base with Isopropyl alcohol.
Compressed air to blow dust.
I have been putting 4 little blobs near center for each corner and one in center.
( these chips have alot of surface )
Then, use a tooth pick to smear out flat to the edges.
( use something that will not leave debris - smear it like your plastering )
I visually look for no metal showing, then place it in the mb.
Just be careful not to add to much paste.
( just remove excess, and smooth out again )

My Noctua's are very sensitive to dust.
( max fan rotation is 1500rpm and my idle is 1450 )
Cleaning is a monthly occurrence for me.
( removing the fans to clean the blades also )

Some have reported that Noctua's have bent at the base.
They are flexible, and need to be handled carefully.
Well, yesterday I went and bought some thermal paste, the Corsair XTM50. I was using the Noctua NH1 thermal paste before this one. No difference in CPU temperatures during full load. When I checked the old thermal paste application on the CPU after removing the heatsink before I was going to apply the new thermal paste I just bought, there was barely any thermal paste in the center of the CPU and some of it ended up near the edges of the heat spreader of the CPU. Also most of the thermal paste ended up on the heatsink instead of the on CPU upon inspection. Also the thermal paste had several cracks on it and looked kind of dried up upon inspection. This time I applied a bigger glob of thermal paste than the previous application onto the center of the head spreader of the CPU, hoping that it would improve CPU temperatures, because I read that the CCD and IO die is not in the center of these Ryzen CPUs and a pea size in the center of the heat spreader might not be enough to cover the area of these dies, so maybe I thought putting a bigger blob this time would solve these temperature issues with my CPU, which really didn't afterall. I even tightened up the CPU's heatsink brackets tighter onto my motherboard as the screws still had some play left in them before I went to apply the new thermal paste. I also made sure that the heatsink was screwed on tightly onto these brackets so there was no play left in the screws. The only thing I can think of that I have left to try is to purchase a 360mm AIO, such as the H150i and front mount it as my case will only accept a 360mm AIO if it's front mounted, but that's such a big expense and hassle especially if I already purchased a Noctua U12A specifically for this CPU and an expensive tube of thermal paste.
 
Last edited:

ASK THE COMMUNITY

TRENDING THREADS