My experience with the Asus PRIME X370-Pro - Page 955 - Overclock.net - An Overclocking Community

Forum Jump: 

My experience with the Asus PRIME X370-Pro

Reply
 
Thread Tools
post #9541 of 9614 (permalink) Old 09-15-2019, 01:09 PM
New to Overclock.net
 
makatech's Avatar
 
Join Date: Sep 2017
Posts: 250
Rep: 0
Quote: Originally Posted by Acidstorm View Post
That's the plan. It may not be worth it, but once I get things I'm going to see what I can do with the board and keep track of performance. Gonna see what my Corsair Vengeance memory can do as well. I actually have it running at 3000, when the previous BIOS (42xx) I forget the specific number, but I hadn't updated since December, it seems to really made a difference. Until that point, I could only get it to OC to 2800 stable. My RAM is 2400 Corsair Vengeance, with Hynix MFR chips, so it was by no means a good overclocker. Now that RAM is sanely priced again, I want to get some 3600 RAM, possibly that new Trident Neo stuff, but I am not sure if the X370 board is going to handle that or not with my R5 1600. It would be interesting to see what happens when I put a 3000 series Ryzen in there and see if the memory controller is any better than my 1600. I haven't had a chance to test any high speed RAM, it's possible my 1600 has a decent memory controller, and I just don't know it.

The other reason I was going to upgrade the board was so that I could take this platform and kick it down to my younger brother. He has an old Q6600 CPU, and I already gave him an RX 580 GPU. I had two RX 580 GPU in X-fire, and both died. Both were XFX boards, something fried in the circuit board traces. I could see dark spots on the board, was very odd because I kept the cards in a well ventilated chassis. While they were on RMA, I decided to pick up an RTX 2060, because I wasn't entirely sure XFX would honor the warranty on the cards or not. Anyhow, they sent me 2 brand new XFX RX 580s and gave one to my bro and the other is still sealed in the box. Once I set him up with this mobo I could set up x-fire for him, but I don't think it's worth it these days. Most games avoid it don't enable it, or optimize for it, so probably best to just keep as a spare.
Being nice to your brother is a very good thing. :-)

For me the difference is big between 1600X and 3700X on this board but I do have Samsung B-die sticks, currently running 3600MT/s. I had major issues getting anything faster than 3200MT/s stable for my 1600X. Ryzen 3 series suddenly makes this board a much better option than before.

Last edited by makatech; 09-16-2019 at 04:18 AM.
makatech is offline  
Sponsored Links
Advertisement
 
post #9542 of 9614 (permalink) Old 09-16-2019, 07:20 PM
New to Overclock.net
 
mat9v's Avatar
 
Join Date: May 2014
Posts: 416
Rep: 10 (Unique: 8)
Quote: Originally Posted by makatech View Post
Quote: Originally Posted by mat9v View Post
My thoughts on ZEN2 [IMG class=inlineimg]/forum/images/smilies/smile.gif[/IMG]
I bought 3900X and after literally waiting 2 months for it I was finally able to install it. In the meantime I got X570 Prime Pro because.... well I had money waiting for a CPU ant it was late to the party [IMG class=inlineimg]/forum/images/smilies/wink.gif[/IMG] ... but anyway from what I observed VRM is not an issue in any way whatsoever. From the beginning:
- I have an Alphacool Eisbaer 420 cooler on it mounted on liquid metal in push-pull config
- CPU on latest beta bios is able to boost to 4.6Ghz on 1-2 cores, but only for a moment and on auto CB20 reaches about 7100p (on older around 6900p)
- but it can be increased to 7650p with manual OC
- but in this ZEN2 is really problematic, for example, AMD lied that AVX2 does not require any offset compared to AVX or simple Float operations, and so while I can run CB20 in a loop or CPU-Z benchmark or anything really at [email protected] and keep temps at most 82C any attempt to run Prime95 AVX2 code at that speed results in instant 95C and reboot a second later (thank you Asus for the reboot or my CPU would probably by fried), to run AVX2 code I have to lower clock to 4.1-4.2Ghz on 1.15-1.2V to keep temps below 85C and 92C respectively
- and no, my cooling solution is not inadequate (I think) since the temp rise is almost instantaneous
- and I can keep regular temps below 80C for over 220W reported from CPU
- memory on ZEN2 (and X570) is working great, my old modules that refused to go faster than 3000 on 2700X on X370 are working easily at 3733 in synchro mode with Infinity Fabric

Reading reviews that state that they pushed 3900X to 4.4Ghz on 1.4V makes me wonder how they cooled it or what they were running on it.
What all above was listed for? Well, I could never push even 250W into the CPU, which was possible for 2700X, without temps going through the roof so no worries about 3900X running on X370, frankly at stock it only uses 145W of power and up to 190W with PBO enabled and scalar 10X forced - in the same ballpark as 2700X with PBO.
You wrote: "any attempt to run Prime95 AVX2 code at that speed results in instant 95C".

My question: Was this in stock or running your system overclocked?
Overclocked to 4.3Ghz of course. At stock it would just downclock correctly 🙂
What I was pointing out was that at 4.3Ghz Prime95 with AVX2 is an instant 95C for my setup while it can run Cinebench R20 or other tasks at full CPU utilization without any problems and temps at most 82C. I do not know what was the power use in that moment that the benchmark started and I'm not inclined to try again 🙂
So there is a difference between loads and AMD was not correct when they pointed out that while Intel requires an offset to run AVX2, Ryzen does not need it.
Well based on how bad ZEN2 is with keeping or even reaching stated clocks.... just read Tom's Hardware article and watch GamersNexus and HardwareUnboxed videos on AGESA 1.0.0.3ABBA "fix" and you will understand. The clocks AMD claims their CPU reach are almost always at idle, and while ABBA brought some single core increases it is still far from sustained turbo clock as found in Intel CPUs.
Not that it matters for me anyway 🙂 as performance of 3900X is more then sufficient for the price I paid.

mat9v is offline  
post #9543 of 9614 (permalink) Old 09-17-2019, 12:56 AM
New to Overclock.net
 
Join Date: Nov 2018
Posts: 33
Rep: 2 (Unique: 1)
5220 beta is spicy! My 3600 is now basically a budget 3600X with +200 MHz auto oc. It would literally be 4.4 GHz if we could get BCLK from 99.8 to 100. A real 3600x probably performs much better in Cinebench but we can pretend. Also my AIDA benchmark is slightly different than my last one here.
https://www.overclock.net/forum/28098712-post9415.html

I didn't run many tests so don't jump to conclusions. The bios update reset all my drivers for some reason and they were being initialized in the middle of run 1 so that's probably why the scores are so low. I'm going to look into turning the OC features on and off because my scores look low.

Cinebench R20
BIOS 5204
Run 1 3562 M / 485 S

BIOS 5220
Run 1 3561 M / 482 S (scores were low because drivers were not loaded properly)
Run 2 3584 M / 487 S


Here's a history of my 3600's boosts with the different bios releases.
5008 - ~4.35 GHz
5204 - ~4.3 GHz
5220 - ~4.4 GHz

Maximum voltage increased from 1.45v to 1.475v so you could say AMD is cheating by throwing more volts.
Attached Thumbnails
Click image for larger version

Name:	clock.png
Views:	8
Size:	103.9 KB
ID:	296508  

Click image for larger version

Name:	cachemem.png
Views:	20
Size:	292.6 KB
ID:	296510  


Last edited by Futsal; 09-17-2019 at 10:44 AM.
Futsal is offline  
Sponsored Links
Advertisement
 
post #9544 of 9614 (permalink) Old 09-17-2019, 07:18 PM
New to Overclock.net
 
mat9v's Avatar
 
Join Date: May 2014
Posts: 416
Rep: 10 (Unique: 8)
Ok, with fresh install on NVMe RAID0
With all settings stock (except memory at 3733 and IF forced to 1:1 at 1866) the 3900X is finally boosting somewhere close to 4.6Ghz with low voltages even, below 1.5V. With CB20 single core test running I observer correlated core load and frequency at about 4525-4550 on cores that the task was running on so while HWiNFO happily shows 4.6Ghz+ it was not while under CB20 load. Not that it is a problem since performance was satisfactory, but it is the numbers that irritate
Note to anyone trying to set up bootable RAID0 on NVMe - stupid AMD made separate drivers for NVMe RAID and SATA RAID so there is no way to make a system backup, create RAID and restore it to the new drive. Even if the restore part works (and it is a problem since most utilities, be it based on Linux or Windows, do not allow adding drivers by hand to bootable image), Windows ends up unbootable after restore since it does not have correct drivers for NVMe RAID. Hence clean install for me after many hours of fighting with AMD stupidity.
Another note, having SATA RAID0 active while trying to create NVMe RAID results in all SATA drives changing to eSATA and invalidating RAID config for SATA drives. Guess what? Lost data. Fortunately there are RAID recovery tools that correctly find out which drive is which in RAID0 sequence and allow for data recovery from destroyed array. But the process is painfully slow, and will take me about 10 hours to recover 3TB of data.
I know that this is not a standard config with multiple RAID arrays but why did AMD make it so painful to use?
Watching maximum and average CPU frequencies over few hours (and installing many programs, browsing, downloading, benchmarking etc) it looks that 3900X does have different quality chiplets, meaning one high quality and one low quality, after over 10 hours I have not seen secondary chiplet ever reach over 4.4Ghz while the primary one does up to 4.625Ghz (46.3 multiplier), I suppose that it makes no difference to end user since most tasks will run happily on CCD0 at up to 4.6Ghz and only heavy multithreaded tasks will use CCD1 along with CCD0 and that kind of load will force CPU to lower clocks below 4.4Ghz anyway. It's easy to see when even CB20 with almost 200W power use lowers CPU frequency to about 4.15-4.2Ghz under full load.
I will be checking different all core loads in regard to undervolting with offset (taking into account known issues with it and lowering performance) trying to find out a sweet spot between undervolting to keep in power envelope, temperature, stability and performance. Lots of fun before me
Attached Thumbnails
Click image for larger version

Name:	3900X - stock benchmarks.jpg
Views:	56
Size:	1.35 MB
ID:	296646  


mat9v is offline  
post #9545 of 9614 (permalink) Old 09-19-2019, 02:47 AM
New to Overclock.net
 
makatech's Avatar
 
Join Date: Sep 2017
Posts: 250
Rep: 0
Quote: Originally Posted by mat9v View Post
Ok, with fresh install on NVMe RAID0
With all settings stock (except memory at 3733 and IF forced to 1:1 at 1866) the 3900X is finally boosting somewhere close to 4.6Ghz with low voltages even, below 1.5V. With CB20 single core test running I observer correlated core load and frequency at about 4525-4550 on cores that the task was running on so while HWiNFO happily shows 4.6Ghz+ it was not while under CB20 load. Not that it is a problem since performance was satisfactory, but it is the numbers that irritate
Note to anyone trying to set up bootable RAID0 on NVMe - stupid AMD made separate drivers for NVMe RAID and SATA RAID so there is no way to make a system backup, create RAID and restore it to the new drive. Even if the restore part works (and it is a problem since most utilities, be it based on Linux or Windows, do not allow adding drivers by hand to bootable image), Windows ends up unbootable after restore since it does not have correct drivers for NVMe RAID. Hence clean install for me after many hours of fighting with AMD stupidity.
Another note, having SATA RAID0 active while trying to create NVMe RAID results in all SATA drives changing to eSATA and invalidating RAID config for SATA drives. Guess what? Lost data. Fortunately there are RAID recovery tools that correctly find out which drive is which in RAID0 sequence and allow for data recovery from destroyed array. But the process is painfully slow, and will take me about 10 hours to recover 3TB of data.
I know that this is not a standard config with multiple RAID arrays but why did AMD make it so painful to use?
Watching maximum and average CPU frequencies over few hours (and installing many programs, browsing, downloading, benchmarking etc) it looks that 3900X does have different quality chiplets, meaning one high quality and one low quality, after over 10 hours I have not seen secondary chiplet ever reach over 4.4Ghz while the primary one does up to 4.625Ghz (46.3 multiplier), I suppose that it makes no difference to end user since most tasks will run happily on CCD0 at up to 4.6Ghz and only heavy multithreaded tasks will use CCD1 along with CCD0 and that kind of load will force CPU to lower clocks below 4.4Ghz anyway. It's easy to see when even CB20 with almost 200W power use lowers CPU frequency to about 4.15-4.2Ghz under full load.
I will be checking different all core loads in regard to undervolting with offset (taking into account known issues with it and lowering performance) trying to find out a sweet spot between undervolting to keep in power envelope, temperature, stability and performance. Lots of fun before me
Congratulations, great scores and a powerful computer. :-) Maximum singlecore score for me (so far) in CB 20 on the Asus Prime X370 Pro board is 509 I think (3700X).

I believe this is my highest score in Geekbench 5 (stable setup):
https://browser.geekbench.com/v5/cpu/148720

3700X brought new life into my "old" Asus Prime X370 Pro board, that's for sure. I have seen slightly better scores for other 3700X & X370 but they have slightly better motherboard, ram and higher gpu clock than me.

Last edited by makatech; 09-19-2019 at 06:27 AM.
makatech is offline  
post #9546 of 9614 (permalink) Old 09-19-2019, 04:28 AM
New to Overclock.net
 
Join Date: Nov 2018
Posts: 33
Rep: 2 (Unique: 1)
Btw guys there's a piece of shrink wrap plastic on top of the chipset heatsink. I didn't notice until I saw that German website it's practically invisible. You should peel it off if you haven't.
Attached Images
 
Futsal is offline  
post #9547 of 9614 (permalink) Old 09-19-2019, 12:39 PM
New to Overclock.net
 
Join Date: Dec 2018
Posts: 7
Rep: 0
Thanks, I missed that too.

Tested bios 5216: 3600CL16 is almost stable, memtesthelper2 works few minutes but the performance in benchmarks is a lot worse than 3466CL14. Ryzen 2700x.
A_free is offline  
post #9548 of 9614 (permalink) Old 09-20-2019, 05:37 AM
New to Overclock.net
 
Facepunch's Avatar
 
Join Date: Mar 2014
Posts: 1
Rep: 0
I'm gonna need an adult to chime in here.

Just upgraded/sidegraded from my r7 1700 to an r5 3600, and i can not for the life of me get PBO+Auto-OC to work at all.
I've tried both UEFI version 5216 and the 5220 beta, and no matter what i do, it won't go above a 41.75 multiplier at all (and this is NOT sustained at all). Cinebench sustains between 40.5 and 40.75 mult.

I have set up the OC settings in all of the 2 or 3 different places it's located in the UEFI, and ryzen master correctly reports the maximum boost clock as 4400, as well as the raised PPT,TDC & EDC (395W, 255A & 255A iirc, not at main PC atm). Also tried using all of the different power plans in windows, makes not difference.

I can set a manual OC at 1.35V for 42.5 mult and it's stable in non-AVX scenarios, but this is obviously not a useable OC. (And lets not pretend 3000 series doesn't heavily downclock under small FFTs in prime, mine goes to around 39 mult under PBO+Auto-OC).

Any clues as to what i can do? did i just receive **** silicon, or does anyone else experience the same?

Spent quite a few hours bashing my head against this wall, without any results.

On another note, my LPX 3200MHz Hynix AFR memory is running great OOTB using XMP settings, haven't tried tightening the timings just yet.
Facepunch is offline  
post #9549 of 9614 (permalink) Old 09-20-2019, 08:15 AM
New to Overclock.net
 
mat9v's Avatar
 
Join Date: May 2014
Posts: 416
Rep: 10 (Unique: 8)
Quote: Originally Posted by Facepunch View Post
I'm gonna need an adult to chime in here.

Just upgraded/sidegraded from my r7 1700 to an r5 3600, and i can not for the life of me get PBO+Auto-OC to work at all.
I've tried both UEFI version 5216 and the 5220 beta, and no matter what i do, it won't go above a 41.75 multiplier at all (and this is NOT sustained at all). Cinebench sustains between 40.5 and 40.75 mult.

I have set up the OC settings in all of the 2 or 3 different places it's located in the UEFI, and ryzen master correctly reports the maximum
boost clock as 4400, as well as the raised PPT,TDC & EDC (395W, 255A & 255A iirc, not at main PC atm). Also tried using all of the different power plans in windows, makes not difference.

I can set a manual OC at 1.35V for 42.5 mult and it's stable in non-AVX scenarios, but this is obviously not a useable OC. (And lets not pretend 3000 series doesn't heavily downclock under small FFTs in prime, mine goes to around 39 mult under PBO+Auto-OC).

Any clues as to what i can do? did i just receive **** silicon, or does anyone else experience the same? [IMG class=inlineimg]/forum/images/smilies/smile.gif[/IMG]

Spent quite a few hours bashing my head against this wall, without any results.

On another note, my LPX 3200MHz Hynix AFR memory is running great OOTB using XMP settings, haven't tried tightening the timings just yet.
Check CPU-Z if it shows higher clocks. In my case HWiNFO for some reason only occasionally show anything higher then 4.4Ghz while CPU-Z has no problem with showing even over 4.6Ghz. There is something strange with HWiNFO startup options I think but I can't figure it out at the moment.

mat9v is offline  
post #9550 of 9614 (permalink) Old 09-20-2019, 09:10 PM
New to Overclock.net
 
chevalier433's Avatar
 
Join Date: Jan 2014
Posts: 1
Rep: 0
Quote: Originally Posted by Futsal View Post
Btw guys there's a piece of shrink wrap plastic on top of the chipset heatsink. I didn't notice until I saw that German website it's practically invisible. You should peel it off if you haven't.
Thanks bro,thats why my chipset temps was 65-70,i used the motherboard with this **** on for some months now during the summer period,and im on 4.05 on ryzen 1600x,i hope i didnt cause any degradation.
chevalier433 is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off