AB350 Gaming-iTX/ac - freezing (oh no...)

Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic

   First of all- Good info here.

Let me share my experiences so far. I have played around with
my new computer for a week now and I must say That it has been a bit
challenging but also a time for learning since I must say it is not exactly plug
and play at the moment :-)

For the new ones reading I wanna confirm that the windows
version on install is very important! I wasted a lot of time there :-)

First of all- I can confirm the cold boot bug, where it
loose all the settings.

I can also confirm the bug mentioned that sometimes the CPU
is not able to downclock but keep pushing on at a constantly high level after
being challenged (Have especially seen that in relation to 3D-Mark.

I have been messing around quite abit and I am not done yet
:-) but let me share a few of my experiences so far.

I dont OC the CPU at all anymore (above the 3600 Mhz) because
this breaks the whole boost funktionality- So unless you OC to 4000 Mhz you
will only get higher base power consumption and less peak Mhz! That basically
sucks. 

An interesting thing however that I have noted is that the
standard voltage that the BIOS set for the CPU voltage seams to be quite high.
many people have been scared about that and have clocked it down. I have
noticed however, that Ryzen master also give it a high default value (3,5 V. if
I remeber correctly). In the example with screenshot on AMD's homepage with the
2400g on ryzen master the CPU Voltage is put to 1,4 V for an 4000Mhz overclock.
I suspect that this chip with the boost funktionality need a somewhat higher
CPU Voltage to accomodate for the boosts than what many expect. 

I have played alot with the ram settings and the APU because
I really wanted my ram to run with the 3466 Mhz that is the maximum for my
board. I have managed to make it post and I can also finish some tests, but it
is not stable in games. I really want to try and OC the ram as much as possible
because the performance from this is much better than increasing the clock on
the APU-GFX relatively. Besides you dont create the same temperature issues with
high ram speeds compared to OC the APU-GFX. I tried to increase the timings
without much succes- Just some trial and error. If anyone have some experience
to share please do. My ram are G. SKill Ripjaws V. 2X8 3600Mhz. 16-16-16-36 (F4-3600C16D-16GVK).

As a note I measure my temperature with CPUID Hardware
monitor and I have 1 crazy temperature 115 C on one of the measurements (cannot
remember the exact naming bu T??something) but I guess it is some bug in the
software - Can anyone confirm this? The other ones are around 40 +/-.

I had my ram clocked to 1,4v. (Back to 1.35V. again). but
more interesting I have had the VDDCR SOC clocked to 1,28.. V - I know that many
will consider this crazy high- which is probably true, but since that SOC feeds
both the RAM and the APU I wanted to try and see if I could get it more stable
by raising it a bit. The reason for this was, that I found it very difficult to
OC the APU higher than 1350 Mhz. So I wanted to see if the issue was that the
RAM took all the juice. I could get some quite good numbers in benchmark, but
it was not stable so now I have clocked the SOC and RAM down a bit again.

Now I am back to 1,3V. for the CPU (Not OC because I want
the boost funktionality and keep the power saving when idle) I have the ram
running at 3333 Mhz with 1,35V. The VDDCR SOC is 1,2 and the APU GFX is 1300
Mhz. After reading Cristys last comment I will also try and see what happens if
I lower the voltage a bit for the ram- Maybe to 1.25V or 1.30V.

It is however not very stable. I can play Crysis 2007 good
on medium settings 1920X1200 for about 5-10 minutes and then I get a windows
popup with the program has to close down??

If I try to raise the APU GFX to 1400Mhz it will crash by
some grafic mess screen or freeze- So I am thinking that I need to lover the
RAM more, increase the VDDCR SOC or maybe I am getting to hot. Any ideas?

Can anyone recommend some software to measure temps within
games and tests? When I rush out of a game or test to see the temperature
measurements it has already cooled down a lot, so not easy to really see. I
have tried the MSI afterburner  but I
cannot make it show the right sensor- Just shows 0 degree.  

Sorry for the long post but I am not native english speaking
so I am excused for not being able to keep it short and exact :-)

Author	Message Topic Search Topic Options Post Reply Create New Topic Printable Version Translate Topic
Mobawe Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 16 Mar 2018 Location: Denmark Status: Offline Points: 13	Post Options Post Reply Quote Mobawe Report Post Thanks(0) Quote Reply Posted: 20 Mar 2018 at 11:17pm
	First of all- Good info here. Let me share my experiences so far. I have played around with my new computer for a week now and I must say That it has been a bit challenging but also a time for learning since I must say it is not exactly plug and play at the moment :-) For the new ones reading I wanna confirm that the windows version on install is very important! I wasted a lot of time there :-) First of all- I can confirm the cold boot bug, where it loose all the settings. I can also confirm the bug mentioned that sometimes the CPU is not able to downclock but keep pushing on at a constantly high level after being challenged (Have especially seen that in relation to 3D-Mark. I have been messing around quite abit and I am not done yet :-) but let me share a few of my experiences so far. I dont OC the CPU at all anymore (above the 3600 Mhz) because this breaks the whole boost funktionality- So unless you OC to 4000 Mhz you will only get higher base power consumption and less peak Mhz! That basically sucks. An interesting thing however that I have noted is that the standard voltage that the BIOS set for the CPU voltage seams to be quite high. many people have been scared about that and have clocked it down. I have noticed however, that Ryzen master also give it a high default value (3,5 V. if I remeber correctly). In the example with screenshot on AMD's homepage with the 2400g on ryzen master the CPU Voltage is put to 1,4 V for an 4000Mhz overclock. I suspect that this chip with the boost funktionality need a somewhat higher CPU Voltage to accomodate for the boosts than what many expect. I have played alot with the ram settings and the APU because I really wanted my ram to run with the 3466 Mhz that is the maximum for my board. I have managed to make it post and I can also finish some tests, but it is not stable in games. I really want to try and OC the ram as much as possible because the performance from this is much better than increasing the clock on the APU-GFX relatively. Besides you dont create the same temperature issues with high ram speeds compared to OC the APU-GFX. I tried to increase the timings without much succes- Just some trial and error. If anyone have some experience to share please do. My ram are G. SKill Ripjaws V. 2X8 3600Mhz. 16-16-16-36 (F4-3600C16D-16GVK). As a note I measure my temperature with CPUID Hardware monitor and I have 1 crazy temperature 115 C on one of the measurements (cannot remember the exact naming bu T??something) but I guess it is some bug in the software - Can anyone confirm this? The other ones are around 40 +/-. I had my ram clocked to 1,4v. (Back to 1.35V. again). but more interesting I have had the VDDCR SOC clocked to 1,28.. V - I know that many will consider this crazy high- which is probably true, but since that SOC feeds both the RAM and the APU I wanted to try and see if I could get it more stable by raising it a bit. The reason for this was, that I found it very difficult to OC the APU higher than 1350 Mhz. So I wanted to see if the issue was that the RAM took all the juice. I could get some quite good numbers in benchmark, but it was not stable so now I have clocked the SOC and RAM down a bit again. Now I am back to 1,3V. for the CPU (Not OC because I want the boost funktionality and keep the power saving when idle) I have the ram running at 3333 Mhz with 1,35V. The VDDCR SOC is 1,2 and the APU GFX is 1300 Mhz. After reading Cristys last comment I will also try and see what happens if I lower the voltage a bit for the ram- Maybe to 1.25V or 1.30V. It is however not very stable. I can play Crysis 2007 good on medium settings 1920X1200 for about 5-10 minutes and then I get a windows popup with the program has to close down?? If I try to raise the APU GFX to 1400Mhz it will crash by some grafic mess screen or freeze- So I am thinking that I need to lover the RAM more, increase the VDDCR SOC or maybe I am getting to hot. Any ideas? Can anyone recommend some software to measure temps within games and tests? When I rush out of a game or test to see the temperature measurements it has already cooled down a lot, so not easy to really see. I have tried the MSI afterburner but I cannot make it show the right sensor- Just shows 0 degree. Sorry for the long post but I am not native english speaking so I am excused for not being able to keep it short and exact :-)
	Ryzen 5 2400g/ ASRock Fatal1ty AB350 Gaming-ITX/ NH-L9A-AM4 L.PROFILE/ INWIN-CHOPIN 150W/ G.Skill Ripjaws V-DDR4,2x8-3600MHz(F4-3600C16D-16GVK)/ Samsung 850 EVO MZ-75E500-SSD-500GB/ UEFI-bios 4.51A

cristy6100 Members Profile Send Private Message Find Members Posts Add to Buddy List Groupie Joined: 15 Mar 2018 Location: Romania Status: Offline Points: 269	Post Options Post Reply Quote cristy6100 Report Post Thanks(0) Quote Reply Posted: 20 Mar 2018 at 6:17pm
	I got a cold boot today too, its the first time it happened, I powered on the PC then it shutdown after 2-3 seconds then it started OK. The thing I changed was lowering the memory voltage from 1.35V to 1.20V, I am using Corsair XMS 2400MHz with Hynix dies which I took up to 3200MHz with no problems @ 1.35V but it seems 1.20V is too low from a cold start so Ryzen memory training kicks in and adjust some timings to ensure a successful boot. Try to disable AM4 Boot memory training and see if it helps, if no boot at all after disabling this, the memory is the problem and you will need to clear the CMOS to restore functionality. Another thing you should do before disabling boot training is setting the memory at 2133/2400MHz with 1.20 or even 1.35V an see if that helps

Thenightstalker Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 03 Mar 2018 Status: Offline Points: 57	Post Options Post Reply Quote Thenightstalker Report Post Thanks(0) Quote Reply Posted: 20 Mar 2018 at 6:24am
	Ans welcome to another round ... Don't get me wrong, I love my PC but some of these issues are killing me! OK, we seem to know why the stuttering appears now I wanted to get rid of the boot issue. It's clearly the RAM, I had it again after cold booting my system today. After that I can set whatever I want, at least as long as I stay in the supported frequency, I tried to set 4000 but that resulted in exactly the same power on/off loop. But this confirms that the problem is the RAM, but I still don't understand why because after that first boot everything is OK, I had my RAM now set to 3200 CL14 which was fine until cold boot today. Now I set it down to 2933 just to see if the Issue has to do with RAM overclocking or if it's simply a bug that appears on every cold boot.

Thenightstalker Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 03 Mar 2018 Status: Offline Points: 57	Post Options Post Reply Quote Thenightstalker Report Post Thanks(0) Quote Reply Posted: 19 Mar 2018 at 9:50pm
	Amen!

cristy6100 Members Profile Send Private Message Find Members Posts Add to Buddy List Groupie Joined: 15 Mar 2018 Location: Romania Status: Offline Points: 269	Post Options Post Reply Quote cristy6100 Report Post Thanks(0) Quote Reply Posted: 19 Mar 2018 at 9:37pm
	I saw about 6-12FPS boost from 1240 to 1500 on the GPU, but until I can OC the igpu in the UEFI and input a normal voltage range (it should be 1.00V up to 1.30V for the iGFX voltage) I will stick to the default clock, you also loose the gpu power states by going manual on the GPU clock, and that is a big shame it should work the same we need to alter only the highest GPU power State from 1240 to any value chosen. If we count to GPU voltage and frequency bug we are up to 9 major bugs so far.... I can add one more bug to those 9 so far: When going manual on the CPU frequency and voltage it should still use the power states not keep the CPU at a fixed frequency and voltage, again the same happens here as on the GPU side, but we are not interested in keeping a fixed voltage and frequency on the CPU/GPU when overclocking, we just want the overclock to apply when needed, that is when the CPU is requesting the highest all core frequency. On Intel side we can adjust the All Core Turbo multiplier, we need to be able to do the same here without stupid complicated P States Tables, And on the GPU side, the GPU clock needs to apply again only to the highest P State of the GPU, that is the highest 3D clock, this works as expected on all AMD GPU's but not here.... Nobody wants a fixed CPU/GPU clock and voltage, but I mean nobody, maybe 1% of users, and again nobody wants to edit the full P States tables, its stupid and applies again to maybe 1% of users We want adaptive voltage and frequency to work when chaiging the CPU frequency and voltage from the standard menu that has been with AMD from the socket AM2 days. Why break something when it worked for the last 10 years? Edited by cristy6100 - 19 Mar 2018 at 10:37pm

Thenightstalker Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 03 Mar 2018 Status: Offline Points: 57	Post Options Post Reply Quote Thenightstalker Report Post Thanks(0) Quote Reply Posted: 19 Mar 2018 at 6:16pm
	I used the WIN 10 drivers first, then reinstalled everything using the vendors drivers, nothing changed so far. I think those drivers are most likely the same. I'll try these voltage settings. I had it running at 1250-1300 before without changing voltage. Hopefully it will go up to about 1600 mhz, that shoud bring me the 5-10 FPS I'm actually missing sometimes.

cristy6100 Members Profile Send Private Message Find Members Posts Add to Buddy List Groupie Joined: 15 Mar 2018 Location: Romania Status: Offline Points: 269	Post Options Post Reply Quote cristy6100 Report Post Thanks(0) Quote Reply Posted: 19 Mar 2018 at 4:24pm
	In Ryzen Master try 1.15 for APU GFX voltage and 1.15 for SOC voltage and fix the gpu to 1240MHz. You should not try to overclock the GFX from UEFI beacause on the ASRock motherboards they do not provide a normal voltage input option, instead they provide a value from 0 to 100 that we can only guess what voltage will provide.... (good job ASRock again) As far as drivers go, uninstall audio/chipset/wlan drivers from AMD/Intel/Realtek, leave the drivers that came with Windows 10, and just install the AMD GPU driver 17.7. I never in the last 6 years had any problems with the drivers that came with Windows for chipset/audio/lan etc etc, but I had numerous problems with the dedicated ones... but anyway those problems we are both reporting have nothing to do with any of the drivers (maybe the GPU driver is causing the throttling, that is a another likely cause, but there is no other driver for the 2200/2400G, at this time) Edited by cristy6100 - 19 Mar 2018 at 4:27pm

Thenightstalker Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 03 Mar 2018 Status: Offline Points: 57	Post Options Post Reply Quote Thenightstalker Report Post Thanks(0) Quote Reply Posted: 19 Mar 2018 at 4:00pm
	Ok another update: I'm constantly getting thise coold power on loop issues now, everytime it seemst to reset my RAM settings and GFX framebuffer. Everytime I go into uefi after this it shows a real low RAM Speed on the left side (don't know which side means what), on the right it shows the set value (3200). And timings seem to change as well but not to XMP default value, I don't understand what's going on there. Also the frambuffer setting is gone each time this happens. You'd think it's just because these settings aren't stable or something but NO, when I set it again and boot everything runs perfect. I played Origins with these settings for more than an hour without any problems. Can anyone tell me about this problem? I rechecked the SATA bug, deactivating definetely does NOTHING. I looked at the drive listing again and yes, there's this AHCI P... string, but it shows only for the 2nd partition of my drive. I have Windows 10 Bootmanager at fist and then AHCI P0 Samsung ... as second drive listed. But that's ONE phyisical drive connected to SATA port 1 on the board. Last but not least some more testing on the stuttering / thortteling bug. Setting any (CPU or GPU) clock speeds manually seems to fix the problem. I tried both yesterday and both worked fine. I monitored the temps during gameplay and it never went much above 60°C (FANs in performance mode). I don't think there's any thermal problem. Drivers: I'm using the latest drivers as far as I know. At least for chipset, GPU, WLAN and AUDIO. Not sure if I did it for LAN because I don't use it. I downloaded them from the vendors page. Can anyone tell me which voltages I need to inrease for GPU overclocking? I tried to raise the GPU clock yesterday but I think it wasn't stable because the game chrashed when I switched between game and ryzen master tool. Maybe it wasn't the GPU but I'll need to raise the voltage anyway if I want to go higher up.

hifihedgehog Members Profile Send Private Message Find Members Posts Add to Buddy List Newbie Joined: 27 Jun 2017 Status: Offline Points: 33	Post Options Post Reply Quote hifihedgehog Report Post Thanks(0) Quote Reply Posted: 19 Mar 2018 at 3:22am
	Just to verify, where did you download and which version do you have of: 1) AMD Chipset driver 2) AMD Catalyst Graphics driver 3) Intel Wireless driver 4) Intel LAN driver 5) Realtek audio driver Try using the latest versions direct from the vendor, not any old drivers from the ASRock motherboard page. Please provide an answer for each so I can see if I can reproduce this with the latest drivers and a clean install of Windows 10 Fall Creators Update I will be doing in the next week. Edited by hifihedgehog - 19 Mar 2018 at 3:38am

cristy6100 Members Profile Send Private Message Find Members Posts Add to Buddy List Groupie Joined: 15 Mar 2018 Location: Romania Status: Offline Points: 269	Post Options Post Reply Quote cristy6100 Report Post Thanks(0) Quote Reply Posted: 18 Mar 2018 at 7:17pm
	Thank you for confirming these issues. For the SATA it should tell you something like AHCI P and after P a number, but if you have CSM disabled and windows installed in UEFI I think it will only list the disk name without the corresponding port number, so that is another bug,if you have 4 disks of the same model with 4 UEFI installs on it you will have to guess in the boot screen and bios menus which disk is which, so the 7th bug so far. The inability to disable the SATA controller is not intended, it should be disable and the whole menu needs do disapear from there, the only thing that menu needs to read after you switch the SATA controller to Disabled is just that "Sata Controller - Disabled" it should not list the ports/disks or any boot menus. Well clearly we need to wait for a new UEFI version from ASRock, then we need to test those 7 bugs out, including the throttling issue. I suggest you Load Default UEFI options, thats what I use now, I loaded the default options, disabled CSM and Enabled Secure Boot, thats all I changed. I even dropped down my memory from 3200 @ 1.35V to stock 2400 @ 1.2V, until I new UEFI version any overclock is useless.... Edited by cristy6100 - 18 Mar 2018 at 7:26pm