AB350 Gaming-iTX/ac - freezing (oh no...)
Printed From: ASRock.com
Category: Technical Support
Forum Name: AMD Motherboards
Forum Description: Question about ASRock AMD motherboards
URL: https://forum.asrock.com/forum_posts.asp?TID=8070
Printed Date: 07 May 2024 at 2:41am
Software Version: Web Wiz Forums 12.04 - http://www.webwizforums.com
Topic: AB350 Gaming-iTX/ac - freezing (oh no...)
Posted By: Thenightstalker
Subject: AB350 Gaming-iTX/ac - freezing (oh no...)
Date Posted: 15 Mar 2018 at 3:47pm
|
" rel="nofollow - And there's another one freezing ... Hello everyone, just finished my iTX Build with r5 2400G on the ASRock AB350 Gaming-iTX/ac and after installing everything and getting into testing everything my system begins freezing. Just out of nowhere ... System specs: ASRock AB350 Gaming-iTX/ac - BIOS 4.50 Ryzen 5 2400G APU - Noctua NH L9x65 G.Skill Trident Z 16 GB PC4000 CL19 (running @3300 / CL14 1,4 V) Samsung 830 128 GB SATA SSD (old but that's what I just had lying around) Philips 47" LCD TV over HDMI (again, that's what was just there when I was done) Windows 10 Pro (latest built, just downloaded it 1 week ago) Latest drivers for ATI Radeion, LAN/WLAN, AUDI etc. (downloaded via ASRock Shop Tool) First I started playing Assassin's Creed Black Flag which I just had there to try out. It runs OK as far as I can tell but it could use some Ryzen optimizations I guess. However I was fine but there was a specific mission where the game crashed just after completing it, screen stayed black, audio plays then stops and some wired artefacts appear on screen. I played it again and it freezed at the same point. I went to uefi and changed the RAM timings from CL14 to 15 but doesn't help, it freezes again just at the same point. After 4-5 tries I gave up but because everything else was ok and temps where just fine I thought it's just a bug in the game or an issue with the drivers because it's a quite old game on a quite new system. After that I wanted to go into stress testing a bit and began to download 3D Mark. It should take quite while because my WifI is bad here in the basement. After a while I looked at the download status and it didn't change anymore, I thought the WiFi just dropped and it's resyncing but when I grab the mouse I noticed it was frozen!? I thought ok, that may happens with a brand new System where maybe some settings are not optimal or thermals may be an issue because cooler not sitting perfect. I did reset and restarted the download and after another while again same thing but this time I watched it. The connections seemed to be lost - it stopped downloading but not abborting it, just stays at the actual speed but no data coming through. Everything else looked normal, I moved my mouse around to see if it was frozen but it wasn't yet but a few seconds later it did freeze again. After that I went to the uefi once more and turned the RAM down to it's default timings (CL19) but at 3200 ofc and with default 1.35 V. After this I had no more freeze for the next half hour before I shut down the system. I completed the 3D Mark download and ran the 3D Mark Time Spy test. The first time it stopps with an error after the first demo but no details about the error. Maybe my switching between 3D Mark and AMD Master tool caused the problem. I ran it a 2nd time and it was OK! CPU Temperatures where waaay below any critical (about 50-60° at max). One interessting thing, 3D Mark GPU test ran quite bad, I was wondering why but at the end it says my GPU was unknown and drivers not acceptet!? But on the start screen it showed up correctly!? Could there be any connection between the freezes and an eveutal broken gpu driver? I don't think so but who knows...!? Okay guys, that's quite a bit information but I think you'll need as much Information as possible to find the problem. Maybe it's just the RAM becaue after setting it to default timings it worked but maybe it was just coincidence. I will do some more testing tonight, I plan to go for RAM testing, they SHOULD make 3200 CL14/14/14/14/34 1.4 Volts and as they are Samsung B-dies I belive they will. Just as the FlareX do. Maybe I need to go down to 2933!? I don't know? Any ideas? Otherwise I'll try the new Bios, I heard of an WiFI issue, could there be any connections? Because the freezing only happens while downloading over WiFi!? EDIT: Upon further investigations I think this freeezes could really be caused by the WiFi bug. I wasn't awate of that bug before because I thought there's just a problem with wifi not working at all (that's what I had read before) but now I found a post on reddit where some people described exactly what happened to me. They say streaming videos or other network heavy things like just downloading a huge file will cause bsod or freeze almost 100% after a couple of minutes or at least half an hour / hour. So that's exactly what I experienced last night while downloading 3D Mark. This should USUALLY be a quick thing over my cable network but because I was in the basement LUCKILY my WiFi connection was bad enough to keep the download running for about 15-20 minutes. I say luckily because otherwise I may wouldn't be able to see the freeze and would have a much harder time figuring it out later when the system runs stable but begins to crash out of nothing when, for example a windows update is loading in background or something like that. This way I was able to eliminate parameters quite easy, I'll flash 3,51 tonight and see if this fixes the freezing issue! |
Replies:
Posted By: stree
Date Posted: 15 Mar 2018 at 7:01pm
|
" rel="nofollow - Could you state exactly, the chipset and gpu (Radeon) drivers installs? This must be chipset first and then the Radeon drivers, because if you install the Radeon first, the video drivers in the chipset pack will overwrite with graphics drivers that will give problems. ------------- ASRock X370-ITX BIOS 4.50 R5 2600 Cryorig C7 EVGA GTX 950 75w 2x8GB Ballistix Sport LT 2933 960Evo M.2 256GB, Firecuda 1TB Win 10 Pro 64 1803 G-Unique Archdaemon 300 Watt Lian-li Q21B |
Posted By: Thenightstalker
Date Posted: 15 Mar 2018 at 8:42pm
I'd bet I did it in this order mainly also because of my slow WiFi connection. But I installed all the drivers through the ASRock App Shop. I remember downloading all drivers except ATI Radeon AiO because I wanted the smaller ones to be done first. Also I knew that display drivers usually should come last because they often interfere with others. So I think I did it that way. But I can't really tell which drivers the ASRock Utility offered me, there was a HD Audio as well as an Intel Networking device driver and a 3rd one. Are there 2 different Network drivers for LAN and WLAN controllers? In that case there was no specific chipset driver at all. But I used the latest Windows 10 built and did all available updates straight after installing Windows. So they myght already have been installed with the system? However, I don't think this caused the freezing, maybe that's why AC BF crashed and 3D Mark didn't know my GPU and driver...!? |
stree wrote:Posted By: stree
Date Posted: 15 Mar 2018 at 9:31pm
|
that sound more like All-in-one drivers than chipset drivers, and you are very vague about which drivers and in what order. You need to get this established really before anything else. latest chipset drivers, complete package, no picking aand choosing, the whole package as is, and only from AMD, nowhere else. Same with Radeon drivers, only from AMD site, latest version is still 17.7 I think. MUST install chipset first, then graphics, and you must uninstall current drivers first before intalling latest. https://support.amd.com/en-us/download" rel="nofollow - https://support.amd.com/en-us/download ------------- ASRock X370-ITX BIOS 4.50 R5 2600 Cryorig C7 EVGA GTX 950 75w 2x8GB Ballistix Sport LT 2933 960Evo M.2 256GB, Firecuda 1TB Win 10 Pro 64 1803 G-Unique Archdaemon 300 Watt Lian-li Q21B |
Posted By: Thenightstalker
Date Posted: 15 Mar 2018 at 9:48pm
|
" rel="nofollow - I was pretty sure everything was/is ok just because there where noc issues, no error messages in my device manager and all devices seem to work as they should. But you're right, that was a bit to easy. I didn't chose anything, I installe the ASRock App Store app and in this there's a drivers/updates section where it shows me ATI Graphics drivers, LAN Drivers and HD Audio drivers which I installed because they defenitely where missing. But I think I should try to do it again the way you described. Therefore I'll unsinstall the ATI Drivers completely, as well as the LAN / WLAN Drivers and download the specific chipset and latest ATI Radeon Drivers directly from ATI. So I'll get this first: https://support.amd.com/de-de/download/chipset?os=Windows%2010%20-%2064 Then go there: https://support.amd.com/en-us/kb-articles/Pages/Radeon-Software-for-Ryzen-Desktop-Processors-with-Radeon-Vega-Graphics-Release-Notes.aspx and get the drivers for Ryzen APU Graphics. After that I should be done, right? Or do you think a clean Windows 10 reinstall would be better? |
Posted By: stree
Date Posted: 15 Mar 2018 at 9:51pm
|
Stick with just chipset and graphics drivers for now, see what you get..........can take it from there...... ------------- ASRock X370-ITX BIOS 4.50 R5 2600 Cryorig C7 EVGA GTX 950 75w 2x8GB Ballistix Sport LT 2933 960Evo M.2 256GB, Firecuda 1TB Win 10 Pro 64 1803 G-Unique Archdaemon 300 Watt Lian-li Q21B |
Posted By: Thenightstalker
Date Posted: 15 Mar 2018 at 10:04pm
|
K, I'lll give it a try tonight ;-) I'll also upgrade Bios to 4.51 to fix any possible wlan issues. Get back here with results. |
Posted By: Thenightstalker
Date Posted: 16 Mar 2018 at 3:36pm
|
" rel="nofollow - Ok, looks like it's working now, kind of. Here's what I did: First I downloaded new beta bios and flashed it. Worked fine. Then I set back the RAM to my "medium" settings > 3200 CL15/15/15/15 @1,4 V Also set the frame buffer back to 2G where it was. Then booting up, everything fine. I downloaded the chipset drivers and display drivers from AMD as well as the cleanup utility. No freezes or bsod so far! After that I ran cleanup utility and removed all display drivers and tools. Then I installed the chipset drivers followed by the display drivers. Always properly restarting after each install, ofc. After that I downloaded Assassin's Creed Origins which was a 43 GB and everything worked fine. The system fellt into suspend mode once but it turns back on immediately and worked fine. Now comes the sad part ... I wanted to try Origins because I've seen a couple of tests where the game ran quite well for an APU and I thought that's great, I'm ok with that performance and want to play it. The game runs, I played the first part on a medium quality which was quite good except a few drops in fps but mostly OK. Even when you come to town it was playable but then after a while it started to stutter massively, everey 5-10 seconds it seemed to almost stop for 1-2 seconds before going on. I looked at the ingame benchmark and saw the awful result. I set everything to low (not lowest) hoping it would help but no, same issue! It's like the game ist loading tons of data every few seconds and slows down like hell. That can't be normal!? The ingame benchmark show a max. FPS of about 40 (medium) and 50 (low) so that's what I was expecting after seeing the tests but the average fps was about 20 because of the massive stuttering every few seconds. I don't think that's a mainboard issue but maybe a settings issue, I don't know. Can anyone help? Thanks |
Posted By: Xaltar
Date Posted: 16 Mar 2018 at 5:51pm
|
Assasin's creed origins is BRUTAL on systems thanks to it's excessive DRM and less than ideal optimization. This is a well known issue that brings any system even remotely falling short of recommended specs to it's knees. Give it a google, there may be some workarounds/tricks to pick things up. Generally you will get better results lowering the resolution than dropping settings from medium to low. If there are any view distance sliders or foliage etc you can lower these, this will reduce the amount of data swapped between the CPU, GPU, HDD and RAM. On the up side 50fps on low means your APU is working and stretching it's legs. You can also try setting your share memory as low as you can. The only benefit of having more dedicated vRAM is that is faster than system RAM, on an APU that is completely moot. When you run out of frame buffer the game swaps to RAM, which, is the same speed as the frame buffer. Given you have 16gb of RAM I doubt it will make much of a difference but it may play better (or worse) with some games. Don't be shy to experiment with different share memory sizes with different games. Some will run better with smaller numbers and others with larger, it all depends on what aspect of the engine is best optimized. Some games optimize RAM swapping to the extreme to facilitate smoother gameplay on older systems, in these cases you will benefit from smaller share memory sizes for example. ------------- 
|
Posted By: Thenightstalker
Date Posted: 16 Mar 2018 at 6:57pm
|
Yeah that's one reason why I chose it to try on this system but not without seeing others trying it first. There's a couple of tests and videos that stated it runs "OK" even on a 2200g (ofc on low settings) and my overal OK FPS are telling me that the system can run it pretty smoth. Those drops are horrible, since they appear frequently and without any clear reason. I tried to quit any pograms that may interfere but nothing changes. However I did some further investigations on that problem and found this: https://www.reddit.com/r/Amd/comments/82wmo3/2400g_users_turbo_off_helps_with_slowdowns_who/" rel="nofollow - https://www.reddit.com/r/Amd/comments/82wmo3/2400g_users_turbo_off_helps_with_slowdowns_who/ This video shows basically the same behavior I'm experiencing with Origins. The frequenc it appears is different but it's exactly the same thing, smooth FPS, suddenly stuttering for ~3 seconds and again smooth FPS. They say it's a bug in the CPU/APU turbo mode, which at least sounds plusible. I'll try this today and get back with results. Thanks |
Posted By: cristy6100
Date Posted: 16 Mar 2018 at 7:36pm
|
" rel="nofollow - I reported this yesterday here http://forum.asrock.com/forum_posts.asp?TID=8075&PID=47916묬 The only temporary fix is to lower your CPU voltage (with a negative offeset, try -100 first cause its the lowest offeset available and test from there) this will lower the frequency of the throttles but not eliminate them
|
Posted By: Xaltar
Date Posted: 16 Mar 2018 at 7:36pm
|
Early on when Ryzen launched I would get hanging in games on my 1600x for about 3 - 5 seconds then back to normal (usually with me dead lol). In the end a windows update fixed the issue, after months of scratching my head, trying different components and OS installs. The turbo theory certainly seems to be in the same vein, worth a shot. Good luck  -------------
|
Posted By: Thenightstalker
Date Posted: 16 Mar 2018 at 7:52pm
|
I will try both, custom setings and reducing voltage / offset. Let's see what we can do. It would be great to play that game on this machine, freakin great to be honest. |
Posted By: cristy6100
Date Posted: 16 Mar 2018 at 8:00pm
|
See here a little more info http://www.apu.graphics/tipps-and-news-for-ryzen-2400g-throttling/" rel="nofollow - http://www.apu.graphics/tipps-and-news-for-ryzen-2400g-throttling/ taken from here https://www.reddit.com/r/Amd/comments/82wmo3/2400g_users_turbo_off_helps_with_slowdowns_who/" rel="nofollow - https://www.reddit.com/r/Amd/comments/82wmo3/2400g_users_turbo_off_helps_with_slowdowns_who/ There is also some talk about SMU 30.68.0 fixing the trottling, the latest available bioses have 30.67.0 so we dont know for sure until ASRock releases updated bios You can check your SMU (which might stand for System Managment Unit) version by going in Advanced/AMD PBS/Firmware versions
|
Posted By: Thenightstalker
Date Posted: 16 Mar 2018 at 8:27pm
|
Ok I'll check that too and let you all know... so it may be though a bios/board issue in that case. Fortunately MOST of those things can and will be fixed someday. Given the fact that the Ryzen APUs are quite new I'm positive and looking forward to seeing some major improvements in the near future. Anything else would be a shame for everyone... This doesn't only mean board manufatorers and AMD themselfes, also Microsoft and all the game developers should get their asses up ;-) |
Posted By: Thenightstalker
Date Posted: 17 Mar 2018 at 5:15pm
|
" rel="nofollow - Ok, here?™s an update! First of all, no more freezing since 4.51! That?™s great! I checked SMU version, it?™s 30.67.0 as we expected. Next I set the CPU to 3700 at stock voltage. As expected everything worked, frequency was fixed to 3,7 Ghz which is ofc not optimal but for testing ok. I fired up the game and played about an hour, no stuttering anymore! That worked! Performance @1080 is still not the bes on medium settings bit way better and playable! Further testing needs to be done now. |
Posted By: cristy6100
Date Posted: 17 Mar 2018 at 7:28pm
|
" rel="nofollow - Interesting finding, could you please tell us what CPU cooler do you use, and cpu fan profile on uefi? Are the stutterings gone with Default UEFI settings, or with fixed voltage and CPU clock at 3.7GHz? Could you also please test the SATA bugs? Enter Advanced/Storage Devices and switch SATA Controller to disable then reboot and press F11 and tell us if the uefi still lists your SATA drives. The 2nd SATA bug is if you put your SATA drives in the 4 different ports and press F11 to boot from them it will list the port number wrongly (the port number is after it states AHCI Px, x being the number of the port) EDIT: See here http://forum.asrock.com/forum_posts.asp?TID=8075&PID=47818 all the bugs I personally found, if you could check to see if your configuration exhibits any of these and report back I would appreciate it
|
Posted By: ajnosek
Date Posted: 17 Mar 2018 at 10:55pm
|
I have the throttling issue too, but I am using the ASRock AB350 gaming ITX/ac motherboard with the 2400g processor. I was getting really bad, regular drops in framerates in PUBG. Here is the fix I discovered, and I have no more problems at all. I simply applied a 350MHz frequency to the GPU with AMD Ryzen Master Utility. The CPU still has turbo enabled. I'm guessing all this did was turn the GPU turbo off. Using a higher GPU frequency also fixes the issue. I would have liked to use a higher frequency, but I'm smashing up against really high CPU temperatures under load when using the stock cooler, like over 90C. I have a new cooler coming and I hope to be able to crank the GPU frequency back up. Here's my theory: It's possible that the issue is that the turbo for the GPU hits some kind of safety limit, and it throttles way back. Once the whole system is choked to almost a stop, the temperature (or power or whatever) drops way down. Then turbo goes back to normal operation, and repeat.
------------- ASRock AB350 Gaming ITX/AC L4.51a, Ryzen 5 2400g 3.6GHz, 2x8GB Patriot Viper 4 3200MHz (PV416G320C6K), Win10 Education, 1TB SSD Samsung 850 EVO, 3TB HDD WD Blue, EVGA 500W (100-W1-0500-KR) |
Posted By: Thenightstalker
Date Posted: 18 Mar 2018 at 2:20am
|
" rel="nofollow - [URL=][/URL][URL=][/URL]Ok, here?�s more details for you I fixed the CPU frequency and voltage by enabling manual mode in uefi. I overclocked the cpu 100mhz just to make sure it uses my manual settings. My fan profile is now set to performance mode and I turned off the cool&quiet feature. I will now systematically test which of the settings did the job. My temps are not causing the issue, I checked temps while running the game and it never went over 60-65 with stock settings. Now with fixed 3.7 Ghz it goes up to ~70 after an hour of AC Origins which is still OK for an air cooled SFF PC. I will also try setting a manual GPU clock speed with the Ryzen tool, this would be much more comfortable because I could simply do this when needed. I can also check the first SATA issue, unfortunately I can?�t do the 2nd because my PSU cables are blocking almost all the ports and I really don?�t want to rework all the cablemanagement today :-) I will have to do this soon because I?�m going to insert a M.2 SSD... |
Posted By: cristy6100
Date Posted: 18 Mar 2018 at 3:58am
|
Interesting, I think the issue was fixed by going manual on the CPU/GPU frequency and voltage, so it would seem to be a AGESA or motherboard issue, ASRock needs to look into this, because in its current state CS GO is unplayable for example and many more, please do check the 1st SATA issue and report back. Regarding the 2nd one, you don't need to move anything, just look at the motherboard where the SATA ports are, they have the numbering printed on the PCB, then just press F11 when booting and confirm if the naming is wrong for the respective ports :) Thanks for checking these out |
Posted By: Xaltar
Date Posted: 18 Mar 2018 at 7:11am
|
Awesome, thanks for all the extra info guys. This is really helpful to both other users an ASRock's support staff. -------------
|
Posted By: Thenightstalker
Date Posted: 18 Mar 2018 at 8:17am
|
" rel="nofollow - Short update When I booted up today I got that stupid power on/off loop bug. After a few tries iit turned on finally but my uefi settings where gone. This happens randomly idk First the SATA stuff, I hit F11, it listed my SSD as first, I can?™t see any port numbers there, it shows the name (Samsung ...) and maybe cuts the rest? Disabling the controller didn?™t seem to do anything too. The drive is still there and even booting. Maybe that?™s intended idk?! I wasn?™t able to specify the performance issue because I made a dumb mistake ;-) I set the CPU back to auto but left the Cool&Quiet off. Set my RAM timings manually to 14/14/14/14 @1.4 V and ran the game... it didn?™t stutter, great but I realized soon that something else was wrong, frames where horrible. I checked temps and watched system load but everything was just fine. So I opened Ryzen Master and set fixed Frequencies. I realized that my RAM timings were wrong and I set them correctly again. But the game still ran bad. I played around with settings about half an hour but even 720p was awfully slow. But then just when I was about to give up I saw the GPU clock speed was just about 420 Mhz!!! Damit how could that happen why didn?™t Insee this earlier?! I manually set the clock to ablut 1300 mhz and now the game was running quite ok even ln medium 1080p. I will check if setting only the GPU manually will help to prevent the stuttering but not today ;-) |
Posted By: cristy6100
Date 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.... |
Posted By: hifihedgehog
Date 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. |
Posted By: Thenightstalker
Date Posted: 19 Mar 2018 at 4:00pm
|
" rel="nofollow - 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. |
Posted By: cristy6100
Date 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) |
Posted By: Thenightstalker
Date 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. |
Posted By: cristy6100
Date Posted: 19 Mar 2018 at 9:37pm
|
" rel="nofollow - 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?
|
Posted By: Thenightstalker
Date Posted: 19 Mar 2018 at 9:50pm
| Amen! |
Posted By: Thenightstalker
Date 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. |
Posted By: cristy6100
Date 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 |
Posted By: Mobawe
Date Posted: 20 Mar 2018 at 11:17pm
|
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 |
Posted By: cristy6100
Date Posted: 21 Mar 2018 at 1:07pm
|
We need to wait for a new bios and graphics driver, they should release Adrenaline for Vega 8/11 with the next WHQL driver release. No point in overclocking when the platform is not stable from the beginning and currently there is no fix for the bugs reported here and the constant stuttering/cold boots/crashes My default core voltage is also very high, on a Cinebench R15 run, the core boosts to 3800 but pulls 1.425V for that, seems a little on the high side Regarding the memory I have not tried lowering the latencies on mine yet, but I dont think it will make a big difference if you only go from lets say 16-16-16 to 15-15-15, bigger difference if going from 3000 to 3466, my Cosrair LPX wont boot past 3200 @ 1.35V but I'm happy cause they are 2400 modules. You can set an offest for the CPU voltage -100 is max that way you will still have adaptive voltage control instead of a fixed voltage, see if that helps I tested today but as both if you said very hard to get a stable GFX clock past 1300, I won't play with the SOC/GFX voltage until we have a bios entry from where we can check in realtime the voltage reading, the CPU offers SOC/GFX voltage monitoring in realtime but ASRock decided not to implement this reading in the BIOS, damn another problem....
|
Posted By: Thenightstalker
Date Posted: 22 Mar 2018 at 5:14am
|
Hi, there's some serious bs going on with this whole plattform right now ... My "cold boot issues" are definetely caused by my RAM settings, I get it every time. However my RAM runs just fine as far as I have testet it. I played AC Origins for hours with medium and low timings (low CL 14/14/14/14). I'm actually using G.Skill 4000 C19 with Samsung B-Dies which are confirmed to run 3200 mhz / cl14 with 1.4 V. Now there could be 2 reasons for the issue. 1. There's a bug in the timing or voltage settings that causes the failure at boot 2. There's a bug in the AM4 RAM Training fuction. I turned this off now just to see what happes because my RAM runs absolutely stable with these settings. It also boots and reboots with this settings everytime until I unplug my PSU and reboot the next day. Weird... EDIT: OK, to make sure my RAM is stable I just ran a full pass MemTest86 > no errors with 3200 / CL14 @1.4 V. It's definetely NOT my RAM. |
Posted By: cristy6100
Date Posted: 22 Mar 2018 at 7:22am
|
" rel="nofollow - Could you try another thing? Enter UEFI, and change overclock settings from ASRock Settings to AMD CBS then go to /AMD CBS/Zen Common Options/ and switch Core Performance Boost to Disabled, this should fix 90% of stuttering and frame drops, you will still have Adaptive Frequency and Voltage but you will loose Turbo and XFR so CPU part will be limited to 3.6GHz on all cores, the GPU part remains at the same frequency as before. I dont know if it will help with the cold boots but worth checking EDIT: you will need to switch to AMD CBS Settings from ASRock for this setting to stick
|
Posted By: Thenightstalker
Date Posted: 22 Mar 2018 at 7:26am
OK, going to check this tomorrow, have to get some sleep now! I'll be back with an update. |
Posted By: cristy6100
Date Posted: 22 Mar 2018 at 7:34am
| Another thing that happens after this is the MAX CPU core voltage will go down from aprox 1.425V to 1.256V a substantial jump, that confirms that this is indeed a power throttle or something not a cpu cooling throttle. |
Posted By: Thenightstalker
Date Posted: 23 Mar 2018 at 6:44am
|
" rel="nofollow - Only a short update ... Today again cold boot issue, again the RAM settings where reset as well as GFX settings. I went through the process of calculating the HEX/DEZ numbers for my RAM timings and enterd them in AMD CBS settings. Also turned off Core Performance Boost. Now the stuttering is gone as you described it but turbo also is off now but ok that's at least something. Again, my RAM was absolutely stbale with my settings, I'll try lowering the voltage next time too because I think it would run those timings on 1,35 V as well. I hope ASRock will patch these problems soon, it seems to be a specific problem here, isn't it? AMD is working closely with MSI, that's garbage... |
Posted By: Mobawe
Date Posted: 23 Mar 2018 at 4:45pm
Hi Guys I tried to do the same and it was definetely a step forward.
Yes, I dont have the Turbo on the CPU anymore but now I am able to clock the
APU much higher than before. Before I was not able to clock the apu above
1300Mhz without a crash and it was not a temperature issue. My guess is that there is some bogus going on with the power
distribution When boosting. What makes me confused is that it somehow seems to affect
the RAM and APU performance- I explain further down. I thought that the CPU
voltage and the SOC Voltage for the Ram and APU are not influencing each other?
Is that a misunderstanding? For now I can live with the CPU limited to the 3600 Mhz
because I think the OC of the APU and the RAM is much more important on this
chip ??Performancewise. The thing I am strugling with is the SOC Voltage that feeds
both the ram and the APU. I now run with my ram on 3400 Mhz but the tradeoff is
that I cannot OC the APU much (MAX 1300 Mhz). If I lower the ram frequenzy I
can OC the APU more. If I OC the APU and maintain the high RAM frequency it close
down Crysis after around 10 minutes with the windows popup, about closing down
the program. If i lower the APU clock it runs the game without problems. It seems that the SOC Voltage is directly influencing this -
The higher SOC voltage the higher I can go with both the RAM and the APU. If I dont
touch the SOC voltage, I have to lower the Ram if I raise the APU Clock and the
other way around. This makes me think a few things: I.
I think that my 2x8 Ripjaws 3600 needs a lot of
SOC voltage to work with the high frequencies. II.
I think that the APU also needs a quite high SOC
voltage and that it kind of compete with the ram for it. https://www.gamersnexus.net/guides/3251-raven-ridge-soc-voltage-guidelines-how-to-kill-cpu-with-safe-voltage" rel="nofollow - https://www.gamersnexus.net/guides/3251-raven-ridge-soc-voltage-guidelines-how-to-kill-cpu-with-safe-voltage Here it is mentioned that you could go as high as 1.25 V to
1.30V on the SOC Voltage (Ram and APU) with the RavenRidge (Seems like more
than the old Ryzen that didn't have the onboard APU). The problem that the
article highligts is that you cannot always trust the boards setting so you could
end up giving to much without being aware of it! Now I am on 1,25 for the SOC
Voltage and 1,4V for the CPU. I think that I could run My ram on the maximum 3466MHZ (Board maximum) and also have a high APU OC if I could go higher on the SOV Voltaget han the 1,25V. I would love some more hard facts about the RavenRidge voltage
limits but I guess it would surface eventually. ------------- 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 |
Posted By: cristy6100
Date Posted: 23 Mar 2018 at 6:59pm
| Indeed you are right, I mentioned this before, on the AB350/X370 Fatality ITX there is no SOC/GFX voltage monitoring entry in the UEFI, you should not exceed 1.25V SOC/GFX, a voltage monitoring entry should be added for SOC/GFX because the CPU has sensors for this and has outputs for displaying these voltages but ASRock decided its easier not to add these entries in the UEFI.... |
Posted By: Thenightstalker
Date Posted: 24 Mar 2018 at 6:48am
|
This whole thing is complete garbage, I'm sorry ... I'm not even talking about the SOC Voltage and trouble on OCing above 3200 (RAM) or OCing APU at all.... The fact that they can't get this damn system to cold boot with absolutely f***ing stable RAM settings is just ....  As I said yesterday I set all my RAM settings in CBS now but still I can't get it to boot. I turned of tha RAM training as well but it also still resets my RAM settings as well as my GFX settings in the CBS. If it would set the RAM at leas to something "usable" but it clocks it down to 2133 .. MAN, THOSE MODULES ARE 4000 and retail at about 250 bucks, are you kidding me ASRock!? Sorry guys but I'm pissed off, I had plenty of systems in the past but that's so bad, get that fixed asap! |
Posted By: Thenightstalker
Date Posted: 26 Mar 2018 at 6:57am
|
Little update ... Running the system now on CBS OC settings since 2 days with AUTO settings for RAM timings. This way I had no more cold boot issues but the timings are terrible ... at least I can boot whenever I want and reboot with Ryzen Master to my prefered timings ... sucks but works for now. Also got my NVMe Samsung 960 now, hopefully this will add a bit more speed as well. Here's a quick bench result for my system http://www.userbenchmark.com/UserRun/7982487" rel="nofollow - http://www.userbenchmark.com/UserRun/7982487 |