Overclock.net - An Overclocking Community - Reply to Topic

Thread: [Official] - X58 Xeon Club - Reply to Thread
Title:
Message:

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


  Additional Options
Miscellaneous Options

  Topic Review (Newest First)
08-14-2020 05:05 PM
xxpenguinxx The RAM lives! I stuck them in the freezer overnight, then let them slowly come back up to room temp. I put them in a sealed bag with most the air sucked out to prevent condensation.

Found a post dating back to 2008 from @PizzaMan , linking to this thread.
http://www.xtremesystems.org/forums/...cron-D9-Owners

Maybe there was some residual power preventing the spd data from being read and it just needed to sit, but either way it's working.
08-14-2020 04:55 AM
PotatoVonEpicus Needed 1.296v to get 4 GHz working without freezing with LLC 1. LLC 2 does 4.1 GHz on 1.328v, not that bad, however higher frequencies likely would want more volts.

Load and idle voltage is the same, so this is not bad, C-States seemingly work as well, including EIST and all other power saving features. Before I had 4 GHz working on 1.280v with LLC 2. This seems stable, by now this went through more than 2 hours of LinX and around 1 hour + of Prime95. Played some GTA 5 as well, without any issues. Only time can show if this is going to be stable long term.
08-14-2020 03:33 AM
PotatoVonEpicus I will give LLC another look, though I been using LLC 2 for longest of times without any issues. Also used it with x5670 as well, which I had stable on 4.3 and 4.1 GHz with some generic RAM. Today ran some LinX ( 256 MB ) tests, got error even after adjusting Uncore frequency and voltage to be stable ( used settings from before that are known to be stable, although I am not sure what is stable anymore ). Reduced multiplier, going down to 4.2 GHz instead of 4.4 ( kept same voltage ) and ran more LinX, no errors. Ran Memtest86 ( with reduced multiplier for CPU ) for a bit, no errors. It could be LLC or it could be CPU degradation and page file BSODs can easily be corrupted windows. But I am not sure if memory management BSOD can be caused by corrupted windows.
08-14-2020 02:31 AM
theister
Quote: Originally Posted by xxpenguinxx View Post
You really can't see what LLC setting works best without an oscilloscope. On modern motherboards you want some vdroop to give the vrm time to adjust. Your idle voltage will be higher, but your voltage under load will be more stable.
This goes with every vrm cause it is just physics and the vrms from back in the days of x58 might be slightly worse in terms of transient response compared to modern designs due early implementations of digital vrms (evga, asus) or simply beeing analog ones (gigabyte boards) with relative low pwm switching frequencies. (this is why x58a oc has user control pwm per dips).

And the voltage behaviour is exactly that what might cause the trouble for him, with LLC2 he has a low idle voltage thats spikes even lower due a short workloadchange (interrupt with lower workload), with less llc you are coming from a higher voltage level, it drops under load and than goes up if an interupt is handled, so you will never get into "to less vcore" zones.
08-14-2020 12:00 AM
PotatoVonEpicus I highly doubt it's magnetic storms that cause this, but who knows...

Regarding LLC and stress testing, I sometimes use my PC to watch videos or read stuff online while stress testing. While LLC can be possible culprit for voltage going too low, it's unlikely to be the cause of these crashes. I experienced these crashes on multiple profiles and most of them pointed to memory ( Memory_management and Page_file_fault ). That aside my main profile always was 4.1 GHz and I always used same voltage with LLC 2, which is 1.328v full load. Now this specific voltage and frequency in total has seen around 200 hours of LinX over long time that is. Only possible explanation would be CPU degradation and it needing more voltage, today I ran more LinX ( 256 MB ) only with 2 sticks on my 4.4 GHz overclock. It BSODed with uncorrectable error, it's Uncore, no worries I was running 3800 MHz on 1.43v ( 256 MB LinX tests helped here, LinX max memory passes seem to be less useful while testing Uncore ). So this one is solved, however this doesn't explain crashes with other profile, especially memory related crashes and magical way to fix them by removing one stick. 4.1 GHz profile has been tested with Prime95 and LinX as well, no errors found with all 3 sticks, just to randomly crash at any time. Another weird thing I observed is that crashes come in waves so to say, it can be stable for longest of time, it will start crashing even after crash PC would reboot just to crash again on boot and then again would be working just fine even have no issues with LinX or Prime95. As for LLC I can't really use LLC 1 since this CPU likes voltage a lot, if I use LLC 1 it will be spiking and running rather high voltage.
08-13-2020 11:25 PM
xxpenguinxx You really can't see what LLC setting works best without an oscilloscope. On modern motherboards you want some vdroop to give the vrm time to adjust. Your idle voltage will be higher, but your voltage under load will be more stable.

The vraise you're seeing is probably the VRM overcompensating because it's set to not let the voltage drop on LLC2. It's hard to say exactly without seeing the voltage ripple.
08-13-2020 06:52 PM
theister
Quote: Originally Posted by PotatoVonEpicus View Post
I haven't tried going back to Win7, it's been around 3 years since I had Win7 installed. Win10 simply has some features that I don't want to miss out on. Taking one stick out seemingly fixes everything, this is rather weird, however there might be issues with chip itself so triple channel doesn't work properly. Possibly board is not liking mixing of sticks, although more often than not it seems fine. I tried several things windows related to fix windows corrupt files and looked at drivers as well, everything seems fine there. Right now using two sticks without any issues at all with exactly same settings, this is just weird, but hopes for me to run triple channel properly stable and not randomly crashing are slim. I rather have stable working machine instead of thinking that it is stable just to randomly crash, also another funny thing is that this PC seen a lot of stress testing these past two weeks. Everything seemed fine and then it randomly crashes in things like GTA 5. RIDICULOUS !
Are you using LLC Level 2 ? This will lead to vcore undershoot where your cpu is going to have not enough vcore anymore if workload changes happen due windows event timer trigger. Every time something is happening (like mouse movment) userwise windows will "pause" the actual workload that I results in a voltage drop. Play a video while you are running LinX / Prime95 or better move your mouse exesivly and i am sure you will see a crash with your "stable" setting. The windows event timer is always running but without user activity it runs in a constant way that may not cause instability, running videos or moving the mouse will do, there is also a tool to trigger windows event timer more frequently, named TimerTool.

In general i found llc2 with 400 pwm freq gigabyte boards is not very good for dailly and switching workloads, in the beginning vdroop is like on llc 1 and a short moment later it acts like a vraise (vcore is higher than set in bios) what causes the mentioned issues and leads to way more overshoot that you also not want in terms of cpu health.
08-13-2020 02:43 PM
xxpenguinxx I killed 3 sticks of RAM, Samsung M393B255273DH0-YH9, Inphi registers. They worked at 1.65V up to 2200Mhz 9-10-10-28. I restarted after passing a few rounds of prime95 and the motherboard got stuck in a boot loop on/off. It cycled though error codes 01, 02, 68, EA.

They were barely over room temperature so not a heat issue. Maybe a voltage spike killed the registers? Makes no sense for all of them to instant die at a restart.

EDIT: Definitely dead. Tried them in the server and it hangs before enabling the RAM. Other registered RAM work fine.
08-13-2020 01:58 PM
Fobia2000 I also struggled with the crash of Windows today, maybe it's magnetic storms?
08-13-2020 12:20 PM
PotatoVonEpicus I haven't tried going back to Win7, it's been around 3 years since I had Win7 installed. Win10 simply has some features that I don't want to miss out on. Taking one stick out seemingly fixes everything, this is rather weird, however there might be issues with chip itself so triple channel doesn't work properly. Possibly board is not liking mixing of sticks, although more often than not it seems fine. I tried several things windows related to fix windows corrupt files and looked at drivers as well, everything seems fine there. Right now using two sticks without any issues at all with exactly same settings, this is just weird, but hopes for me to run triple channel properly stable and not randomly crashing are slim. I rather have stable working machine instead of thinking that it is stable just to randomly crash, also another funny thing is that this PC seen a lot of stress testing these past two weeks. Everything seemed fine and then it randomly crashes in things like GTA 5. RIDICULOUS !
This thread has more than 10 replies. Click here to review the whole thread.

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