Jump to content

Overclocking stability

ArushM

Sorry for the repost but no one answered my first issue in 16 hrs and the issue still exists, mods take this down if you want to 

My PC: https://pcpartpicker.com/user/ArushM/saved/#view=XTXfpg

So I was overclocking and I randomly crashed after trying to watch a youtube video (it didnt load the video and just fully crashed), this continued happening doing anything normal but seemed to be fine during stress tests,

Overclocks(everything is done in XTU minus the memory which was done in BIOS):

i5-13600K @ 5.5/4.5 Ghz - Default voltage(1.25 idle, 1.23 under load) , LLC mode 4 (MSI), No avx2 ratio decrease, ~250W max (It thermal throttles like hell but im working on it)

Intel UHD 770 iGPU @ 21Ghz - 1.15V (static)

Ring Ratio (Uncore clock) - was 4.7 but now at 4.5 (Stock) and still crashing

Kingston 5200 CL40 @ 5800 CL36 - Stock voltage

The GPU was linpack xtreme'd for about 24 hrs

Memory was Memtest86+ for 13 hours (20 passes)

GPU was tested in Fortnite for 2 Hrs

Ring ratio was tested with the CPU for the same time

 

Problem:

  • Any reason why this is happening, and how do I fix it and also what can I do to prevent this again
  • Any recommendations to test Ring Ratio and GPU with error checking to avoid something like this again
Link to comment
Share on other sites

Link to post
Share on other sites

Its only been 16 hours dont be so impatient, else pay someone to fix it. 😄
Its just made of people that want to be here, not a help desk.

When i ask for more specs, don't expect me to know the answer!
I'm just helping YOU to help YOURSELF!
(The more info you give the easier it is for others to help you out!)

Not willing to capitulate to the ignorance of the masses!

Link to comment
Share on other sites

Link to post
Share on other sites

4 minutes ago, HanZie82 said:

Its only been 16 hours dont be so impatient, else pay someone to fix it. 😄
Its just made of people that want to be here, not a help desk.

Fair enough, mb

Link to comment
Share on other sites

Link to post
Share on other sites

If you run locked speed on the cpu it doesn't need a power limit.
It sounds like memory instability, if it's 5200mhz it's probably not a very good kit. You need a 6400+ kit then you can get A-die which will probably overclock to the limit of your imc/mobo (Usually 7200-7600 if you are good at tuning and have a half decent z790 board).
Also test memory speed, with DDR5 I think if it's unstable it'll actually be slower. Use the aida64 cache and memory test for this. if you can get 55ns or lower on your memory that's pretty good and below 52-53 I don't think there's much performance scaling at least for games. You might not get that low with your current kit though.

PC Specifications: Intel i9-14900KF, 5.7GHz all core locked, 5GHz ring, 1.35v Medium LLC, E-cores and HT disabled | MSI RTX 4090 Gaming X Trio | Arctic Liquid Freezer II 360 + Thermal Grizzly contact frame | 2x16 G.Skill Trident Z5 7400MHz 34-44-44-34 1T 1.45v (Maxed Subtimings, 52ns latency) | Gigabyte Z790 AORUS Elite AX | Windows 10/11 EVGA SuperNOVA 1000 T2 Phanteks P400A (Black non-rgb version, Phanteks T30 fans 3 intake (On AIO), 1 exhaust) | SK Hynix Platinum P41 2TB PCIe 4 SSD

 

Displays: MSI MAG 271QPX 1440p 360Hz 27" QD-OLED

 

Desktop Audio: STAX SR-007 MK2 Electrostatic Headphones (Current revision) | STAX SRM-400S Amp | Schiit Bifrost 2/64 (NOS mode, USB in, XLR out)

 

Mobile Audio: Sennheiser IE 900 IEMs w/ 4.4mm cable | FiiO KA13

 

Peripherals: Razer Huntsman V2 Full size wired with linear optical switch | Logitech G502 Hero

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, rippy4500 said:

If you run locked speed on the cpu it doesn't need a power limit.
It sounds like memory instability, if it's 5200mhz it's probably not a very good kit. You need a 6400+ kit then you can get A-die which will probably overclock to the limit of your imc/mobo (Usually 7200-7600 if you are good at tuning and have a half decent z790 board).
Also test memory speed, with DDR5 I think if it's unstable it'll actually be slower. Use the aida64 cache and memory test for this. if you can get 55ns or lower on your memory that's pretty good and below 52-53 I don't think there's much performance scaling at least for games. You might not get that low with your current kit though.

Is memtest86+ stable not good enough? Thought that was the gold standerd, as for the kit I bought it like 2 years ago when DDR5 was really new so yeah its not the best. I got like 71 ns of latency on the memory, as I said not a great kit.

Link to comment
Share on other sites

Link to post
Share on other sites

27 minutes ago, ArushM said:

Kingston 5200 CL40 @ 5800 CL36 - Stock voltage

I'm sorry, but do you refers to mix and match 2 different rated sped of RAM or testing different XMP profile on the RAM to see if the OC is safe or not?

 

When your rig is crashed, does it only happened on the related app/game?
Or is there BSOD?

If there is BSOD, you can check at 'Event Viewer' to at least know what is the  trigger of the BSOD

My System: Ryzen 7800X3D // Gigabyte B650 AORUS ELITE AX // 32GB 6000MHz DDR5 Silicon Power Zenith CL30 // Sapphire Pulse AMD Radeon RX 7900 XT OC with mod heatsink on the metal plate  // Phanteks P300A  // Gigabyte Aorus GEN4 7300 PCIE 4.0 NVME // Kingston NV2 Gen4 PCIE 4.0 NVME // 

Seasonic Focus GX-850 Fully Modular // Thermalright Frost Spirit 140 Black V3 // Phanteks M25 140mm // Display: Bezel 32MD845 V2 QHD // Keychron K8 Pro (Mod: Gateron black box ink; Tape mode on PCB and Keycaps) // Razer Cobra Wired Mouse // Audio Technica M50X Headphone // Sennheiser HD 650 // Genius SP-HF180 USB Speaker //

 

And Laptop Acer Nitro 5 AN515-45 for mobility

Phone:

iPhone 11 (with battery replaced instead of buying new phone for long term and not submitting (fully) to Apple Lord

Link to comment
Share on other sites

Link to post
Share on other sites

Just now, ArushM said:

Is memtest86+ stable not good enough? Thought that was the gold standerd, as for the kit I bought it like 2 years ago when DDR5 was really new so yeah its not the best. I got like 71 ns of latency on the memory, as I said not a great kit.

Use multiple tests if you can, it's not the one I use so I don't know how good it is.
An A-die kit would be a good investment in your case, intel performs pretty bad if you have slow memory. You'll probably get a 10% performance jump minimum if you go to a 6400 kit and oc it to 7200, not to mention it'll probably be far easier to work with and find oc guides for A-die.

PC Specifications: Intel i9-14900KF, 5.7GHz all core locked, 5GHz ring, 1.35v Medium LLC, E-cores and HT disabled | MSI RTX 4090 Gaming X Trio | Arctic Liquid Freezer II 360 + Thermal Grizzly contact frame | 2x16 G.Skill Trident Z5 7400MHz 34-44-44-34 1T 1.45v (Maxed Subtimings, 52ns latency) | Gigabyte Z790 AORUS Elite AX | Windows 10/11 EVGA SuperNOVA 1000 T2 Phanteks P400A (Black non-rgb version, Phanteks T30 fans 3 intake (On AIO), 1 exhaust) | SK Hynix Platinum P41 2TB PCIe 4 SSD

 

Displays: MSI MAG 271QPX 1440p 360Hz 27" QD-OLED

 

Desktop Audio: STAX SR-007 MK2 Electrostatic Headphones (Current revision) | STAX SRM-400S Amp | Schiit Bifrost 2/64 (NOS mode, USB in, XLR out)

 

Mobile Audio: Sennheiser IE 900 IEMs w/ 4.4mm cable | FiiO KA13

 

Peripherals: Razer Huntsman V2 Full size wired with linear optical switch | Logitech G502 Hero

Link to comment
Share on other sites

Link to post
Share on other sites

14 minutes ago, ImWilly said:

I'm sorry, but do you refers to mix and match 2 different rated sped of RAM or testing different XMP profile on the RAM to see if the OC is safe or not?

 

When your rig is crashed, does it only happened on the related app/game?
Or is there BSOD?

If there is BSOD, you can check at 'Event Viewer' to at least know what is the  trigger of the BSOD

No they are the same kingston 5200 CL40 sticks, both increased to 5800 CL36 and then memtest'd. There is no BSOD when it does crash, just the whole thing freezing and then restarting.

13 minutes ago, rippy4500 said:

Use multiple tests if you can, it's not the one I use so I don't know how good it is.
An A-die kit would be a good investment in your case, intel performs pretty bad if you have slow memory. You'll probably get a 10% performance jump minimum if you go to a 6400 kit and oc it to 7200, not to mention it'll probably be far easier to work with and find oc guides for A-die.

I cant really get anything cause I am a minor and the whole PC was a gift (hence no GPU)

What tests can I use?

 

 

Just changed RAM to 5600 CL36

Link to comment
Share on other sites

Link to post
Share on other sites

8 minutes ago, ArushM said:

I cant really get anything cause I am a minor and the whole PC was a gift (hence no GPU)

What tests can I use?

I use testmem5. But as I said you'll be able to somewhat tell by the speed of the ram cause if it's unstable it'll be unusually slow. If you use the aida64 benchmark you can also see cache speed so you can judge ring frequency stability and speed too, increase it until it becomes slower then you'll know what your limit is on that. Also system agent voltage is very important for ring, but also memory if you do have the ability to get a faster kit some day, use like 1.15 to 1.2v if you want to increase uncore speed.

PC Specifications: Intel i9-14900KF, 5.7GHz all core locked, 5GHz ring, 1.35v Medium LLC, E-cores and HT disabled | MSI RTX 4090 Gaming X Trio | Arctic Liquid Freezer II 360 + Thermal Grizzly contact frame | 2x16 G.Skill Trident Z5 7400MHz 34-44-44-34 1T 1.45v (Maxed Subtimings, 52ns latency) | Gigabyte Z790 AORUS Elite AX | Windows 10/11 EVGA SuperNOVA 1000 T2 Phanteks P400A (Black non-rgb version, Phanteks T30 fans 3 intake (On AIO), 1 exhaust) | SK Hynix Platinum P41 2TB PCIe 4 SSD

 

Displays: MSI MAG 271QPX 1440p 360Hz 27" QD-OLED

 

Desktop Audio: STAX SR-007 MK2 Electrostatic Headphones (Current revision) | STAX SRM-400S Amp | Schiit Bifrost 2/64 (NOS mode, USB in, XLR out)

 

Mobile Audio: Sennheiser IE 900 IEMs w/ 4.4mm cable | FiiO KA13

 

Peripherals: Razer Huntsman V2 Full size wired with linear optical switch | Logitech G502 Hero

Link to comment
Share on other sites

Link to post
Share on other sites

7 minutes ago, rippy4500 said:

I use testmem5. But as I said you'll be able to somewhat tell by the speed of the ram cause if it's unstable it'll be unusually slow. If you use the aida64 benchmark you can also see cache speed so you can judge ring frequency stability and speed too, increase it until it becomes slower then you'll know what your limit is on that. Also system agent voltage is very important for ring, but also memory if you do have the ability to get a faster kit some day, use like 1.15 to 1.2v if you want to increase uncore speed.

So 5600 CL36 crashed, running 56000 CL40 now

The testmem5 page doesnt seem to be working correctly, IDK why but the download button doesnt download anything

How exactly can I use AIDA64 to check cache speed? Explain like I'm 5 please

System agent voltage was at default (IDK what it was) and now at 1.2V

Link to comment
Share on other sites

Link to post
Share on other sites

29 minutes ago, rippy4500 said:

I use testmem5. But as I said you'll be able to somewhat tell by the speed of the ram cause if it's unstable it'll be unusually slow. If you use the aida64 benchmark you can also see cache speed so you can judge ring frequency stability and speed too, increase it until it becomes slower then you'll know what your limit is on that. Also system agent voltage is very important for ring, but also memory if you do have the ability to get a faster kit some day, use like 1.15 to 1.2v if you want to increase uncore speed.

Also what is considered normal latency vs slow latency for both cache snd memory?

Link to comment
Share on other sites

Link to post
Share on other sites

2 hours ago, ArushM said:

Also what is considered normal latency vs slow latency for both cache snd memory?

Just use stock as a baseline, 4.5 seems to be the default usually for ring. Also set minimum and maximum uncore/ring speed to the same if that's an option.
For memory same thing, use stock xmp or a known stable oc as a baseline. 71ns is slow but for your memory speed I suppose it's not too unusual. We also don't know what die type your memory is, bios will usually say it somewhere or at least the manufacturer.

3 hours ago, ArushM said:

How exactly can I use AIDA64 to check cache speed? Explain like I'm 5 please

After you download aida64, click tools -> Cache and memory benchmark -> start benchmark. Do not have anything else in the background because it will interfere with the result.
You can also test memory latency by itself if you go to "Benchmark" on the left menu.

3 hours ago, ArushM said:

The testmem5 page doesnt seem to be working correctly, IDK why but the download button doesnt download anything

TestMem5 v0.12 (best configs).zip
This is the file I have. Also drag the "Extreme1" profile to the TM5.exe, it'll give a heavier/more accurate test than the default testmem5 settings. There are other profiles included but I don't use those.

PC Specifications: Intel i9-14900KF, 5.7GHz all core locked, 5GHz ring, 1.35v Medium LLC, E-cores and HT disabled | MSI RTX 4090 Gaming X Trio | Arctic Liquid Freezer II 360 + Thermal Grizzly contact frame | 2x16 G.Skill Trident Z5 7400MHz 34-44-44-34 1T 1.45v (Maxed Subtimings, 52ns latency) | Gigabyte Z790 AORUS Elite AX | Windows 10/11 EVGA SuperNOVA 1000 T2 Phanteks P400A (Black non-rgb version, Phanteks T30 fans 3 intake (On AIO), 1 exhaust) | SK Hynix Platinum P41 2TB PCIe 4 SSD

 

Displays: MSI MAG 271QPX 1440p 360Hz 27" QD-OLED

 

Desktop Audio: STAX SR-007 MK2 Electrostatic Headphones (Current revision) | STAX SRM-400S Amp | Schiit Bifrost 2/64 (NOS mode, USB in, XLR out)

 

Mobile Audio: Sennheiser IE 900 IEMs w/ 4.4mm cable | FiiO KA13

 

Peripherals: Razer Huntsman V2 Full size wired with linear optical switch | Logitech G502 Hero

Link to comment
Share on other sites

Link to post
Share on other sites

send thaiphoon burner screenshot for both sticks to identify ic cause overclocking blind is stupid when other ppl have done the work already and you can use other ppls overclocks as references or for slower speeds like <6400 plug and play, at the very least it isnt trash micron 16gbit rev a as it clocks past 5500 but im suspecting trash samsungs or something like that

 

overclocking blind is not gonna be very fun and will absolutely drive you insane as ive seen for myself trying to stabilize 2800+ triple channel on 32nm x58 on aircooling, actually did succeed and get 2832c10 stable ish but i was hoping for 2900 or 3000, oh and it took 3 days testing which sticks worked best as a trio cause some sticks really hate eachother, welp better than the two weeks or so that trying to stabilize 1520c7 ddr2 took =p

 

also stock voltage? not a good idea, you should be running as much as you need, thats probably why the rams arent stable at those rather slow speeds, cpu imc can give less of a shit regarding any ram voltages aside from the cpus internal imc voltage so no worries from the cpu side of things, means your only real limits are how far the ram scales alongside temps

 

imma summon @RONOTHAN## as he has experience with ddr5 and i only have rather general knowledge of ddr5 since im still stuck on ddr3

 

Link to comment
Share on other sites

Link to post
Share on other sites

16 minutes ago, Somerandomtechyboi said:

at the very least it isnt trash micron 16gbit rev a as it clocks past 5500 but im suspecting trash samsungs or something like that

There is some well binned Micron 16Gb A die that does boot and run some benchmarks at 5800, so it's not impossible. It's unlikely, but possible. The kit is also Kingston, and Kingston kits are usually either Micron or Hynix, so odds are it's more likely to be a really good Micron kit than it is to be a really bad Hynix kit, or even a really bad Samsung kit wouldn't make sense as even the worst Samsung kits do 6000MT/s. 

 

3 hours ago, ArushM said:

Is memtest86+ stable not good enough?

No, it isn't. MemTest86 in generally isn't that great of a memory stress test, especially on 13th gen Intel, as there's been plenty of kits I've seen where it passes on MemTest86 then fails in something like TM5, Karhu, HCI Memtest, or Y-Cruncher VT3. You want to run something else instead, probably VT3 and TM5. 

 

4 hours ago, ArushM said:

Thought that was the gold standard

It's the most recommended one because of its other advantages, not because it's amazing at finding errors. Because it runs off of a USB, it allows you to test the entire memory pool rather than losing ~2GB like you would if you're running a stress test inside Windows, making it in theory better at detecting faulty memory sticks (in practice though I've never seen a memory stick fail MT86 while passing one of the good in-Windows tests). The advantage I more agree with is that because its running off of a USB stick, it doesn't have a chance at corrupting data on your drives like bad memory with an in-Windows test can do (this still isn't that big a risk, though it is something that most overclockers have experienced at least once when stress testing inside Windows). It might have also been much better back in the older days of DDR and DDR2, though admittedly I don't have enough experience with those platforms to say for certain.

 

Timing or frequency issues like you'd get from manually overclocking your RAM though, it pretty much never picks up on, especially on DDR5. You need to use one of the in-Windows tests if you want to actually see those issues. 

 

 

Anyway, for how to setup TM5. I get it from the Overclock.net thread as it comes with a bunch of config files as well

https://www.overclock.net/threads/memory-testing-with-testmem5-tm5-with-custom-configs.1751608/

Go there, click the Download Link, then download that file from Mega. Extract the RAR file, open the TM5.exe, then click "Load Config and Exit." To select the config, go into that folder, go to "bin" then select one of the config files. The one I like is the 1usmus_v3.cfg as it's what I've had the most success with finding errors with DDR5, though you will need to modify it to have it run for a long enough time to get a valid stress test. After the config has been loaded, if you want to modify it to have it run for longer, go into the bin folder and right click the file you want to edit, then select "Edit with Notepad." When the notepad window comes up, Look under the Main Section for the line "Cycles" and change the value that is set to. I personally would run 30 cycles minimum with 1usmus_v3 or 3 cycles with the Extreme1 preset (1usmus cycles are significantly shorter than Extreme1 cycles, where each Extreme1 cycles is about half an hour while the 1usmus cycle time is usually between 3-5 minutes for 32GB of RAM), though feel free to go longer if you really want to be sure of stability. After the configs are setup, right click on TM5 on the desktop (the first time you run it, it automatically creates a desktop shortcut) and run that as Administrator. If you get errors the first time you run that as administrator, restart the system and they should go away. 

Link to comment
Share on other sites

Link to post
Share on other sites

I crashed trying to run 5600 CL40 too, running just 4800 CL40 now

1 hour ago, rippy4500 said:

This is the file I have. Also drag the "Extreme1" profile to the TM5.exe, it'll give a heavier/more accurate test than the default testmem5 settings. There are other profiles included but I don't use those.

No issues reported with this

1 hour ago, Somerandomtechyboi said:

send thaiphoon burner screenshot for both sticks to identify ic cause overclocking blind is stupid when other ppl have done the work already and you can use other ppls overclocks as references or for slower speeds like <6400 plug and play, at the very least it isnt trash micron 16gbit rev a as it clocks past 5500 but im suspecting trash samsungs or something like that

 

overclocking blind is not gonna be very fun and will absolutely drive you insane as ive seen for myself trying to stabilize 2800+ triple channel on 32nm x58 on aircooling, actually did succeed and get 2832c10 stable ish but i was hoping for 2900 or 3000, oh and it took 3 days testing which sticks worked best as a trio cause some sticks really hate eachother, welp better than the two weeks or so that trying to stabilize 1520c7 ddr2 took =p

 

also stock voltage? not a good idea, you should be running as much as you need, thats probably why the rams arent stable at those rather slow speeds, cpu imc can give less of a shit regarding any ram voltages aside from the cpus internal imc voltage so no worries from the cpu side of things, means your only real limits are how far the ram scales alongside temps

 

imma summon @RONOTHAN## as he has experience with ddr5 and i only have rather general knowledge of ddr5 since im still stuck on ddr3

 

How do I use Thaiphoon, downloaded it, now what

1 hour ago, rippy4500 said:

After you download aida64, click tools -> Cache and memory benchmark -> start benchmark. Do not have anything else in the background because it will interfere with the result.
You can also test memory latency by itself if you go to "Benchmark" on the left menu.

This result is with 5600 CL40 and 4.5 GHZ uncore and then 4800 CL40 4.5GHZ uncore (pictures below)

54 minutes ago, RONOTHAN## said:

No, it isn't. MemTest86 in generally isn't that great of a memory stress test, especially on 13th gen Intel, as there's been plenty of kits I've seen where it passes on MemTest86 then fails in something like TM5, Karhu, HCI Memtest, or Y-Cruncher VT3. You want to run something else instead, probably VT3 and TM5. 

Passed Y cruncher too, and when I lowered the clocks a ton still crashed so IDK anymore

54 minutes ago, RONOTHAN## said:

Anyway, for how to setup TM5. I get it from the Overclock.net thread as it comes with a bunch of config files as well

https://www.overclock.net/threads/memory-testing-with-testmem5-tm5-with-custom-configs.1751608/

Go there, click the Download Link, then download that file from Mega. Extract the RAR file, open the TM5.exe, then click "Load Config and Exit." To select the config, go into that folder, go to "bin" then select one of the config files. The one I like is the 1usmus_v3.cfg as it's what I've had the most success with finding errors with DDR5, though you will need to modify it to have it run for a long enough time to get a valid stress test. After the config has been loaded, if you want to modify it to have it run for longer, go into the bin folder and right click the file you want to edit, then select "Edit with Notepad." When the notepad window comes up, Look under the Main Section for the line "Cycles" and change the value that is set to. I personally would run 30 cycles minimum with 1usmus_v3 or 3 cycles with the Extreme1 preset (1usmus cycles are significantly shorter than Extreme1 cycles, where each Extreme1 cycles is about half an hour while the 1usmus cycle time is usually between 3-5 minutes for 32GB of RAM), though feel free to go longer if you really want to be sure of stability. After the configs are setup, right click on TM5 on the desktop (the first time you run it, it automatically creates a desktop shortcut) and run that as Administrator. If you get errors the first time you run that as administrator, restart the system and they should go away. 

tried doing this and got some critical debug error when selecting the config and the whole thing exited out

 

 

 

 

Sorry for the late reply went for lunch with my dad (happy fathers day to anyone it applies to) and spent some time with him

IMG_0819.HEIC IMG_0820.HEIC

Link to comment
Share on other sites

Link to post
Share on other sites

Also for reference right now I am running no OC whatsoever and RAM at 4800 Cl40

Link to comment
Share on other sites

Link to post
Share on other sites

3 minutes ago, ArushM said:

How do I use Thaiphoon, downloaded it, now what

1 hour ago, rippy4500 said:

Open the program, Hit the read button, then select one of the memory chips available. Do note though that Thaiphoon Burner has issues reading DDR5, so don't be surprised if it doesn't work (I've only ever had it read the SPD properly twice, and I have absolutely no idea why it worked those two times). 

 

4 minutes ago, ArushM said:

Also for reference right now I am running no OC whatsoever and RAM at 4800 Cl40

Alright, so there could be three things that cause this: The memory is in the wrong slots (though given that you were able to boot 5800MT/s I'd consider that unlikely), unstable memory corrupted the Windows install, or something is dead (likely from the factory as nothing you described should have damaged your components). Double check that the RAM is in slots 2 and 4, and if you're up for it do a Windows reinstall, and if neither of those work then its time to start trying to diagnose what's faulty. 

 

Also, did you run no OC by setting everything back to stock manually or by clearing CMOS? If it was trying to stock manually, it's very easy to forget something you changed and it's still running the OC, or there are sometimes even hidden settings that need a Clear CMOS to reset. Try to do it with the CMOS jumper as well, as the "load defaults" button in the BIOS isn't always 100% reliable. 

Link to comment
Share on other sites

Link to post
Share on other sites

7 minutes ago, RONOTHAN## said:

Open the program, Hit the read button, then select one of the memory chips available. Do note though that Thaiphoon Burner has issues reading DDR5, so don't be surprised if it doesn't work (I've only ever had it read the SPD properly twice, and I have absolutely no idea why it worked those two times). 

 

It worked

7 minutes ago, RONOTHAN## said:

Alright, so there could be three things that cause this: The memory is in the wrong slots (though given that you were able to boot 5800MT/s I'd consider that unlikely), unstable memory corrupted the Windows install, or something is dead (likely from the factory as nothing you described should have damaged your components). Double check that the RAM is in slots 2 and 4, and if you're up for it do a Windows reinstall, and if neither of those work then its time to start trying to diagnose what's faulty. 

 

Also, did you run no OC by setting everything back to stock manually or by clearing CMOS? If it was trying to stock manually, it's very easy to forget something you changed and it's still running the OC, or there are sometimes even hidden settings that need a Clear CMOS to reset. Try to do it with the CMOS jumper as well, as the "load defaults" button in the BIOS isn't always 100% reliable. 

I mean I just didnt load XTU, should be good enough right? As of right now though seems to be ok but also it did crash once with no OC and just RAM OC, I mean wait and see IDK. Im not the expert here

Kingston KF552C40-16.txt

8 minutes ago, RONOTHAN## said:

unstable memory corrupted the Windows install

Id rather not but I could

Link to comment
Share on other sites

Link to post
Share on other sites

2 minutes ago, ArushM said:

I mean I just didnt load XTU, should be good enough right?

Not always. Personally I'd uninstall XTU and do a Clear CMOS just to be sure (admittedly I hate XTU and would advocate for just doing everything through the BIOS, so take that for what you will). 

 

4 minutes ago, ArushM said:

It worked

Alright, you have Samsung 16Gb B die, that's pretty uncommon for these Kingston kits. It should do 6000 CL36 with ease though, so I'm not sure why you were having issues. 

 

 

Try to see if it crashes and report back to confirm. 

Link to comment
Share on other sites

Link to post
Share on other sites

3 minutes ago, RONOTHAN## said:

Not always. Personally I'd uninstall XTU and do a Clear CMOS just to be sure (admittedly I hate XTU and would advocate for just doing everything through the BIOS, so take that for what you will). 

A lot of people have said that is bad but honestly ive liked it and its infinitely faster then BIOS OC

4 minutes ago, RONOTHAN## said:

Try to see if it crashes and report back to confirm. 

Crashes what 6000 CL36, it wont even boot 6000 CL40, ive tried

Link to comment
Share on other sites

Link to post
Share on other sites

11 minutes ago, ArushM said:

Crashes what 6000 CL36, it wont even boot 6000 CL40, ive tried

Sorry, I meant confirm whether you're having issues with the system with no OCs at all. I probably should have made that a bit clearer. 

 

12 minutes ago, ArushM said:

A lot of people have said that is bad but honestly ive liked it and its infinitely faster then BIOS OC

It can be a bit faster, sure, though it does have other issues. If you really want to use it, use it just for setting up the OC, then copy all the settings into the BIOS. 

Link to comment
Share on other sites

Link to post
Share on other sites

26 minutes ago, RONOTHAN## said:

Sorry, I meant confirm whether you're having issues with the system with no OCs at all. I probably should have made that a bit clearer. 

No problem, yeah it seems to be fine, also could be some other BIOS setting though, LLC maybe? I set the entire BIOS to be stock

27 minutes ago, RONOTHAN## said:

It can be a bit faster, sure, though it does have other issues. If you really want to use it, use it just for setting up the OC, then copy all the settings into the BIOS. 

I will, any things I need to look for while using XTU?

Link to comment
Share on other sites

Link to post
Share on other sites

I think it was iGPU LLC but I will get back to you guys on that, how high should these dies clock to, tried 600 CL40 and crashed (1.35V stock)?

 

Update: it wasn't it still crashed, trying CPU llc now

 

@RONOTHAN## @Somerandomtechyboi @rippy4500

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, ArushM said:

how high should these dies clock to, tried 600 CL40 and crashed (1.35V stock)

Samsung B die I've seen upwards of 6600 working, though most top out at 6000-6200. The stock voltage of your kit is 1.25V, 1.35V should be necessary to get to 6000MT/s and above. 

 

1 hour ago, ArushM said:

it wasn't it still crashed

Was that before or after trying different memory settings? If it was before, what was the workload you crashed during? 

Link to comment
Share on other sites

Link to post
Share on other sites

1 hour ago, RONOTHAN## said:

Samsung B die I've seen upwards of 6600 working, though most top out at 6000-6200. The stock voltage of your kit is 1.25V, 1.35V should be necessary to get to 6000MT/s and above. 

 

Was that before or after trying different memory settings? If it was before, what was the workload you crashed during? 

So I just have bad dies?

 

The computer crashed on 4800 MHz so I started messing with other voltage related setting, I changed CPU LLC to mode 3 and iGPU LLC to auto. Seems to be ok for now. When it crashed I was doing literally nothing.

 

Running Y cruncher right now 5800 CL5800, haven't had any issues since the CPU LLC edits though.

 

Update: ycruncher came back clean like 3 times, then tried to load into OCCT and before I could even start it crashed for some reason (with full oc and LLC mode now 3, iGPU LLC auto, 5800 cl36). What now

 

Event viewer says that it crashed because of Kernel-Power (completely forgot this was a thing, shoulda checked it sooner)

Log Name:      System
Source:        Microsoft-Windows-Kernel-Power
Date:          6/16/2024 9:20:06 PM
Event ID:      41
Task Category: (63)
Level:         Critical
Keywords:      (70368744177664),(2)
User:          SYSTEM
Computer:      ArushDesktop
Description:
The system has rebooted without cleanly shutting down first. This error could be caused if the system stopped responding, crashed, or lost power unexpectedly.
Event Xml:
<Event xmlns="http://schemas.microsoft.com/win/2004/08/events/event">
  <System>
    <Provider Name="Microsoft-Windows-Kernel-Power" Guid="{331c3b3a-2005-44c2-ac5e-77220c37d6b4}" />
    <EventID>41</EventID>
    <Version>9</Version>
    <Level>1</Level>
    <Task>63</Task>
    <Opcode>0</Opcode>
    <Keywords>0x8000400000000002</Keywords>
    <TimeCreated SystemTime="2024-06-17T01:20:06.1773323Z" />
    <EventRecordID>12564</EventRecordID>
    <Correlation />
    <Execution ProcessID="4" ThreadID="8" />
    <Channel>System</Channel>
    <Computer>ArushDesktop</Computer>
    <Security UserID="S-1-5-18" />
  </System>
  <EventData>
    <Data Name="BugcheckCode">0</Data>
    <Data Name="BugcheckParameter1">0x0</Data>
    <Data Name="BugcheckParameter2">0x0</Data>
    <Data Name="BugcheckParameter3">0x0</Data>
    <Data Name="BugcheckParameter4">0x0</Data>
    <Data Name="SleepInProgress">0</Data>
    <Data Name="PowerButtonTimestamp">0</Data>
    <Data Name="BootAppStatus">0</Data>
    <Data Name="Checkpoint">0</Data>
    <Data Name="ConnectedStandbyInProgress">false</Data>
    <Data Name="SystemSleepTransitionsToOn">0</Data>
    <Data Name="CsEntryScenarioInstanceId">0</Data>
    <Data Name="BugcheckInfoFromEFI">false</Data>
    <Data Name="CheckpointStatus">0</Data>
    <Data Name="CsEntryScenarioInstanceIdV2">0</Data>
    <Data Name="LongPowerButtonPressDetected">false</Data>
    <Data Name="LidReliability">false</Data>
    <Data Name="InputSuppressionState">0</Data>
    <Data Name="PowerButtonSuppressionState">0</Data>
    <Data Name="LidState">3</Data>
  </EventData>
</Event>

Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×