PC Problems

Vamino

Insert coin(s)
vacBacker
Feedback
21 (100%)
Credits
2,606CR
You got them by the balls if the Ram isn't on the QVL mate, they should have checked that out beforehand.

Certainly mention that your sticks are not on the QVL and if the SN are not consecutive, also mention that they are not a matched pair.

You may get upgraded to some G.Skill ram, from my testing that's the most reliable ram imo.
 

Vorbis

Active member
Feedback
6 (100%)
Credits
579CR
They are the same part number but as far as I can see they aren't sold anywhere in a kit and don't show up in the qvl. They do show the same manufacture date in CPU Z and are running the default JEDEC timings despite being set to EXPO. I guess the only way to be sure it is an issue with them being mismatched is to remove one and check the memory speed
 

Vorbis

Active member
Feedback
6 (100%)
Credits
579CR
They'll probably claim it has 64 GB of 6000mhz ram but they didn't say it would run as such. If it runs 32gb at that speed then that's probably the issue but will wait for them to authorise me to try that as I don't want to give them any excuse not to sort the issue. The ram not being on the qvl even at 32gb is concerning though.
 

Vamino

Insert coin(s)
vacBacker
Feedback
21 (100%)
Credits
2,606CR
As you are probably aware, you will lose performance when testing with just one stick, and it will put a lot less stress on the memory controller on the CPU. I know ZX said kits these days should run on a potato, but running 2 x 64gb DR sticks at 6000 CL30 is not the same as running some 4800 CL40 DR sticks. I could make some suggestions on what has worked for me in the past, but best to let them sort it out as you are still in the 30 day return to base window.

Corsair are known to release the same modules with different heatspreaders with or without RGB lighting, this 64gb kit is on the QVL.

https://www.corsair.com/uk/en/p/mem...-cl30-amd-memory-kit-white-cmp64gx5m2b6000z30

Corsair are also known to change revision numbers on their kits, with later revisions having worse/different memory chips than the earlier revisions. is there a revision number on your sticks?

I'd be running some stress tests to prove to Specialist PC that something is wrong. Karhu memory test is one of the best, but it costs 10 Euro.
https://www.karhusoftware.com/

I run Karhu for 24hrs to prove stability.

I also use Y.Cruncher to test the memory controller on the CPU. VST or VT3 is the test to prove IMC stability.
https://www.numberworld.org/y-cruncher/

I run 2 versions of YC concurrently, an earlier version that has VST and the current version that has VT3. This isn't something I recommend anyone do as it puts tremendous stress on the memory controller and will make most coolers cry, but I do a 1 hour combo test to prove stability on my rigs.

I also use P95 blend test, and run that overnight.
https://www.mersenne.org/download/

I take stabilty very seriously on my PC's, a really unstable PC can pass memtest86 but will fail in seconds using Karhu or YC. Yes I go overboard and it's probably not necessary to most people, but I don't care what other people think as I just want a stable PC that has almost a bulletproof overclock that can handle anything I throw at it, even in a worse case scenario.

I've stuck with G.Skill from DDR1, and I've never had any issues having them run at their rated speed on half decent motherboards so I don't mind paying a premium for them.
 
Last edited:

Vamino

Insert coin(s)
vacBacker
Feedback
21 (100%)
Credits
2,606CR
I just want to make it clear that I don't recommend anyone running the tests that I run.

The CPU cooling on my rig can handle over 450w without thermal throttling, I delidded the CPU and am cooling it Direct Die using a DD waterblock with liquid metal, when you have that amount of cooling you can get too confident and it's easy to degrade a CPU at over 350w when CPU temps are in the 60's/low 70's. I've got a degraded CPU caused by benchtesting at over 450w to prove it! lol

If you can, just run Karhu memory test mate.
It's well worth the 10 euro.
 
Last edited:

mr grissom

Active member
Feedback
8 (100%)
Credits
167CR
Take a stick of the ram out, then run pc see how it fares.
then swap over ram see how it fares again, as there seems to be an issue with both in , id suspect not a matched pair timings may well be different on each one.
 

Vorbis

Active member
Feedback
6 (100%)
Credits
579CR
Weirdest thing. I turned off expo in bios, rebooted and then went back to bios and enabled expo and set to EXPO II (I gather this is the more stable version). Now according to CPU Z the ram is running at 3000mhz 4 x 32 bit which I assume is the same as 6000mhz. Not had a chance to run any stability checks but could it be this simple?

Dram status is still showing as 4800 in bios though which seems weird
 

Vamino

Insert coin(s)
vacBacker
Feedback
21 (100%)
Credits
2,606CR
Weirdest thing. I turned off expo in bios, rebooted and then went back to bios and enabled expo and set to EXPO II (I gather this is the more stable version). Now according to CPU Z the ram is running at 3000mhz 4 x 32 bit which I assume is the same as 6000mhz. Not had a chance to run any stability checks but could it be this simple?

Dram status is still showing as 4800 in bios though which seems weird

Yeah, it will show at 3000 in cpuz.

The 4800 in bios may just be relating to the JDEC spec.

I'd certainly run some tests before calling it a day.20251110_095155(1).jpg
 

Vorbis

Active member
Feedback
6 (100%)
Credits
579CR
Well that didn't go well. Definitely not stable.

They updated the bios which to be fair did improve stability but not to a usable level. Ran a couple of games and they ran slower than at the stock 4800hz. Just in the process of running memtest86 for them but already up to 73 errors so not looking good.

The don't stock any 64gb Ram other than this one but are sending me out 2 new sticks to try. If that doesn't work I think I'll give up.
 
Last edited:

Vorbis

Active member
Feedback
6 (100%)
Credits
579CR
So throughout my calls with them when updating the expo settings they've had me hard restart the computer when it hadn't rebooted after about 5 minutes. This hasn't felt quite right and a bit of googling tells me you're supposed to sit and wait as long as it takes. I did this and left it for a while and the machine did actually restart and the ram seems to now be running at a stable 6000mhz in EXPO. The weird thing is though, I seem to actually be getting slightly lower FPS than I was getting previously (around 210 in Cyberpunk rather than the 250). I think I have the same settings but I've tried so many different things it's hard to tell🙄. Maybe it's because I've enabled adaptive sync? I'll run memtest86 again tomorrow and see if I'm still getting errors but it's definitely more stable.
 

eliotcole

Active member
Feedback
7 (100%)
Credits
739CR
You mind me asking a couple of questions?
Totally don't expect/feel entitled to an answer, but the answers could allow me to figure out ways to get you more help, is all.

I've removed the original post as it's so far off topic (totally my fault). Could someone move this from for sale to the reliant forum area. Thanks

👍
New PC arrived this morning. RTX 5090 Ryzen 9950X3D 64gb Ram. I'm actually a bit scared to plug it in 😂
Only 3 questions ...
  1. SOURCE - Where did you buy from, was it:
    1. ONE SHOP - All one place, if so, where?
    2. ALL OVER THA SHOP! - Multiple places?
      no need to specify 😅
  2. BUILD - What 'type' of buy/build did you go for?
    1. DIY - A fully self-build/DIY
      ie. you've bought all your components separately irrespective of shop
    2. BUNDLE - A 'bundle' (ie. a set of components):
      1. Run-In - That's 'run in' (tested together - memcheck, etc)
      2. DIY - Set of stuff for a DIY build
    3. PRE-BUILT - A pre-built (ie. set of components includes more, like a case, etc):
      1. No OS- Shipped without an operating system
        1. Run-In
        2. DIY
      2. With OS:
        ie. components AND operating system
        1. Run-In
          ie. the OS is tested then set back to new user experience
        2. DIY
  3. OS - If there's an OS, which is it?:
    I didn't want to assume win, but this pic could confirm it
    1. Windows
    2. Linux
    3. None/Other
The reason that I'm asking about bundles and suchlike is that some places will put 'known good' combinations of motherboard and memory together, and some will go an extra step to ensure that the memory is 'QVL' and therefore ratified by the manufacturer to work with the MB. This doesn't mean a whole lot, in the grand scheme of things, but, it can REALLY help if you cannot get help from the vendor, but need to ask the manufacturer for assistance.
So, my last build would have been:
1.1
2.2.1
3.3
 

eliotcole

Active member
Feedback
7 (100%)
Credits
739CR
Also, on the ACTUAL gaming stuff ... might I suggest one thing? (again, no expectation/entitlement to being read/responded to)
So throughout my calls with them when updating the expo settings they've had me hard restart the computer when it hadn't rebooted after about 5 minutes. This hasn't felt quite right and a bit of googling tells me you're supposed to sit and wait as long as it takes. I did this and left it for a while and the machine did actually restart and the ram seems to now be running at a stable 6000mhz in EXPO. The weird thing is though, I seem to actually be getting slightly lower FPS than I was getting previously (around 210 in Cyberpunk rather than the 250). I think I have the same settings but I've tried so many different things it's hard to tell🙄. Maybe it's because I've enabled adaptive sync? I'll run memtest86 again tomorrow and see if I'm still getting errors but it's definitely more stable.

SET SOME BASELINES
So, to improve the ability for folks to help with all that ( and there might be better places 😲 ) , it might be best to set a couple of baselines.

Like ... for example ... whilst you are 'proving' your system:

  1. No Fakes - Turn off all the 'fake' stuff like AI / non-AI upscaling, frame generation, etc
  2. Test Runs - If you have games with them use the built in 'testing run' thing ... it's ... it's like a sweep through a city in Far Cry 6 or a pre-fought battle in Total War: Warhammer 1 & 2
  3. Optimise OS - If you ARE on Windows, make sure that you have everything all stable, and the same for each run, and try to ensure memory hogs like browsers are not running
Again, this just helps everyone get a firm grasp on baselines, and what happens WHEN things are changed.
That way, folks can help point out where benefits come in.




Final thing ... maybe start editing a bunch of what's been done/confirmed good, etc, in your first post. It would seem that you're at least working off of this thread as your fix log for this, so having that as your information repository ( and if it runs long, maybe the 2nd post to! 😅 ) will help you, too! :) 👍
 

ZedEx48K

8-Bit
Feedback
2 (100%)
Credits
1,342CR
So throughout my calls with them when updating the expo settings they've had me hard restart the computer when it hadn't rebooted after about 5 minutes. This hasn't felt quite right and a bit of googling tells me you're supposed to sit and wait as long as it takes. I did this and left it for a while and the machine did actually restart and the ram seems to now be running at a stable 6000mhz in EXPO. The weird thing is though, I seem to actually be getting slightly lower FPS than I was getting previously (around 210 in Cyberpunk rather than the 250). I think I have the same settings but I've tried so many different things it's hard to tell🙄. Maybe it's because I've enabled adaptive sync? I'll run memtest86 again tomorrow and see if I'm still getting errors but it's definitely more stable.

I'm presuming that's the memory training, is memory context restore enabled? It shouldn't take too long, but then I'm not sure on the average for 64gig, but 5 mins prob shows there was a fail.
 

ZedEx48K

8-Bit
Feedback
2 (100%)
Credits
1,342CR
You mind me asking a couple of questions?
Totally don't expect/feel entitled to an answer, but the answers could allow me to figure out ways to get you more help, is all.


Only 3 questions ...

  1. SOURCE - Where did you buy from, was it:
    1. ONE SHOP - All one place, if so, where?
    2. ALL OVER THA SHOP! - Multiple places?
      no need to specify 😅
  2. BUILD - What 'type' of buy/build did you go for?
    1. DIY - A fully self-build/DIY
      ie. you've bought all your components separately irrespective of shop
    2. BUNDLE - A 'bundle' (ie. a set of components):
      1. Run-In - That's 'run in' (tested together - memcheck, etc)
      2. DIY - Set of stuff for a DIY build
    3. PRE-BUILT - A pre-built (ie. set of components includes more, like a case, etc):
      1. No OS- Shipped without an operating system
        1. Run-In
        2. DIY
      2. With OS:
        ie. components AND operating system
        1. Run-In
          ie. the OS is tested then set back to new user experience
        2. DIY
  3. OS - If there's an OS, which is it?:
    I didn't want to assume win, but this pic could confirm it
    1. Windows
    2. Linux
    3. None/Other
The reason that I'm asking about bundles and suchlike is that some places will put 'known good' combinations of motherboard and memory together, and some will go an extra step to ensure that the memory is 'QVL' and therefore ratified by the manufacturer to work with the MB. This doesn't mean a whole lot, in the grand scheme of things, but, it can REALLY help if you cannot get help from the vendor, but need to ask the manufacturer for assistance.
So, my last build would have been:
1.1
2.2.1
3.3
A quick scan read of the thread would tell you all of this.
 
Top