APU 1 core is not holding clock multiplier?? CPU, MOBO, or RAM problem??

[ramstock]

Hello, I'm new and this is my first post. I have built systems for over 20 years but over the last 7 years due to circumstances I bought CTO systems. I decided I was missing out on what I enjoyed to do and ordered the parts for my rig after much research and desired need/use for my system. Before becoming disabled I had resources to buy the best and latest components so overclocking wasn't a need as I always had the best system money could by.

To start off technology has changed since my last build thus all the research but still encountered several hardware related issues. Defective mobo or defective ram or cpu/apu that is the question.

Things have changed and OCing and manufacturers encourage it though warranty can be void; noted though Ocing especially Ram is necessary for achieving rated specs. Gone are good old days when you could mix match ram and still have a stable system.



Followed everything by the book. Actually disassembled an older DIY build and rebuilt it just for practice. All ok.
New build 1. Asus A88xPro mobo wouldn't flashback update bios for Godavari chip. Assembled minimum components to see if sys would post and load OS and software to update Bios.
2. Installed G.Skill PC2400 8gtx cl10 2X4gb in recommended dual channel slots.
3. System posted setup bios defaults and was able to update bios.
4. Win 7 64 bit Ultimate- during updates system kept crashing (note: had a good friend C.S. grad, IT tech
assisting). Restarted system and OS picked up where it crashed and started reloading updates. Crashed again.
5. Cleared CMOS and flashbacked bios. Bios default mem SPD 1600mhz... same issues so removed ram and used 1 dimm in slot a1. Bang all ok!! then added dimm2, then constant sys crashes, multiple cmos,bios, os reinstall etc.
6. Called all but AMD tech support, Corsair-set to performance no sleep state;check, Samsung same-set to performance. G.Skill best ram inexperience in timings so I thought OK I'll Just use JEDEC spd defaults to test.
7 Got system to run stable for several hours loading software with 2 dimms installed. Then system fatal shutdown. MB temps were extremely hot chipset heatsink and vrm over 100 F even after about 2 minutes time to go get a temp probe.

IT friend said it's bad mobo dimm slot, mobo co suggest it could be bad ram or cpu memory controller or all said bad PSU. Tested PSU on another system for 36 hours no problem.

Ram memtest showed Ram spd timings different though ram was a kit same build, version code., etc.
TESTED CPU AND CORE 00 or I refer to as core 1, throttling from 17 to 39 when multiplier was set manually to 39.

RMA'd mobo and upon removing CPU noted that Heatsink backplate was noticeably over torqued compared to replacement board and after removing Heatsink the CPU was stuck in socket. Moved slighty side to side no give.]
Then got a small flat head screwdriver and put upward pressure on pcb and it eventually popped out.

Waiting to get thermal compound to reinstall components and test hoping no damage to CPU or is already damaged or is it normal for 1 core when all cores are set to a manual multiplier; to throttle over 100% variance.

Any comments, advice, questions, opinions, are greatly appreciated. I was certainly enjoying it for the few times it was working ok, especially with the WIFI speeds I was getting. My IT Tech friend, an INTEL fanboy was impressed with the AMD APU. But as he said I don't have any issues with my Intel systems I own or service.

Now, disabled my resources are limited and never before had to RMA more than 1 item and then I just bought a new one and kept replacement as a spare. I don't have that luxury now . SO PLEASE HELP ME with what steps I should follow before I proceed.

Thanks for your patience and understanding.

 

[ShrimpBrime]

Man oh man a list of issues I see here.

Core 1 throttles while manually set x39 multiplier - No mention of power savings being enabled or not. Should disable these if overclocking, disabled in windows and bios. 2ndly, shouldn't OC a build before you install OS, this should be done all stock values. Load bios defaults and continue....

Memory in single channel one stick works, yet tested memtest and come up with different values, Is this during overclock or stock loaded default values or clear CMOS (you mentioned clearing cmos, but not exactly when... before or after the overclock or before and or after / during installation of w7.

Then we have big temps on the NB and VRM package areas... A good solution is to actively cool these areas of the motherboard with a fan or two to ensure temps stay in check during the OC process.

The Cpu was stuck in the socket. Stupid question maybe.... but did you flip the retention bracket up? If so what exactly happened? Did the Cpu suffer physical damage to the pins?

Mention of the "few times it worked ok" ..... was "ok" meaning working normally except core 1 was throttling?

Is the Ram tridentX F3?

2400(O.C.)/2250(O.C.)/2200(O.C.)/2133/1866/1600/1333 MHz Non-ECC, Un-buffered Memory Anything over 2133 is a OC on this system probably requiring some manual set up action for proper workings.
Really don't need more than 1866mhz Ram - AMD likes it fast with tight timings, so Cas 9 @ 933mhz is going to be just fine.

System should boot memory stock 667mhz at 9-9-9 timings. If Ram and IMC don't like high speed ram and these timings aren't usually part of 2400mhz Ram timing sets as it's meant for high end overclocking, generally not seen much on APUs, and I'm not saying it's not possible, but likely the TridentX isn't listed on QVL.

The RMA for the motherboard is a good idea. Perhaps the same for the cpu if your in question, likely it's not damaged beyond some bent pins, but you never know.

Also a list of system specs. Is there a discrete graphics card in this system? Or using the iGPU and sharing memory with the system......

Dunno!! You got a gremlin system problems for sure.......

GLHF!!

 

[shawnmcc]

First thing is reapply thermal paste if it looks like you used to much. I had the same problem till I put less thermal paste on my cpu. Plus I used some sand paper to take some brown marks out on my heat sink on the bottom. That helped to I think.