Did you had a computer, serviced under warranty services to see that the technician sucks dicks and mess up your computer to behave worse than when it was given for upgrades?
I had such issue with my Lenovo Thinkstation D20. I gave my Lenovo for warranty services upgrades of RAM, with clear and specific instructions what i want done, instructions based on Intel's ark website for the CPU at hand, found here:
https://ark.intel.com/content/www/us/en/ark/products/52576/intel-xeon-processor-x5690-12m-cache-3-46-ghz-6-40-gts-intel-qpi.html
Based on the CPU specs as noted by the manufacturer, i explicitly requested 12 sticks, 8 GB per Stick of ECC, DDR3, 1333 Mhz RAM. This exact RAM specifically. I also Specifically and explicitly mantioned i want 12 identical sticks - same make and model, same manufacturer, same specs.
where it is clearly stated, that the CPU can manage DDR 3, ECC, 1333 Mhz RAM. So based on the Lenovo's specifications, found Lenovo's Support website, it was stated that after a Bios Upgrade to 2 versions before last, the Workstation becomes capable to support and manage 96 GB ram, using 12 RAM sticks of 8 GB per stick, separated in two pools of 48 GB per pool - 6 sticks, of 8 GB per stick, one memory pool for every CPU. For those of you who struggle to figure it out by themselves alone, i use X58 Server grade chipset with support for 2 CPUs, and both CPU sockets are in use, i run 2 CPUs both are Xeon X5690 - also server grade CPU.
Well i updates to the latest BIOS, so BIOS is now ok. This is where my visit to service came in play. I went to the service, and i requested upgrade or RAM, replacing 12 sticks of 4 GB per stick of ECC, DDR3 ,R1, 1066 RAM, with 12 sticks of 8 GB per stick, ECC, DDR3 1333 MHZ, using Intel Ark site for a guide what RAM to select.
2 days later, i took my computer "done and ready" to find my Linux crashes or random. WTF? it did not crashed before the upgrade, it was stable as a rock. I also notices a bit sluggish performance issue - it is not as snappier as before the upgrade. WTF? Upgrading RAM this way should not slow it down, yes i will not feel bombastic speed upgrade, but it will be able to handle much heavier workloads, without performance penalty.
I pulled my bag of tricks - my bag of bootable USB drives.
I used mu distribution's boot usb, to run the live environment and test. I verified my Linux install is not corrupted.
I used another bootable USB, and I ran GRC's Spinrite 6.0 on Level 2, which is extensive READABILITY test, WITHOUT any writing, verified my SSD is not corrupted,
I rebooted, I used another bootable USB with Hiren's boot DVD, and trimmed my SSD, Then i ran Intel CPU test. The test recognized chipset and CPU setup correctly.
Test crashed. Were Intel Xeon beasts defective? Not likely - Xeon is tough beast, but not impossible.
I removed all the ram and i installed some spare non ECC sticks of DDR 3, 1333 mgz, 2 sticks, 4 GB per stick, one per every CPU pool. I ran Intel CPU test again. It passed with clean bill of health and lower temps, compared to previous setup. Another WTF, how i get lower temps on same workload, for the same duration, and same intensity, same coolers, same fans, same thermal paste, in the same room and same ambient temperature between tests? Well being left without much to do except complaint to service for bad work, i said to my self, f*ck these c*cksuckers, these c*cksuckers are useless pieces of shit.
I decided to see that else they screwed on my computer - going into BIOS, password was placed. I called to ask them for the password, and ask them why they lock me out from my computer. They said "This is a system ares and we reset it to fail-safe defaults, and we locked it, for your own good. I asked what else they did, they said they re-pasted the CPU's, explaining the brand of paste they used. It was a cheapo-crap piece of bulk paste sold i 5 liter buckets to cheap cents. Thermal conductivity around 1.3 watts/Meter-Kelvin - for the non-initiated, in the cooling this crap is so shitty that CPU cools better without thermal paste, compared to this cheapo crap. Ok, I pushed the clear CMOS button to reset CMOS in factory defaults, i examined my fans, all of them were set in non-optimal configuration - front and rear fans acted as intakes and CPU cooling fans were facing up. Well There is no laminar flow in the case to ensure proper cooling, and fast expel of the heat.
Well, i took the CMOS battery out, and i removed all fans, all heatsinks, thoroughly cleaned the CPU and heatsinks, I went out i bought some decent thermal paste - Thermal Grizzly Kryonaut Extreme, which is premium grade, premium performance thermal paste - 14,2 watts/MK, it smokes the living daylight out of the cheapo poop i found on my CPU's after the technician.
I Re-pasted my CPUs again, installed the heat sinks in their correct orientation, and i reinstalled all fans to their correct orientation, so the front is intake, the rear is exhaust, and both CPU fans blow in the same direction as the general airflow created by the intake and exhaust fans. there now i have good paste, and optima cooling based on even laminar airflow, trough the case, expelling heat faster, and maintaining even lower noise, and cooler case and cooler processors.
I placed the CMOS battery back, i went into BIOS, and readjusted power options, turbo ptions, P states and C states as recommended for Westmere generation CPU, again information on that - ark.intel.com for the CPU in question.
Well so far so good. I put 8 GB ram again, the 2 x 4 gb as known good ram and compliant to specs RAM, intel cpu test for 190 minutes buning torture - My CPU's were much cooler - 58C on the package for CPU 0 and 54 C for CPU 1.
Well i put the RAM from the warranty service again, ran tests again intel CPU test crashed. It took it deeper into the test, it took it longer before it crashes, but it crashed. CPU temps: 68 C on the packace for CPU 0 and 64 C on the package for CPU 1. WTF?
Then i pulled one more USB flash drive, running Memtest X86. This is where it all came clear - all 12 modules were random mix of ECC, DDR3, 1600 and 1866, Neither of which are supported by the CPU and they are mixed speed, mixed settings mixed timing, mixed steppings. So The CPU used the lowest common denominator to make them work somehow with non-compliant RAM. The slow down is coming from mixing and matching speeds to the lowest common denominator, the crashes came because none of the sticks were supported as noted in the ark.intel.com specs for the CPU. All of the 12 sticks were unsupported.
I collected all the evidence i could, i filed a complaint against the service into authorities, and with the same evidence i demanded my money back accusing them for being dishonest cheap cheats, and telling them that i already gave all evidence to authorities and i already have the complaint filed and working. Well they did not gave me my money offered free repairs, i asked why? to screw up my computer again? i just rebuilt the whole thing to get it into correct build. I demanded my money back and i told them i will not recall authorities. Well Waiting for authorities to come and make their judgement.
I just purchased 12 identical Samsung RAM Sticks of ECC, R2, DDR3, 1333 Mhz RAM, as recommended on ark.intel.com and the RAM Sticks are in delivery. Once they arrive i will replace them, and i will continue to work with authorities to get my refund. As for incorrect modules, i will give them back to the service. I do not need them, and selling server grade RAM is not easy. It does not run on consumer grade computers, it is not supported, the consumer grade PC will not even post with ECC server ram, it does not know what is this, what to make of it and how to use it.
This is the mentality of Bulgarian PC repair shops.
This is the quality of Bulgarian PC repair shops.
