Probably / Possibly Dead 5700G - Diagnosis help please

I had transferred my "old" 5700G socketed in an Asrock Steel Legend Motherboard to my business partner for her to use as her main office computer when her existing machine no longer was useable online. Other important specs are 32GB (2x16) Team Group 3600 18-22-22-42 1.35v RAM and a WD Black SN850 500GB OS drive. The PSU is a Season X-650 gold and there is no discrete GPU (outputting to two monitors from the board via HDMI and Display port to HDMI), which was one of the reasons for a "G" series chip in the first place when I got it back in 2021. I paired the 5700G with an X board rather than a B series, for full USB connectivity despite using a G series chip.

She got the 5700G because it has an active license for Office 2007, which is the most recent word processor she can tolerate using. I really can't lose that license and I don't believe I can activate a new copy of 2007 should I manage to find one.

Anyways, when we came back from a conference in Saturday, she went to the office in the evening and her computer would not boot. She called me to come to the office late at night saying the computer wouldn't turn on, so I presumed it was just either unplugged or It was plugged in and she, again, forgot which button was the power button on the case.

Upon getting to the office, the rainbow puke LEDs from the motherboard were cycling and visible through the unoccupied PCIe grate in the case, but pressing the power button would not do anything. I tried pulling out the RAM and re-seating it, one stick at a time in different slots, but no change. I tried re-seating the 24 pin and 8 pin MB cables, but no change. I pulled another PSU out of a computer to use that PSU's 24 pin and 8 pin and there was no change. I tried pulling the motherboard out of the case and putting it on carboard in the event there might be a short due to the board touching the case, but no change. When I pulled out the CMOS battery, let it sit for 10 minutes (after pressing the power button repeatedly), and then put the battery back in, pressing the power button on the case did cause the case's power light to turn on and the motherboard's chipset fan and the AMD box cooler fan (it's just a 5700G's heat load) to start spinning... but there was no further steps in booting and the Post Status Checker LEDs for CPU and DRAM were solid red.

I took the RAM home and tested the kit in my only other DDR4 motherboard, a working Z590 with an 11900k (don't judge it had a valid use case of best single-threaded performance at the time). Each stick separately and then together worked just fine with the Intel Z board. Today at the office, I removed the CPU cooler and unsocketed the CPU to check for physical damage to the pins and/or the socket. There was no damage I could see with my naked eye and no dust or gunk on either the socket nor touching the chip's pins. After resocketing the CPU, the motherboard's decorative LEDs came on but pressing the power button did nothing further. After removing the battery again, the fans would spin and Post Status Checker LEDs for CPU and DRAM were solid red again.

So, in conclusion, dead CPU or dead board or both?

Can you swap the 5700G in another system? Also, you could try using a single, slower stick of RAM, like a 2400MHz or 2666MHz. Sometimes something happens, and it justs needs a slower stick of RAM to get going.

loosen the timings on the ram and/or up its voltage a bit. could also try reseating the cpu and make sure you put a good amount of pressure on it as you lock it in.

El_Capitan said:

Can you swap the 5700G in another system? Also, you could try using a single, slower stick of RAM, like a 2400MHz or 2666MHz. Sometimes something happens, and it justs needs a slower stick of RAM to get going.

Click to expand...

No swapping available, my only other DDR4 board is an Intel Board: I have no AM4 boards to test the CPU nor Zen 3 processors to test the Asrock board with. My only DDR4 RAM kits are 3600 18-22-22-42 1.35v (The 5700G's kit) and 4000 18-22-22-42 1.40v (the Rocket Lake's kit). The 3600 defaults to 2400 when XMP is not enabled, or at least it did when I put one stick in the Rocket Lake system. I do not remember when the non-XMP specs are for the 4000 kit.

pendragon1 said:

loosen the timings on the ram and/or up its voltage a bit. could also try reseating the cpu and make sure you put a good amount of pressure on it as you lock it in.

Click to expand...

Earlier today, I tried re-seating the CPU as stated in the original post. That put the system's condition back to not doing anything beyond standby power running the decorative LEDs.. I obviously can't adjust RAM timings or voltage until I can get into the BIOS.

Teenyman45 said:

Earlier today, I tried re-seating the CPU as stated in the original post. That put the system's condition back to not doing anything beyond standby power running the decorative LEDs.. I obviously can't adjust RAM timings or voltage until I can get into the BIOS.

Click to expand...

as stated in my post, did you?:

pendragon1 said:

put a good amount of pressure on it as you lock it in.

Click to expand...

that was the point...


good luck.

Yes I put a good amount pressure in the seating and the locking.

Sounds like you've been pretty thorough, but have you tried running a single stick of RAM in the 5700G computer?

Jherek said:

Sounds like you've been pretty thorough, but have you tried running a single stick of RAM in the 5700G computer?

Click to expand...

Yes, to clarify from my initial post's "I tried pulling out the RAM and re-seating it, one stick at a time in different slots, but no change." The actual process in detail was: one stick at a time in each of the four slots, I picked one of the two sticks at random and started with the slot you are supposed to use (The A2 slot) moving in sequence through all four DIMM slots as indicated by the Steel Legend's downloadable user manual. Then I tried the other stick in the same pattern. This was done pre-removal of the CMOS battery and the computer would not being to boot. After pulling the battery and thereby clearing the CMOS, the Post Status Checker lights did not change when I also tried each stick in the A2 and B2 slots.

At this point, I just ordered a new 5700G since I really hope it's the chip rather than the board. I can't buy a reasonably priced new X570 board and am always leery of buying used motherboards. I am almost certain that the copy of Windows is a full retail license rather than an OEM one, but sometimes Microsoft's validation of Windows can be twitchy when swapping to a different motherboard rather than a replacement of the exact same model.

Good luck with that! Sounds like you've covered all the bases to me. In the olden days, the odds of a bad Cpu were minimal. It was usually the board. Not anymore.

An update. The new 5700G just arrived from Amazon. I unscrewed the existing box cooler without bothering to clean the NT-H1 paste I freshly applied yesterday, removed the old chip, socketed the new one, put the cooler with paste residue firmly on top and had one stick of RAM in the A2 slot. After attaching a monitor via an HDMI cable to the I/O panel, plugging in the 24 pin and the 8 pin, and then pressing the power button... it booted right into the BIOS after a bit of what was probably RAM training based on the PSC lights flashing all over the place.

Now to do a proper paste job and screw down the cooler, add a second stick of RAM, and a keyboard then see what happens. If all goes well I can put the SSD back in, run cinebench twice, set the memory to XMP, run cinebench again, put the motherboard completely back in its case, and hopefully call this matter resolved.

Well, so it was a bad CPU. Glad you got that machine going again. What is it with modern CPU's that makes them fail at rates unknown compared to years gone by?

Jherek said:

Well, so it was a bad CPU. Glad you got that machine going again. What is it with modern CPU's that makes them fail at rates unknown compared to years gone by?

Click to expand...

all the stuff thats now in them that used to be board side, like mem controllers.

Also, the tiny transistors have more leakage current and are more susceptible to damage from voltage transients, and defects which could cause failure after delivery are more likely.

Is their "auto overclocking" - running up the clock speeds to meet a thermal or power limit also just burning some out prematurely? If so, when I finally upgrade to a newer Ryzen, I may just run it in 65 watt eco mode.

My 8400F has been running fine with a slight overclock. I think it's just luck, mostly. If you get a good one, you're good. Get a bad one, good luck.

possibly. but probably luck^^

I had a 3770k CPU running from new till last year in my partner's computer when she was transferred over the 5700G. Bought it, I believe, during Thanksgiving 2012's Black Friday sale. The chip was fine, the software would not update any more with it hence... when I got a 9950X3D, my personal 7950X became my office PC and my office 5700G went to her. Before that, I hadn't lost a chip since a storm took my 3930k which Intel replaced under warranty.

Also, a little update on the paste side of things, when the CPU booted into the BIOS as there was no SSD or HDD installed at the time, with the basic cooler that comes with a 5700G meeting the new chip with just the residual NT-H1 paste and not being tightened down, the motherboard report 62.5 degrees C in a 24C room. My tube of NT-H1 paste only had a teeny tiny ball left so that plus the residual paste on the cooler and tightening down led to an MB temp reading of 58.5. Replacing that paste with the Alphacool Apex paste that came with my 5090's waterblock (and not doing an exceptional job in the application) lowered temps to 56C. When I actually booted into Windows, HWinfo was showing 35C.