I'd suspect a motherboard problem, there is a thread about a person with the same 970 board who also had problems. I also assume you are using a liquid cooler to go with your new board, else it's going to overheat.
I'll be using the new board in my main PC which has a custom water cooling loop for CPU and GPU, and it wouldn't overheat without water, it'd overheat without sufficient cooling and there are a number of air coolers which are up to the job, please don't try to scare people into thinking they need water.
1 of 1 people found this helpful
1) A robust CPU liquid cooling system. The Antec Kuhler H20 925, Cooler Master Nepton 120XL, and Corsair H80i are examples of products with sufficient cooling capacity.
1 of 2 people found this helpful
As noted previously in several other threads the referenced AMD Requirements are NOT in fact actual requirements. The original intent of the document was to make AMD FX-9000 users aware of three important areas that need greater capacity than what is required for the standard 125w FX processors.
Those three areas are:
1. A mobo that is specifically designed and approved for 220w AM3+ CPUs
2. 220w or better CPU cooling
3. Additional mobo VRM cooling via fan is often necessary to prevent overheating and throttling of the CPU frequency
I personally also recommend at least a 650w quality PSU and/or one capable of at least 54 amps. on the 12v rail. If you have dual GPU cards then you'll likely need a higher output quality PSU. JonnyGuru and Hardware Secrets do excellent PSU reviews so people can tell the quality PSUs from the not so great PSUs.
A liquid cooler is NOT required for FX-9000 series CPUs period ! This statement comes directly from AMD. In fact it would be useful if the incorrect "requirements" doc being referenced above were corrected to state what the AMD product group actually desired it to say/mean.
There are numerous top end tower style coolers that can properly cool an FX-9000 series CPU. The AMD product group has stated they have NO preference as far as the type of cooler used on the FX-9000 series CPUs. The important point is to use a CPU cooler that has sufficient cooling capacity to prevent the CPU from running hotter than 61C under full load. I have been using the Xigmatek Aegir SD128264 for years running 4.7 GHz. including numerous 25 hour Prime95 stability tests, without any issues at all.
Due to the fact that liquid coolers introduce a very real and unnecessary coolant leak liability to a PC, I don't recommend liquid coolers to anyone, especially the AIO/CLC models which have leaked, had water pump failures and have cause significant damage to other PC hardware. Due diligence would be to read actual customer issues in Corsair or similar user forums to understand the real world problems that exist with AIO/CLC liquid coolers. If after technically educating yourself you still desire a liquid cooler, go for it. You won't have anyone to blame if you suffer the same damage that many AIO/CLC owners have experienced.
Now to the OP's FX-9000 issues...
The Gigabyte 990FXA-UD5 R5 model does in fact support the FX-9000 series CPUs. If you do NOT have the "R5" version of the GA-990FXA-UD5 mobo, then your mobo does not support the FX=9000 series AM3+ CPUs. I believe Gigabyte originally stated that the GA-990FXA-UD3 / UD5 / UD7 all supported the FX-9000 series CPUs, but they probably learned the hard way they were wrong as the VRM circuits probably overheated and / or burned up resulting in the "R5" versions of these mobos.
If you do have the "R5" mobo then I's suggest that you will likely need to set all of the BIOS settings via manual mode as "auto" mode does not work well IME and that of many other people on the FX-9000 series and even on some FX-8000 series CPUs. The following settings have worked well for many FX-9000 users, overclocker's and folks with unstable 8-core FX processors.
1. Disable C6, C1E and Cool & Quiet - this will NOT hurt anything, they are just power saving modes
2. Manually set the CPU to the specified default vcore shown for your specific CPU
3. Manually set the DRAM timings listed on the DIMMs - assuming they were purchased as ONE DIMM kit and NOT two individual DIMMs. If bought as two independent DIMMs you likely will need slower timings to prevent issues.
4. If the DIMMs are 1.5v increase the voltage to 1.55v
5. Set CPU NB to 1.25v and HT to 1.25v-1.28v
6. Set Hyper transport frequency to 2600 MHz.
7. Set Northbridge frequency to 2400 MHz.
Applets like OCCT or similar that will record the CPU temp/vcore/frequency are very helpful in fixing operational problems
I don't believe I have the R5 gigabyte 990 FXA-UD5, I just checked the box and it's 'rev 3.0', I just checked the spec of the new board I've ordered and it states 'CPU's up to 125W'.
The spec for my gigabyte board says:
"Supports AMD AM3+ FX/ AM3 Phenom™ II, Athlon™ II series processors
8+2 phase CPU VRM power design for AMD high-TDP CPU support"
which is disappointing, and the 9590 is in their CPU Support List, so I will try it again today with the settings you recommend. I've also got some 40mm fans I can put over the vrm's and see how it goes, but I expect I'll be returning the Asus and get an R5
I reinstalled the 9590 into my gigabyte 990 FXA-UD5 today with the setting suggested, power extras disabled, manually set frequencies and voltages on the CPU and RAM, also tried XMP and Default for the RAM. I fitted some fans to the vrm heatsink to ensure good cooling, it's right next to an exhaust fan anyway. I initially tried booting into my main OS but then when it was similarly unstable I switched to my fresh Windows 10 install disk.
I tried CPU voltages from 1.4 to 1.55 and ram voltages from defaults to 1.5 to 1.55 but I was getting bluescreen crashes at windows login screen 'IRQL not less or equal', that's when I switched to XMP profile for the ram and also defaults but still no-go.
My supposed replacement motherboard came today but as I found out it's only good for 125W cpu's I've asked the supplier to RMA and arrange a Gigabyte 990FXA-UD5 R5. Hopefully that will be arranged soon and hopefully it will solve my problem, if it doesn't then it's looking like a bad CPU
I had some trouble finding good info on CPU voltages, I found people reporting success with 1.45, 1.48, 1.5 which I've tried, CPUboss lists that cpu as 1.91-2v, which seems high so I didn't try it.
Unfortunately Gigabyte has had issues with the VRM designs on both AMD and Intel mobos in recent years. The R5 model appears to be a suitable design for the 220w minimum power required for the FX-9000 series CPUs. For two years Gigabyte Tech Support assured me they had no issues with their earlier AM3+ model mobos even after I explained to them how to test and observe the CPU throttling the overheated VRM circuit caused to CPU performance... with any 8 core FX processor.
As far as vcore you definitely do NOT want to be above 1.55v. When you boot your system and go into the BIOS it will show you the true default vcore voltage for your CPU. If it isn't listed in the CPU section you can temporarily switch to "manual mode" then the default vcore will show and be set by default. You can record the vcore voltage and not exceed that value if you run in manual mode. AMD does extensive testing of all processors immediately before they are boxed. Unless the processor has physical damage it's highly unlikely the CPU is defective.
Hopefully the new mobo will resolve the issues.
Thanks buddy, that's reassuring.
R5 board should be here today and I may be able to fit it over the weekend, I'll try default settings as well as those suggested above and see where I get, I'll report back results here.
Okay I had a 9590 at one time . It definitely needs some kind of water cooling or AIO . Also turn the turbo core off on the board - Not all the 9590 can hit the 5.0GHZ and stay stable. Mine could run between 4.7 to 4.8 stable. Make sure your VRMS are kept cool as well. The VRM on the mother board will heat up more then most from this processor. Good luck
Thank you, I have a custom water cooling ready for this chip and I'll be adding three 40mm fans to help keep the VRM's cool.
I'll be trying without the turbo, probably all manual settings initially but once stable I will try letting it go to 5GHz.
Currently running with new motherboard using default settings with disabled extras, no C&C, no turbo, etc etc.
9590 in Gigabyte 990FXA-UD5 R5 normal use so far min 34.5c max 59.1c
I'm not sure of those temps are normal but I noticed the flow-rate in my system is quite slow so I'll be overhauling it in a few weeks.
I may have time to torture it tomorrow and I'll generally get a better idea of stability over the coming week, I will report back again if I have more problems or to report it stable.
First the good news, it has been working faultlessly since getting the new motherboard with basic settings (plus XMP for the RAM).
Now the bad news, I noticed an interesting burning smell yesterday morning, but I couldn't pin it down, however I came home to find my computer frozen and trying to restart it was met with only a blank screen, no POST, no beeps, nothing but spinning fans and silence,
to cut a long story short I found some burnt-out wires in the plug feeding the CPU 12v, see the plug here, and inside the socket here, all of the 12v+ wires were burned through and one of the pins had melted it's plastic shroud. I have no idea what happened, maybe a bad connection caused it to overheat?
I've replaced the motherboard and cpu with my old 8350 and it's working fine, no damage to the RAM or PSU or anything else, I'll be trying to clean-up the plug socket and see if I can get the board working again, if not I'll put the 9590 in my cant-cope-with-that-much-juice motherboard I'm currently using to see if it's still alive, hopefully it's going to require only a replacement motherboard, which I can manage with a little saving, if the CPU is also dead then I'm very unlikely to be replacing both.
So, if anyone has any greater insight than I, please share, if anyone else is running a 9590, it's probably wise to check if your CPU power cable is properly seated as my best-guess is that was the problem. Thankfully I was using an extension cable that came with my case so I don't have to replace the PSU plug, it only just reaches without the extension.
The CPU and Board are still alive, I'm going to need to replace the 8-pin power socket on the board due to damage to be sure of a good connection, my guess is still overheating due to poor connection, somehow...
1 of 1 people found this helpful
Wow, definitely report that to Corsair and get that RMAd, do not try and use a PSU which has had a fault. That power supply is rated for several times the wattage you were pulling, and the CPU doesn't pull everything from the EPS8, some comes from the ATX24's 12v supply, so it's most likely there was some kind of delivery side fault with the PSU, it happens even to top tier manufacturers. This is also why you should -always- shut down, or at least sleep, a computer when you are away for an extended period, as that could have easily evolved into an electrical fire.