• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

AMD FX8350+M5A99 Evo R2 -> Persistent System FREEZE

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

...READ to READ Delay is in your bios ram settings. it is 4 now and would try 5. It will not hurt to go 6 but should not need to do so since changing 4 to 5 is likely enough.

That one setting by itself is what I use to run 2x4gig AND 2x4gig GSkill DDR3-1600 Ares and DDR3-1866 Ares together in my CHV-Z mobo. I have one brand new CHV-Z that has 4x4gig GSkill Ares 2400 in it running at DDR3-1866 set by "auto". So having that setting available is a HUGE bonus for mis-matched ram sets and 4 stick kits.

I would up ram voltage. I would up CPU-NB a little at a time. I have to run mine at 1.23V on one mobo and another board with another FX 8 core it takes 1.287V on CPU_NB. There is NO one size fits all. I make adjustments and then run P95 Blend mode for at least 20 mins and view HWMonitor Free for Max Temps/Volts.

So set READ to READ Delay to 5. Adjust ram voltage upward. 1.6V is not harmful. But I would try 1.55V first and with CPU_NB at 1.25V. Do at least 20 mins of P95 Blend. I would try this with 21x multiplier on cpu and that would be 4.2Ghz BUT it will have ALL 8 cpu cores on at one time at that same overclocked speed. That is NOT how AMD specs it and much more power hungry and HOT. So am not suggesting higher than 4.2Ghz to begin with. I would set CPU voltage to 1.35V for first testing, but may or may not take a little more or get by with a little less. These later FX 8 cores often can get by with less cpu voltage than 18 mos ago.

Think that is what I would do first and then give us the captures of CPU-z > CPU Tab > Memory Tab > SPD Tab and finally a capture of HWMonitor Free while it has been running logging Min/Max Temps/Volts during at least 20 mins of P95 Blend.

Luck man.

ok I did as you suggested.
RAM @ 1333 I also filled in tWR, TRRD, TRFC, tWTR, tRTP timeings manually, the auto-set timings were WAY to tight compared to the manufacturer specifications of the RAMs

CPU @ 20multiplier (stock) and went up with voltage setting successively till 1.35. -> Blender froze always after like 30 seconds and for 1.35 also Temp went up very quickly

tried something different, went back with RAM@1066 (still manually increased NB voltage, [email protected] and manually set timeings)
CPU @ 20 multiplier (stock) but now low voltage...1.225V -> Blender stable for 20min

more interestingly I went up with multiplier (now 4.3 GHz) and set voltage to 1.285V again blender stable for 20min with reasonable temps...maybe with some further tweeking temps could be more towards green zone at same clock.

seems I got one of those low voltage revisions of the fx...

But 1066 RAm is probably seriously bottlenecking...1333 is even not optimal...but I think it is safe to assume the major cause of my instability issue is the RAM and there is no need to RMA board or chip which is good news and you guys gave me a lot of useful information.

Edit: set core [email protected] Ghz also stable for 20min blender finished at CPu 57° Package 41°
Last edited:
64c socket and 54c on the package at 1.32 Cpu V? Something doesn't seem right. Celta please add a signature to your posts it will help us help you if we do not have to go back to the first post to see what is in the computer. To do so click on settings and edit signature, then add the components in the rig. What are your ambient temperatures? What case and how many fans?
Get a better case and a high end CPU cooler, that case you posted is trapping a lot of heat in it, also you need better cooling on that CPU.
64c socket and 54c on the package at 1.32 Cpu V? Something doesn't seem right.

Looking at the temps and 'having' the long expected delta points me to something I have noticed in these cpus that take lesser Vcore than expected...they are h0t. I have thought this and it is hard to catch the...make that the many changes we have seen in the FX cpus over the last 10 months or so.

Remember the ones in the last year that would not do anything 'as expected' and the user had to settle for what he/they could get? Been n0 more than four of those that I can remember. That is not a lot of such...but knowing they have been there and then this one would make another. Well let us say I begin to think we are seeing another.

I mean the ram is either so badly incompatible or the cpu IMC is a bow-wow or some combination of same. And then those 'overall' high temps but with the correct delta and Low Vcore? Makes me nearly want to get in touch with AMD and tell 'em I got a dog. Well some statement of problem that they might do an RMA for, but if ram seems wonky, too much heat and cpu might be just stupid...well that is nearly cause for a whole new setup.

I think "celta" is going to have to get stable at some speed with ram at whatever it is that seems to work and then make up his mind if he is fast enough and then know it is going to take "more" parts. Sadly that seems to sum it up...well to date in my mind.