Question FX9590 intermittent no post -- but runs fine once started. Ruled everything but CPU. ?????

[DONINAUSTIN]

Asrock 970a-g/3.1
AMD FX-9590
NvMe SSD
Old generic video card. (PC is business use)
8 gigs Corsair ram
Problem exists with all drives except SSD unplugged.

Intermittent failure to post. Not even video. At one time the fans were clicking and cycling but that has stopped. I have swapped memory, video card, SSD, MB, 4 different power supplies 500w to 1000 watt.
No problem with temperature -- this is not a gaming PC. What is strange is that once I get the PC up and running it can go days or weeks with no faults or crashes. No problems found running burn-in software.
So it seems to come down to the processor. Today i underclocked it from 4.7 to 4.0 and reduced the turbo speed. This MAY have fixed it but it has been unpredictable so hard to say.
But can a CPU be flaky in such a way that the ONLY symptom is failure to post?

 

[Abwx]

Honestly. Let's not derail the thread too much, but:

https://www.overclock.net/forum/10-amd-cpus/1496284-huge-issues-9590-stability-2.html#post22433650

9370s and 9590s had such high leakage that they couldn't be sold as 83xx chips. Trash bin material. So they sold them to enthusiasts who actually wanted the high leakage. .

High leakage mean that the chip has the capability to work at lower voltage at a given frequency and hence to be more efficient with comparable leakage at the end of the day...

So it s actually better chips but AMD couldnt change the voltage/curve of the FX8350, and shift the voltage by say -6%, because of the big installed base, the only mean would had been to release a new 4GHz SKU with a different voltage/frequency curve, wich they more or less did with the FX8370.

That s for whom is a litte aware of semiconductors physics, others will be left creating urban legends as a mean to appear knowledgeable, but certainly in no way accurate, not even semi...

 

[chrisjames61]

It is supposed to support FX9590

In this case "support" and function properly are two totally different ballgames. All this stuff was discussed at Overclock.net and Overclockers.com years ago.

 

[Abwx]

https://www.computerbase.de/2013-07/amd-fx-9590-prozessor-test/9/#abschnitt_undervolting

They use an Asus Crossair Formula V wich was released well before the 9590 and wich is supposedly limited to 140W.

https://www.hardware.fr/articles/905-2/am3-220w-turbo-frein-overclocking.html

Asus Sabertooth, also limited to 140W, and 8+2 phases, same as the OP s Asrock..

 

[DrMrLordX]

High leakage mean that the chip has the capability to work at lower voltage at a given frequency and hence to be more efficient with comparable leakage at the end of the day...

If you put a 9590 and an 8370 side-by-side and tuned them to run at 4 GHz at the lowest possible voltage for each chip, on average, you know which one pulls more power? The 9590. Because it sucks. That. Much. Sure it'll do it at a lower voltage. It'll still pull more power from the socket.

The only advantage to those chips is that they consistently hit clockspeeds of 5.0 GHz and higher moreso than other dice from the same wafer. Period. Otherwise they are crap. Don't give me this "9xxx were superior!!!" nonsense. It just isn't true. Wasn't then, isn't now. It wasn't even true after the late 2014 process refresh of 32nm.

 

[Fanatical Meat]

In this case "support" and function properly are two totally different ballgames. All this stuff was discussed at Overclock.net and Overclockers.com years ago.

and back to my point about all those AMD chips & motherboards being complete crap.

 

[chrisjames61]

https://www.computerbase.de/2013-07/amd-fx-9590-prozessor-test/9/#abschnitt_undervolting

They use an Asus Crossair Formula V wich was released well before the 9590 and wich is supposedly limited to 140W.

https://www.hardware.fr/articles/905-2/am3-220w-turbo-frein-overclocking.html

Asus Sabertooth, also limited to 140W, and 8+2 phases, same as the OP s Asrock..

You obviously don't have a clue about AM3+ boards if you think that ASRock board is up to par with the Sabertooth Gen1 or R2. Comparing the original posters board with either of the Sabertooths is ludicrous.

 

[chrisjames61]

and back to my point about all those AMD chips & motherboards being complete crap.

There were a lot of good AM3+ motherboards. You just had to buy the premium boards. The chips weren't crap either. Were they as good as what Intel had? No. Nobody will argue that point. They filled a certain niche in the market and if you knew what you were doing you could wring good performence out of them.

 

[Fanatical Meat]

There were a lot of good AM3+ motherboards. You just had to buy the premium boards. The chips weren't crap either. Were they as good as what Intel had? No. Nobody will argue that point. They filled a certain niche in the market and if you knew what you were doing you could wring good performence out of them.

That says crap to me but we can agree to disagree

 

[Headfoot]

that whole line were turds, and maybe if you really polished the turd super hard it would shine a little. But it was still a turd. Meanwhile, my overclocked 5820k on the cheapest X99 board with high end memory from late 2014 is still within 10-20% of even overclocked 8700ks in gaming.

 

[Abwx]

If you put a 9590 and an 8370 side-by-side and tuned them to run at 4 GHz at the lowest possible voltage for each chip, on average, you know which one pulls more power? The 9590. Because it sucks. That. Much. Sure it'll do it at a lower voltage. It'll still pull more power from the socket.

.

What sucks is the clulessness about what is compared, hence your non sense..

The FX8350s manufactured after mid 2013 have higher leakage than those manufactured with the previous process, actually leakage almost doubled, and indeed they overclock much better than the older siblings..

FTR a FX8350 of 2012 consume about 3.3W at idle while after mid 2013 they are at about 6.5W with the FX9590 being 15% more leaky than the 8350, so in essence The Stilt was comparing a first process 8350 to an updated process manufactured 9590 without noticing what happened to the new batches of 8350s...

Old review with first 8350 and the newly released 9590, there s the mesurement at the ATX 12V rail :

https://www.hardware.fr/articles/905-3/consommation-efficacite-energetique.html

Updated review with new SKUs including the "low leakage" FX8370E, and of course the "leaky" 9590 :

https://www.hardware.fr/focus/99/amd-fx-8370e-fx-8-coeurs-95-watts-test.html

https://www.hardware.fr/articles/965-4/consommation-efficacite-energetique.html

So it took all this time to point that the discussion in the forum you linked was pointless and just a bad mouthing of the 9590 and AMD...

 

[DrMrLordX]

Sure @Abwx, sure. You win!

So it took all this time to point that the discussion in the forum you linked was pointless and just a bad mouthing of the 9590 and AMD...

Yup, none of those people knew anything. Especially not that The_Stilt guy.