FX-8350, How's yours?

[Anonaru]

Just got one (For free mind you-- see new toys thread), and am having an interesting time overclocking it. I've gotten up to 4.8GHz, and haven't had any issues keeping it cool so far

Now, I'm curious. What's your configuration? How did you get your maximum 24/7 or stable OC? I'm using a multi-only OC with:

200 FSB
2200NB
4800CPU
2600HT
1.105 and 1.205 for NB / HT
1.475 vcore
1.225 vCPU-NB
2.505 CPU-PLL

I can run IBT and use anything intensive without any issues, but this gives me an "Invalid sumout" spam on every core during P95. Wonder what thats all about? Too much multiplier?

 

[PolRoger]

Nice find... Being able to recover a fully functioning chip out of the trash!

I'm expecting the return of my RMA'd CHVF from FedEx sometime today... So I hope to have mine back up and running soon.

The illegal sumout is just another type of "instability" showing up in your o.c.

Here is a thread for new FX owners with some overclocking results:

http://www.overclockers.com/forums/showthread.php?t=719466

 

[Anonaru]

Nice find... Being able to recover a fully functioning chip out of the trash!

I'm expecting the return of my RMA'd CHVF from FedEx sometime today... So I hope to have mine back up and running soon.

The illegal sumout is just another type of "instability" showing up in your o.c.

Here is a thread for new FX owners with some overclocking results:

http://www.overclockers.com/forums/showthread.php?t=719466

Been hearing some funny things with 8350 and P95. Getting a perfect IBT 20 runs three times in a row and then gaming on it while P95 instantly gives illegal sumouts on every core no matter when / how I run it

Btw- Thanks for the link!

 

[PolRoger]

Yes... I find Prime to be a very hard stress test for the Bulldozer/Piledriver arch. I sometimes (but not very often) do 1hr(+) runs but I don't care to do long tests. If my setup has "24/7" folding/crunching capability... Then I consider it good enough.

 

[RGone]

I have the FX-8120 and the FX-series will fool one by just how much Vcore it takes to stabilize P95 Blend mode. I get my cpus P95 Blend stable before " I " consider them stable. That is me.

 

[Anonaru]

I have the FX-8120 and the FX-series will fool one by just how much Vcore it takes to stabilize P95 Blend mode. I get my cpus P95 Blend stable before " I " consider them stable. That is me.

You don't think it could be something other than stability giving "Illegal Sumout" on all 8 cores at exactly the same time, the very moment I begin the P95 run?

I don't get the error if I hold everything the same and lower the fsb to 199, I also don't get the error if I increase fsb slightly and lower multiplier. Voltage doesn't seem to fix it, and IBT will run multiple 20's without any issue.

Frustrating chip

 

[RGone]

Frustrating chip

I have not looked for the 'how you got' the FX-8350, but didn't you say it came out of the garbage can? Maybe it was there for a reason.

 

[Anonaru]

I have not looked for the 'how you got' the FX-8350, but didn't you say it came out of the garbage can? Maybe it was there for a reason.

Didn't get it out of a garbage can-- Got it from a kid in my shop who "bricked it" by attempting to force it into a AM2 board. He said both were busted, and refused my offer to test it for free. A few slightly bent and one severely bent pin, but all made a recovery. (Storys and pic in the "What New toys did you get" thread)

The CPU is completely fine-- I can get it to pass benchs up to 5.2GHz and pass IBT as high as 5.01, but the sumouts in P95 are't really something I can avoid.

I'm thinking I'm just overlooking something, like forgetting to change a voltage or not appealing to the pigmy overloards

 

[RGone]

Okay found your posts in the "new toys" thread. Brain surgery involved. Hehehe. I know what you mean about that if more than two or three pins are bent.

Sumout on all theads seems odd perhaps but 4.8Ghz at 1.45Vcore seems low to me. Realize I said to me. In these forums with as many as use things, well any other idea might be valid for them.

Have you flashed the Vishera bios for the UD3 board? I am guessing you have but checking.

It sounds like a partially dorked chip at first glance at the symptoms. Works but works oddly.

I guess it is usable and you are. Will do a little more thinking on the symptoms. Odd for sure.

 

[Anonaru]

Okay found your posts in the "new toys" thread. Brain surgery involved. Hehehe. I know what you mean about that if more than two or three pins are bent.

Sumout on all theads seems odd perhaps but 4.8Ghz at 1.45Vcore seems low to me. Realize I said to me. In these forums with as many as use things, well any other idea might be valid for them.

Have you flashed the Vishera bios for the UD3 board? I am guessing you have but checking.

It sounds like a partially dorked chip at first glance at the symptoms. Works but works oddly.

I guess it is usable and you are. Will do a little more thinking on the symptoms. Odd for sure.

I got the sumout to go away when I increased PLL voltage to 2.695 and lowered vcore to 1.450 with LLC set to high (Medium or anything more than 'high' caused it to return).

At 1.475 it runs up to 4.8GHz if I OC it using only FSB, but I get sumouts any time I use 23 or higher multiplier.

I think its just something I'm doing wrong. When I'm sitting in front of it I'll do some homework and tinkering

 

[RGone]

One thing man you are not limited to working only from within the box. Hehehe.

I don't think I have heard anyone with a cpu that acts as you describe. Of that so far I am sure. Such has not come across my net screen. However it seems you are doing some winning against the odds and oddities. Good luck to you.

RGone....ster.

I got the sumout to go away when I increased PLL voltage to 2.695 and lowered vcore to 1.450 with LLC set to high (Medium or anything more than 'high' caused it to return).

At 1.475 it runs up to 4.8GHz if I OC it using only FSB, but I get sumouts any time I use 23 or higher multiplier.

I think its just something I'm doing wrong. When I'm sitting in front of it I'll do some homework and tinkering

 

[PolRoger]

Been hearing some funny things with 8350 and P95. Getting a perfect IBT 20 runs three times in a row and then gaming on it while P95 instantly gives illegal sumouts on every core no matter when / how I run it

Btw- Thanks for the link!

I got the sumout to go away when I increased PLL voltage to 2.695 and lowered vcore to 1.450 with LLC set to high (Medium or anything more than 'high' caused it to return).

At 1.475 it runs up to 4.8GHz if I OC it using only FSB, but I get sumouts any time I use 23 or higher multiplier.

I think its just something I'm doing wrong. When I'm sitting in front of it I'll do some homework and tinkering

As RGone already mentioned... It seems you are experiencing some peculiar behavior from your chip. I've also experienced the "illegal sumout" when running Prime but usually it happens only on one or two cores/threads. If I either lower my overclock or further tweak/bump some of my voltages I can usually can get them to go away.

 

[bassnut]

Me too .... plenty of Illegal Sum Out errors but just on one core. 1.45v is low for 4.8. I have turned my OC down a bit to 4.9GHZ and I think I am up around 1.48 ish VCore and that is Prime stable overnight. There is a point with this chip where a big jump in VCore is needed to get stable and that is around 4.7GHZ, or at least that was the way it was for me. It is quite similar to what I found with the FX 8150 though with it it was around 4.6GHZ. There were a few people that noticed this trend. I have found though when pushing past 4.7GHZ using more FSB instead of Multi can help stablize your OC.

 

[Anonaru]

Me too .... plenty of Illegal Sum Out errors but just on one core. 1.45v is low for 4.8. I have turned my OC down a bit to 4.9GHZ and I think I am up around 1.48 ish VCore and that is Prime stable overnight. There is a point with this chip where a big jump in VCore is needed to get stable and that is around 4.7GHZ, or at least that was the way it was for me. It is quite similar to what I found with the FX 8150 though with it it was around 4.6GHZ. There were a few people that noticed this trend. I have found though when pushing past 4.7GHZ using more FSB instead of Multi can help stablize your OC.

Understood as the typical solution.

I figured it out though: With "High" LLC, if I ran 4.8 with 1.475, I'd get sumouts.

If I ran 4.9 with 1.475, they went away. Dafuq? I'm approaching 5.0GHz now, using FSB mainly now. I'll probably be satisfied with a 4.9 24/7, what a strange bugger this chip's been haha

Edit: While sumouts go away at 4.9, BSOD's come at about 10 minutes. lmao >_< I'll be giving a voltage bump now that I got that crap to go away

 

[Anonaru]

Alright, I'm satisfied until I get my custom loop built and tested. I'll be sticking to a FSB-overclocked 4.8GHz for now (The brown score is a 4.8 multi-only OC, note the diff. in scores. Haha, I can see my old 955BE runs on there too )

Note: I peeled one of my 680's out of my intel rig to test out some graphics scores from way back when I only had one 680 in my Deneb rig, just for giggles. Back to my 670s now, though

I paused P95 to run cinebench, and resumed it while loading CPU-Z (Hence the LLC induced voltage hike. Its 1.475 / 1.504 with LLC and increased NB / CPU-vPLL)

Couldn't get off needing to raise vPLL to 2.695, hopefully I'll find a work around for this eventually. But for now, woot! Passes IBT and P95, finally

 

[Anonaru]

Myself, staying stuff.

Was able to push it just a smidgen more-- Biggest gain was the decrease in 100% load voltage, but I got a near .1 gain in cine and cooler temps. Don't think I'm pushing this farther until I get a real cooler set. Fully stable though, didn't bench till after 12 hours of P95 while we had turkey day!

Dropped the 680 in again for some tests, but I think I cut that and the graph out trying to size it similar to my CPU-z shot. Excited now, this chip may be a very good chip to run 4.82 at this voltage

 

[azanimefan]

so you've OC'ed your "bricked" FX8350 to 4.8ghz on a stock fan?

good lord that chip makes me drool...

 

[Anonaru]

so you've OC'ed your "bricked" FX8350 to 4.8ghz on a stock fan?

good lord that chip makes me drool...

Oh god no on the Stock Fan thing-- Its on an H100 haha
But yeah, I got a "bricked" 8350 to 4.82 on less than 1.5v under full load

Edit: Sadly, I still can't avoid the high PLL voltage-- I got it down to 2.585 from 2.695, but any lower and I get sumouts on P95, even if I raise LLC, other voltages, etc. Temps hit at about 65 socket, 58 core, can't go any farther