Sweetheart little FX-8320

[Johan45]

On a whim I picked up anF-X 8320 on Friday since they were on sale. I got it home and started some testing and man was I impressed. Runs P95 blend and IBTAVX on very high with less that 1.45v . Then I uppped the ram and NB a bit and ran Cinebench 11.5 at 5392 with 1.45v. I am just amazed at the voltage requirements of this CPU.

[IMG][/IMG]
[IMG][/IMG]
[IMG][/IMG]

[IronBalls]

Lucky you.

You probably got one of the best FX 8320 out there
What was the room temperature? I see temps are really low for 1.45v.

And how much does your CPU needs for 2800 MHZ CPU/NB ?

Congrats for your new cpu

[Johan45]

Well , I do use a chilled loop that's why the lower temps. The mix is usually in the -10c range but warms up fairly quickly since I just use a deep freezer to cool the res and rad. As you can see in the pics between IBT and P95 the temps were on the rise at that point. Even compared to my 9370 with the same cooling, it needs 1.5v to be P95 stable at 5.0.
The CPU_NB volts were 1.35 for 2800

[IronBalls]

That explains why your 8320 does so good. Because 1 hour prime95 stable @ 5 Ghz and 1.418v is awesome.

The best my chip does is 2 hours OCCT @ 5 ghz and 1.44375v but with ambient water temperature (20-21C).



I wonder what my cpu would be capable of with chilled water

I wonder why you don't run Cinebench with higher voltage? Is 5.4 GHZ the limit of your CPU even with raised voltage?

[Johan45]

That I hopr to find out over the next couple of nights. I certainly hope it keeps scaling but you never know. It would be a real pity if that's where it quits.

[22JHP]

could you show your VID value?

[FlanK3r]

wow, awesome guys. What batch is it? Looking for more benchtests here . Johan, go, go go

Ironballs: nice too

[Daveburt714]

wow, awesome guys. What batch is it?

This ^^^

If it's batch/week 1428 or higher it may have the same die revisions as the new E series chips (C0k)...

[Johan45]

The VID is 1.212 @ 3.5 but the batch # will have to wait till I pull the block off to check.

[tbone8ty]

wow i need to get rid on my ancient FX-8350.... week 1249PGT

[IronBalls]

Well mine has a VID of 1.2625v Batch is 1416PGS. Not one of the new revisions then

Funny thing, my TJmax is 90C compared to the one above.

[Johan45]

Well I spent an hour last night trying to get 5.5 to go. Either I'm at my limit and the chips is getting too warm. The chip just won't run any higher or there's something going on with my board. I had the same sort of issue last week when trying to push my 9370 under much colder conditions and had the same results. Just a lockup when the CPU has to work under heavier load at higher speed. About three weeks ago I had my CPU fan header blow up for no apparent reason and it's been different ever since. Might be in my head but I'll swap out for my Sabertooth just to rule this out so you should have your batch number in the next day or two.

[FlanK3r]

Johan, can u test few wishes ? Thx
1) stability test in AMD Overdrive for 30min, if you can get 5100 MHZ stable here...
2) max Superpi 1M and 32M test (up to 1.55V +-)

Thank you very much

[Johan45]

That might take a bit of time still getting used to the CPU but will do FlanK3r although I find AMD OD a bit weak for my tastes. I did manage to get it to run r11 at 5.5 and 15 5.45. I took my block apart and cleaned it and apparently some dumb**s didn't get it together properly the last time. It's behaving a bit more normal now but it's a hot S.O.B my loop can't handle much over 1.6v. The batch was 1403 PGY

http://hwbot.org/submission/2661301_ as you'll see took a prety good bum in voltage as well.

[IronBalls]

Nice run, you beat my R15 run by 50 Mhz

You should be okay doing a run @ 1.6v, as I went as high as 1.5875 with water temp of 18C. And my watercooling was capable of keeping the CPU temp below 55C. But the socket temp is the problem. It literally sky-rockets above 75C once I bump voltage above 1.5v. Do not know if it is normal for crosshair formula V. But many people complain about this too and cool the backside of their motherboards with a big fan to keep socket temp under control.

Ha know I'm considering dumping my DDR3 1600 that's holding my scores back and buying some decent kit

[Johan45]

Thanks Ironballs.
I swapped ram on that run for my flares. I can get them a bit tighter than that but I'm still feeling out the CPU. For AMD I think lower speed and tight timings are still better but you'll need an older PSC kit or similar to get good ram these days. The new stuff just isn't the same. I have some PI here and Corsair Dominator all with similar results. I have been able to put more volts into other chips with no ill effects but this one is just too tight. It gets unstable when the volts go up. I'm hoping some more cold will fix that. I have a 140 behind the mobo and a couple 92s on the VRM NB cause you're right they really heat up when pushing these CPUs

[Johan45]

Just an update, still making progress. managed to run R15 at 5.5 last night http://hwbot.org/submission/2662260_ and a valid at 5.7 http://valid.canardpc.com/lpqhiy. Still staying under 1.6v

[IronBalls]

Just an update, still making progress. managed to run R15 at 5.5 last night http://hwbot.org/submission/2662260_ and a valid at 5.7 http://valid.canardpc.com/lpqhiy. Still staying under 1.6v

How did you get 100 Mhz more out of your CPU? Higher voltage? Lower temps? Really impressive. Even people using phase change as cooling don't go as far.

[Johan45]

Cold water has it's advantages with such hot CPUs. The heat these things put out can overpower some SS units. The water has more ability to pull heat away. I'm still feeling out the CPU, and finding what I can and can't do with it ran Wprime 1024 at 5.6 last night but didn't have a lot of time. http://hwbot.org/submission/2663039_
@FlanK3r I haven't forgot about AOD just haven't had time last few nights for a stability test. Should have results before Monday though. For settings did you want to see the same as I posted initially 2400 NB and 1866 Ram? How long to test as well 25 minutes , an hour?

[Johan45]

OK FlanK3r here's part of it. 30 mins @5100, 1.428v with a droop to 1.404v under load. I'll get to the SPi over the wekend hopefully. My XP part seems to be broken so have to fix it.

http://tinypic.com/r/j7689w/8

[FlanK3r]

ahhhh, I cant see your screenshot, probably you have wrong tag in forum ...Show me, show me :-D. Thank you. 5100 MHz 1.428V is awesome! What ambient temps or radiator temps?

[Johan45]

I made it just a link, I don't know what it is but I get a hard time trying to upload here. Water -10 or warmer and it ran at 1.404v for the test I was surfing on the side as well. I have a droop set from bios.
I spent some time at this before running the stability test so the water wasn't at it's prime. http://hwbot.org/submission/2663832_...0_9.71_points/
Oh ya and this http://valid.canardpc.com/y1zyki

[IronBalls]

Is this AOD stability test worth something? I mean does you Flank3r use this to test stability and then you use the CPU to bench, play etc. and does it never freeze?

I tested it yesterday, and my cpu passed the test @ 5,1 ghz with 1.45v ... Compared to OCCT where my cpu needs 1.5125v it seems that AOD is really light on the cpu.

[FlanK3r]

And do you think OCCT with AVX or without? In practice I dont use AVX yet. Im not big gamer , so I played only few modern games, but without issues.
What do you think about y-cruncher? This test seems very hard too...

Johan: n1, this chip seems similar as my 9590 ES (5100 MHz AOD 1.515V, but with allinone and 22C temps in room)

[IronBalls]

And do you think OCCT with AVX or without? In practice I dont use AVX yet. Im not big gamer , so I played only few modern games, but without issues.
What do you think about y-cruncher? This test seems very hard too...

Johan: n1, this chip seems similar as my 9590 ES (5100 MHz AOD 1.515V, but with allinone and 22C temps in room)

I don't use CPU:Linpack because it's basically the same as using IntelBurnTest. And IBT is softer on the CPU as OCCT or prime95.

I always stress my CPUs 2 hours with CPU:OCCT large data. For me, it's the most stressing test. I had CPUs passing prime95 blend, but failing OCCT ... OCCT has another advantage, once a core fails, it stops immediately. Prime95 keeps testing the other cores ... It's a waste of energy.

So for me the best is OCCT>Prime95 (blend)>IBT>AOD

I never used y-cruncher, so I can't talk about that one.