FX 8350 5GHz under heavy load,its not that hard on air

[PcCI2iminal]

Room temp is 20c
the CPU is a retail box
boot at 4800MHz
vcore at 1.52v
and the heatsink is Enermax ETD-T60 with stock fan at full speed

Thanks for reading

[punx223]

dude you are lucky,..... mine at 1.47V gets into 64C on water....

[AMDforME]

Congrats ! The temps don't look right however as others are reporting 60+ C when using 1.5 vcore. It looks like many folks may be able to get close to 5.0 GHz. on good air cooling, for 24/7 operation.

[PedantOne]

Congrats ! The temps don't look right however as others are reporting 60+ C when using 1.5 vcore. It looks like many folks may be able to get close to 5.0 GHz. on good air cooling, for 24/7 operation.

only in dreams, 8350 is not much better ocer then 8150, only little

[Miwo]

In before cherry picking, prime stable vs cinebench stable mudslinging.

[PcCI2iminal]

Congrats ! The temps don't look right however as others are reporting 60+ C when using 1.5 vcore. It looks like many folks may be able to get close to 5.0 GHz. on good air cooling, for 24/7 operation.

yes ,mis-read sensor
temperature does not appear to be correct

[FlanK3r]

but score is not much good, looks like APM problem....The same I have with FX-8150 in R11.5 and similar clock (5050MHz)

[punx223]

but score is not much good, looks like APM problem....The same I have with FX-8150 in R11.5 and similar clock (5050MHz)

Yes some benchmarks run very buggy with Bd and also PD.... so i try not to use them unless I see reliable results.

[keithlm]

but score is not much good, looks like APM problem....The same I have with FX-8150 in R11.5 and similar clock (5050MHz)

Doesn't the FX throttle at 70C? If this were to happen during a benchmark run then the results would be totally sporadic.

(And while there may be a problem reporting the temperature externally the CPU itself should not have a problem monitoring it's own internal temperature so that it can throttle.)

[Tomasis]

it means my cooler (Siilver Arrow) could do that better then Enermax cooler? it looks good for me , I would love 5ghz Cinebench benchmark

[FlanK3r]

Doesn't the FX throttle at 70C? If this were to happen during a benchmark run then the results would be totally sporadic.

(And while there may be a problem reporting the temperature externally the CPU itself should not have a problem monitoring it's own internal temperature so that it can throttle.)

yes, about 68-70 in Coretemp begining the problem....Next, APM master mode settings must be disabled etc...

[03hdfatboy]

will the Crosshair IV Extreme run the new 8350 with the bios update that ASUS says is compatible with AM3+?

[EniGmA1987]

yes

[Daveburt714]

Hi PcCI2iminal,

What's the date code on your chip, and do you have any idea what safe max core voltages are (specs)?

I picked up a chip yesterday (batch 1237) and I'm having a hard time getting clocks much higher than what I had on my sweet 8150.
I'm stressing the cores right now @ 4.78 with 1.4625v. Temps look decent and it seems pretty stable but I'm a little concerned about
pushing the voltage much higher (scaling has slowed as well)...

Here's a Prime95 Large FFT run of ~20 minutes (just stressing cores):


Guess I was hoping for a little higher clocks.
I don't see this one getting to 5Ghz stable (on custom water). I've just started, but I'll keep pushing voltage a little as long as temps don't get out of control.
From the week code I get the feeling this one came from a batch where they were just trying to build stock for the launch...

On the bright side, I can see some IPC improvement at these clocks in a few benches.
The x264 encode bench shows a decent improvement over the 8150, and OC'd load power draw (system) has actually dropped by ~90w!

[BeepBeep2]

Impressive findings Dave!

[Zeus]

Nice Dave but it seems like the magical 5 remains elusive for AMD users.

[informal]

Dave are you sure you measured 90W drop in OCed scenario? That's massive difference. I hope you will hit your 5Ghz but efficiency will go down the toilet since you will need to push Vcore higher and power draw will go up noticeably. 4.78Ghz is just 4% lower than 5Ghz,I guess it's the nice round number people are after rather than performance improvement (not saying you won't notice 4% ).

PS In this Russian review you can check out 4.7Ghz on PD and BD and the difference PD brought. Maybe even verify their results now that you have the chip at the same clock . In some cases there is massive perf. difference,so much that 4.7Ghz BD is just on par with stock 4Ghz Vishera.

[Daveburt714]

Nice Dave but it seems like the magical 5 remains elusive for AMD users.

Yeah... It is Zeus (turncoat ), but is 5Ghz really that common on Intel chips?
They may be, but I'm still an AMD guy, so I don't follow what's happening on the blue side that much...

Seems I got a mediocre chip this time, but I've been working with it a little tonight and things are looking better!

I gave it some more voltage (still within reason IMHO), temps are decent and stable clocks are looking better:


When comparing Intel with AMD allot of benchmarks can be questionable, but the one that's always important to me is video encoding...
Here's some x264 numbers from the 8350, using the best (seemingly) stable clocks I've gotten so far:



Wait a minute.... WHUT ?!?!?!

This $200 AMD chip outperforms a stock ($500) 3930K and a seriously OC'd 2600k in a real world (useful) app!
Plus, you know from experience AMD chips are fun when it comes to OC'ing...

This chip is fun, does what I need it to do (quite well) and was cheap!
I don't regret my purchase, even though I do wish I'd have gotten a little better clocker...

[Zeus]

No mate not that common but there are quite a few around. Mine can do it too but it does not justify the Vcore bump it requires imho. 4.8 is where i'm at 24/7. After seeing BD clocking around 4.8 i was hoping PD would do 5.

Still, performs nice in mt benches, pretty sure you will get 9 points in CB 11.5 now Dave, assuming it will do that at 5.25-5.3G. Show me bro!

[Daveburt714]

Dave are you sure you measured 90W drop in OCed scenario? .

That was compared to my 4.7Ghz 8150 informal.
When blending (oc'd) on that chip I hit ~490w, at the same clocks the 8350 was ~400w (system)...

I'm still tweaking this new chip so power consumption could easily go up, but so far it's looking more efficient @ load.

Since GF took over production of AMD's chips VIDs/OC headroom and power seems to be a lot less consistent.
It makes it harder to compare chips, but thats what I've seen from the Kill-a-Watt with my chips...
YMMV more than it used to.

Once I get things dialed in for the best stability, I'll update power numbers.

[FlanK3r]

whats your batch Dave? Again you are maybe lucky man

[BeepBeep2]

Since GF took over production of AMD's chips VIDs/OC headroom and power seems to be a lot less consistent.
It makes it harder to compare chips, but thats what I've seen from the Kill-a-Watt with my chips...
YMMV more than it used to.

It has always been that way, except BD was a power monger with like 100w more at load OC'ed than Phenom II X6, not to mention X4.
VIDs for X4s were all over the place, as low as 1.3v for some 955/965 to 1.425v-1.475v for higher models. Power consumption I'm sure was different too, but the power consumption wasn't through the roof and into space either.

Dave, you need to drive up here on a weekend and do some overclocking with me. You seem to have strong chips

[sdlvx]

Dave, have you considered that some chips may do better with multiplier overclocking and some may do better with bus overclocking? Could you try something like 254 x 19 or 230 x 20? I'm still waiting on my chip but if we have the numbers to change, why not change them and experiment?

[AMDforME]

Dave, have you considered that some chips may do better with multiplier overclocking and some may do better with bus overclocking? Could you try something like 254 x 19 or 230 x 20? I'm still waiting on my chip but if we have the numbers to change, why not change them and experiment?

Generally speaking you can always squeeze a few more MHz. from a CPU/APU by tweaking the FSB but for many folks using the muliplier is quick and easy for a 24/7 stable PC.

[tbone8ty]

judging by how the 8150/8120 stepings went looks like this will be the best we will get? or do u think later stepings will be any better?