• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

**LETS SEE YOUR PILEDRIVER OVERCLOCKS - LET ME START WITH 5GHz+!!**

PCZ said:
Well you aren't going to run extended prime tests at high clocks if you constrain yourself to that degree.

PD's are not Thubans they run at higher reported temps.
I say reported because AMD temps are fiction.
Click to expand...

You have to have some limits to work to. I mean there's plenty of people that have had throttling from their PDs and it's not much over the limits I was talking about. Plus my mobo software gives warnings when the socket temp hits 65degC. Are you regularly letting yours go over that during stress testing?
 
CAT-THE-FIFTH said:
The CPU is not the only part though. I am a SFF PC fan,so things like this have interested me,much longer than for most people running full sized systems.
Click to expand...

For this reason, when I switched to Core i7, I got rid of the CRT monitor (about 6 months before), in favour of 2 LCD monitors. The power usage went down by about 90%, compared with the CRT. I also got rid of the "gaming" video card. I now use onboard video and a low power video card, which allows me to output to 4 screens.

At the plug point, idle power usage went from around 280W, to under 150W.

I understand though, that some people don't care about power usage, in which case, fair enough. My way of thinking is that it's better to get an i7 (pay a little extra at the time of purchase) and then save on electricity costs, with the progression of time.

Anyway, back on topic....
 
Wolvers
I don't worry too much if i see low 70's on small fft's.

Doesn't get anywhere near that when i do long encodes, probably the primary use of this CPU for me.

As to the throttling it may be VRM load rather than socket temp which triggers it.
The SabreTooh is very strong in the VRM department and i don't get throttling.

This isn't my primary PC though so i am not investing a lot of time or money in it.
I think if it was, i would consider going down the WC route with it as you are doing.
 
Last edited:
Sabertooth here too, R2. I have had the VRMs (VCORE-1) running at 70+ and it didn't throttle.

I get the feeling though that these CPUs are more stable when the core is kept under 50C. For example, if I take the PC out of the AV rack that it sits in and add a more powerful fan to the intake, it'll run with temps as much as 10C cooler under full load. I've had OCs fully stable like that, but put it back in the rack with the normal fans where it runs a bit hotter, and into the 60s, and it looses some stability. Then it needs a little more voltage for stability and that obviously means it runs hotter again.

So at the moment it's running at 4.7ghz and I can leave it encoding all night without it breaking 55C. I'm happy with that compromise..............for now. :D
 
basically all anyone needs to know it like any cpu and peoples claims unless you have uber cooling and a magic chip you wont be running 5ghz

most will be around 4-4.5ghz in reality.
 
From the two that I've had, I'd say that 4.5ghz should be the minimum you can expect from one of these. It's 4.6Ghz and over where the hotter temps and higher vcores might be a problem for some people, but even then I would expect a closed loop cooler to cope with 4.6ghz. The bigger closed loop ones and custom loops you're probably looking at 4.8ghz and over.

It's dependent on the RAM speed you want to run too. Anything over 1800Mhz is likely to need a bump in CPU/NB voltage and that adds heat to the CPU as well. I've decided to stick to 1600Mhz for now (keeps my CPU/NB down at 1.175v) with tighter timings but this 30nm RAM is good for all sorts of different clock/timing options so I'll probably be playing with that again soon once I've settled on one CPU speed and it's 100% stable.
 
Wolvers
Does your CPU NB default to 1.4v ?
It does on the MK 1

I manually set it to 1.2 or 1.25.
1.2 seems ok with the samsungs at 1.35v but with the HyperX at 1.575v it needs to be 1.25 for stability.

Pushing the ram speed up to 2133 needs quite a bit more NB volts, when i first did some testing it was set on AUTO and 2133 worked OK.
That is 1.4v though, i wonder if ASUS set it that high on purpose.

Have been using 2400 NB speed, 2500 on occasions when messing with the FSB.
It doesn't seem to make a lot of difference to performance though and i am tempted to bring it down to stock.

I have a suspicion that the cache isn't reliable at 2400 Mhz when the chip gets warmed up.
 
Last edited:
Seems to be trading blows with the an ivy-i7 in Far Cry 3!

CPU_03.png


CPU_02.png


Seems pretty impressive at 4.5Ghz!
 
Most telling thing about that graph is the fact that the IVY can pump out the same FPS at 3.5Ghz that the PD needs 4.5Ghz to do.
Basically stock IVY and overclocked PD.

At the top end of the graph the GPU is pretty much maxed out (GPU Limited)
 
Last edited:
PCZ said:
Most telling thing about that graph is the fact that the IVY can pump out the same FPS at 3.5Ghz that the PD needs 4.5Ghz to do.
Basically stock IVY and overclocked PD.

At the top end of the graph the GPU is pretty much maxed out (GPU Limited)
Click to expand...

Indeed, I just love when graphs are interpreted a certain way.
At 2.5GHZ the i7 gets higher than the FX8350 at stock.

Not that you'd run your i7 at 2.5GHZ, however you have 2GHZ at least left in the tank (That said, I know someone who runs their i5 at 3GHZ), on the FX8350 you have ~1GHZ.
 
Last edited:
Wow, another 8350 being compared with the 3770k. :rolleyes: Considering at max clocks there's 7% fps loss (and improving as clocks get higher) to a chip that is 40% more expensive, that's pretty good in my book. If you really want to make the 8350 look bad in gaming you have to compare it with the 3570k.

PCZ said:
Wolvers
Does your CPU NB default to 1.4v ?
It does on the MK 1

I manually set it to 1.2 or 1.25.
1.2 seems ok with the samsungs at 1.35v but with the HyperX at 1.575v it needs to be 1.25 for stability.

Pushing the ram speed up to 2133 needs quite a bit more NB volts, when i first did some testing it was set on AUTO and 2133 worked OK.
That is 1.4v though, i wonder if ASUS set it that high on purpose.

Have been using 2400 NB speed, 2500 on occasions when messing with the FSB.
It doesn't seem to make a lot of difference to performance though and i am tempted to bring it down to stock.

I have a suspicion that the cache isn't reliable at 2400 Mhz when the chip gets warmed up.
Click to expand...

On auto, the 8350 sets the CPU/NB to 1.175v. The 8320 was 1.3v IIRC, and I had to raise that to 1.35v for stability of higher RAM speeds although I never got round to getting it 100% stable at 2133MHz. From what I can tell, anything over 1866Mhz is basically OCing the mem controller. I've yet to try higher RAM speeds on the 8350, but I'm assuming that it'll need a bump to keep it stable.

As for NB speeds, I think there's little benefit in OCing it. It seems reasonable to me that as long as it's the same as, or higher than, the RAM speed there's no bottleneck. Some guys at OC.net that were testing out higher HT and NB speeds and finding that it improves multi-gpu performance. I can't remember which one it was now. Does the PCI-E go via the NB? That might explain it.

Would you agree that these are more challenging/interesting/satisfying to clock than current Intels?
 
It is the higher HT speed that helps with multi GPU performance.
I did experiment a bit.

As to them being more interesting to clock than intels i would say what Intels.
Sandy's and IVY's yes but SB-E no.
 
This may be a silly question and you will see from my rig I am not high end but.....

When you play in the real world do you notice the difference between 64 and 73 fps?

Sorry if a noob question?
 
I get your last statement and agree but is that a statement of benchmarking?

If I gave someone who did care the game to play on two otherwise identical computers would they actually notice, I know I am simplifying and other variables would infer but I just wondered.
 
I don't get the question, seems worded strangely.

If people didn't care about framerate, they wouldn't be running flagship GPU's.
The people who just want to game will just game, and likely not notice the difference between 64 and 73.
 
Thankyou, I guess your last statement answers the question, those who buy these GPU's do more than just game because it matters to them.

The rest just game!
 
jackboy said:
I get your last statement and agree but is that a statement of benchmarking?

If I gave someone who did care the game to play on two otherwise identical computers would they actually notice, I know I am simplifying and other variables would infer but I just wondered.
Click to expand...

To answer your question; no you wouldn't notice the difference. :)
 
You must log in or register to reply here.
Back
Top Bottom