Power Supplies and Venice overclocks...

[d33pblue]

I got my DFI mobo and Venice 3000+ last week. As far as overclocking goes, it hit a brick wall at 2450Mhz.

A series of strange behaviors leads me to believe that its my power supply (400W Seasonic supersilencer) holding me back. I'm pushing a lot of hardware with it:

Venice 3000+ (9x272, 1.55v)
2GB Corsair ddr400 (2.6v)
2 x X300 video (three monitors)
4 x 250GB Hitachi hard drives (2 500GB raid 0 arrays)
Audigy 2 sound
Pinnacle TV tuner
DVD-RW

The specs on the Seasonic are as follows: +3.3V@28A, +5V@30A, +12V@22A. Yes, I know the 12v rail is weak. That should be whats holding me back, right?

Whenever I try to put more than 1.55v through my processor, my hard drive array wont boot. I also cant boot whenever I push the HTT up past 275 (running 272 now prime stable). Ive got the DFI Ultra-D, so I know its capabale of much more. It takes all of 1.55v to get the processor stable at 2.45Ghz. That seems awfully low. Could lack of amps (or stable voltage) require that my processor need a higher voltage to be stable?

It seems as if my PSU is making more noise these days than it usually does (could be my imagination) - indicating that its closer to its maximum output than it was with my old system.

Anyhow, what I want to know is how high you guys have gotten your 3000+s (or venice chips in general), and what type of power/12v amperage your power supply outputs. I'm debating with myself whether or not to buy a new PSU for the sake of longetivity and a higher overclock.

Thanks

 

[Frown66]

DFI nf4 ultrad + Venice 3200+ at 2.7ghz (270x10), cpu at 1.61v (1.5 x 110%)

PSU is an OCZ MODstream 450, 12v is at 26A. Everything is very stable. All the instability when attempting to reach 2.7ghz was related to RAM... once I found a good setting, everything was fine. My video cart is a 6800GT o/c'd pass Ultra levels, so I've got a lot of power draw there too. Smartguardian does not show anything under or over volting.

I was thinking my PSU would not be enough at first, but I don't see any problems with it.

 

[CMC79]

There are a whole lot of things you can try before that to make sure, but you're running quite a lot off your PSU. Did you make sure to plug in all four power connectors to the board?

Your memory--is it 4x512mb or 2x1GB? If it's the latter than try running with a 2T command rate. What divider/timings are you using?

You may also try upping the voltages on the LDT, chipset, and memory.

I would guess 400W may not be enough for all the you have anyway though. I have a 500W Antec Surepower with my DFI NF4 Ultra and Venice 3200, and I'm running stable at 9x289 on 1.5v.

Maybe you should try running with fewer peripherals and see if your overclock will go higher with less of a draw on the PSU. That'd probably help to isolate if it's a power issue.

 

[Concillian]

Originally posted by: d33pblue

Venice 3000+ (9x272, 1.55v)
2GB Corsair ddr400 (2.6v)
2 x X300 video (three monitors)
4 x 250GB Hitachi hard drives (2 500GB raid 0 arrays)
Audigy 2 sound
Pinnacle TV tuner
DVD-RW

The specs on the Seasonic are as follows: +3.3V@28A, +5V@30A, +12V@22A. Yes, I know the 12v rail is weak. That should be whats holding me back, right?

Honestly, I don't think that 22A on 12v should be limiting you here. You don't have much on your 12v. Two PCI-e cards, but they're weak cards that are pulling well under the AGP spec power (50W). 9600XTs would still run off the AGP slot with no external connector, and people would overclock them even more. I'd be pretty surprised if a x300 pulls 30W each at peak. Memory is coming off 3.3 or 5v rail, which sholdn't be a concern. The hard drives are going to average around 1A each, as their peak is only on spinup. You should have like 120-150W of headroom on the 12v rail for your CPU, and an A64 is not going to pull that kind of juice.


If your 12v is your limiting factor, then try this:
REDUCE your voltage on the CPU by 0.1v. This will quite significantly reduce power consumption on 90nm process CPUs. If you can overclock further, then your PSU was the problem. If you can't even make what you had, then your CPU is the problem. If you can make the same as what you could before, your MB is the problem.

Or just unplug whichever array is not booting to free up a couple amps on the 12v rail. to see how that helps.

The fact that your HTT past 275 isn't stable indicates that it's probably something with the board that is at issue here. You can't do say 8x300? A litle odd for a DFI, but it is possible that it tops out in that range.

What is your HT speed/multiplier?
What is your RAM speed/divider?

Where do you max out on 1.4v on your CPU?
Are you running the memory off the 3.3v or did you move the jumper to give it a max 4v (which runs it off the 5v rail, and is supposedly less stable with DFI NF4 and 20 pin supply)

 

[d33pblue]

Originally posted by: CMC79
There are a whole lot of things you can try before that to make sure, but you're running quite a lot off your PSU. Did you make sure to plug in all four power connectors to the board?

Your memory--is it 4x512mb or 2x1GB? If it's the latter than try running with a 2T command rate. What divider/timings are you using?

You may also try upping the voltages on the LDT, chipset, and memory.

I would guess 400W may not be enough for all the you have anyway though. I have a 500W Antec Surepower with my DFI NF4 Ultra and Venice 3200, and I'm running stable at 9x289 on 1.5v.

Maybe you should try running with fewer peripherals and see if your overclock will go higher with less of a draw on the PSU. That'd probably help to isolate if it's a power issue.

Oh wow... umm... So there are FOUR power connectors that should be connected to the board? I currently only have three connected. I took a look at the DFI manual and it appears there is a 5v/12 connector next to the top PCI-E slot. I should have a power connector from the PSU connected there, right? If so, then i feel profoundly stupid for making this thread.

 

[d33pblue]

Ok, update

Plugging in the second, FDD type 5v/12v power plug made *no* difference at all in the maximum overclock.

To answer a few questions:

1. The ram is 3-4-4-8 T2 (system wont post at T1), 272*.7 divider = 190mhz.
2. At 2.45Ghz, system BSODs at any less than 1.55 vcore.
3. System wont boot at any higher than 1.55vcore.
4. System *WILL* do 8x285. This indicates that its the chip that is the limiting factor here, not the board.
5. 272x9 is still my max. stable overclock. 275x9 still wont boot at any vcore.
6. Memory is still on the 3.3v jumper.

I'm guessing the extra 5v/12 connector plugged in would be supplying a little extra power to the system, right? Since my max. overclock is still exactly the same, I would guess that power isnt my limiting factor.

Am I missing something here? Any ideas?

 

[bjc112]

I'm running the rig in my Sig and it doesn't seem to have any problems..

It does have 26A on the 12v line though.

 

[Sentential]

To be honest the silicon quality plays a much larger role than does the PSU.

 

[Duvie]

12v rail does include the drive motors of the HDDs so it can still be a limiting factor...NOt just vid cards and cpu.....

It may not be quantity either it could be quality and effects of fluctuating power and noise....Have you tested the pSU with a multimeter to see if it is delivering what the board and other s are reporting...

List 12v rail load and idle as well as the +3.3v rail load and idle...

Have you made sure to raise your vdimm (voltage to the ram)???

 

[Concillian]

After getting my Venice 3000+ CPU & DFI LP UT NF4-D, and seeing your further testing, I'm more and more convinced you just got a relatively poor piece of silicon.

If I set voltage to 1.55 on my NF4-D, actual voltage is closer to 1.515 as reported by CPU-Z. At that voltage, I can only get stable just over 2500 Mhz. So I don't think it's a longshot to say that your CPU just isn't stable at speeds above what you are getting. I just can't believe that PSU is going to make a big difference in that.

My PSU is a seasonic S12 430W, which is dual 12v rail. The 12V1 (CPU) rail is 14A and the 12V2 (everything else) rail is 15A.

I still need to test with my UTT chips, which will operate at higher memory speeds. With my current memory (that isn't stable much above 200 MHz with 2-3-2) and BIOS revision, the memory presents a problem at certain dividers. For example, at my current OC (2560 @1.57v read by CPU-Z) I cannot run stable at 9x284 with the 140 divider (200 MHz RAM) but I CAN run at 8.5x300 with the 133 MHz divider (197MHz RAM). The memory runs fine a 1:1 slightly above 200 MHz. However, something is odd in some of the particular RAM divider settings that causes instability.

You may want to play around with your dividers to verify that you are not experiencing something similar. 100 and 133 settings didn't seem to be a problem, but 140 and 180 sure seemed to be. I haven't tested others to see what else may be an issue.

I was going to update to one of the 5-10 BIOSes posted on dfi-street.com before thoroughly checking things out.

 

[bjc112]

Is it possible to unhook some of your devices and try and maximize your overclock?

Maybe a couple HDD's and that TV tuner.. Just to see what happens..

Maybe drop to a gig of ram for a few hours..

 

[ericlala]

I'm using an Enermax 485W noisetaker with dual 18A 12v rails. It is Silent! and i also love the dual fan where it helps suck hot air from the cpu.