0.1V Vcore droop. That's really large right?

[Hulk]

E2460
9x333=3.0GHz
Asus P5E-VM HDMI
Stock Intel cooler (yeah I know I need to upgrade it)

Vcore at 1.35 gave restart after about 15 seconds of Orthos.
Vcore at 1.3625 gave Orthos error after 22 minutes.
I'm Orthos testing right now at Vcore 1.375. TAT shows 55 C. I'm 15 minutes in.
CPUZ 1.42 indicates Vcore of 1.280.

That's a pretty huge Vdroop right?

 

[aigomorla]

mmmm.... thats a bit high....

What board is this? are you using the P5B? there might be a pencil mod to fix that.

 

[Hulk]

Asus P5E-VM HDMI

It's a mATX board.

 

[Idontcare]

For a dual-core that is a rather large VDroop if it is all Vdroop.

When you say "Vcore at 1.3625" is that your BIOS setting or is that idle Vcore reading per CPUZ?

Vdroop is idle Vcore - loaded Vcore. Voffset is Vcore in Bios - Vcore at idle.

If you are taking Vcore in BIOS and substracting Vcore at load then you are observing "Voffset + Vdroop" which could really be quite large as Voffset can be quite large depending on BIOS and power supply.

If you are really referencing Vdroop (Vcore at idle in CPUZ minus Vcore at load in CPUZ) then yes 0.1V is large.

I had a 0.2V Vdroop (on a quad though) and it turned out to be a burned out component on my mobo. Could not see it from above either, had to remove the board from the case so I could see the very small (smaller than a dime) area of brown where the electrical short occurred.

Your issue may be entirely unrelated to mine, just giving you a start on possible suspects for your Vdroop.

 

[Hulk]

Originally posted by: Idontcare
For a dual-core that is a rather large VDroop if it is all Vdroop.

When you say "Vcore at 1.3625" is that your BIOS setting or is that idle Vcore reading per CPUZ?

Vdroop is idle Vcore - loaded Vcore. Voffset is Vcore in Bios - Vcore at idle.

If you are taking Vcore in BIOS and substracting Vcore at load then you are observing "Voffset + Vdroop" which could really be quite large as Voffset can be quite large depending on BIOS and power supply.

If you are really referencing Vdroop (Vcore at idle in CPUZ minus Vcore at load in CPUZ) then yes 0.1V is large.

I had a 0.2V Vdroop (on a quad though) and it turned out to be a burned out component on my mobo. Could not see it from above either, had to remove the board from the case so I could see the very small (smaller than a dime) area of brown where the electrical short occurred.

Your issue may be entirely unrelated to mine, just giving you a start on possible suspects for your Vdroop.

You're right. What was I thinking.

Bios = 1.375
Idle = 1.320
Load = 1.280

So the Vdroop is actually only .04

 

[myocardia]

Originally posted by: Idontcare
Vdroop is idle Vcore - loaded Vcore. Voffset is Vcore in Bios - Vcore at idle.

Actually, that's what we usually call Vdrop. Not saying you're wrong, just saying that's what he'll hear it called around here. Although, now that I think about it, Voffset is probably a better name for it, for the sole fact that it won't be mistaken for Vdroop, like Vdrop always seems to be. Hmm...maybe we should all adopt your term for it.

 

[Idontcare]

Originally posted by: myocardia

Originally posted by: Idontcare
Vdroop is idle Vcore - loaded Vcore. Voffset is Vcore in Bios - Vcore at idle.

Actually, that's what we usually call Vdrop. Not saying you're wrong, just saying that's what he'll hear it called around here. Although, now that I think about it, Voffset is probably a better name for it, for the sole fact that it won't be mistaken for Vdroop, like Vdrop always seems to be. Hmm...maybe we should all adopt your term for it.

Well in all honesty I am not out to invent anything, am merely using the terms for it.

Weren't these the terms used in the Vdroop article published on Anandtech a few weeks ago?

I like their definitions as Voffset is basically the delta between what you wanted to give the CPU versus what the CPU is actually getting.

And since power, heat, and CPU chipkill from overvoltage are all based on what the chip actually sees it makes sense to know how much of your BIOS Vcore setting is actually never making it to the CPU.

For example, I suppose you could in theory have a rather dodgy mobo/PSU setup which combined with a haphazaardly optimized BIOS results in a 0.3V offest.

This would mean if you wanted to give your CPU 1.35V then you'd have to tell the crap BIOS to jolt the sucker with 1.65V...a certain death for the CPU if it actually got that voltage. But given the Voffset for the board/PSU/Bios you know your chip is safe as it is merely seeing 1.35Vcore.

From their you go downhill with Vdroop as you load and unload the power circuits.

At any rate, like I said I don't want to be the inventor of some newfangled terminology, if this is the wrong lingo then let's nip it in the bud now and get me re-educated pronto! 😀

 

[aigomorla]

poor idontcare

People are getting his terms so confused. I got his term polish mixed up yesterday even!

 

[Idontcare]

Originally posted by: aigomorla
poor idontcare

People are getting his terms so confused. I got his term polish mixed up yesterday even!

I suck at communication...but that is a requirement to graduate as an engineer!

Here's the article on Anandtech I am referring to - note Voffset is defined as the delta between VID (what the BIOS is supposed to deliver to the CPU) versus what the CPU is seeing as Vcore at idle.

http://www.anandtech.com/cpuch...howdoc.aspx?i=3184&p=5

I believe I am using the terms correctly, but if no one else uses the terms this way then I will fail to communicate effectively, and communication is the point of a forum.

So let me know folks, what shall we call Vdroop and what (if anything) shall we call Voffset?

 

[VirtualLarry]

I must be lazy then. I call both the difference between BIOS setting and idle, and BIOS setting and load, "Vdroop".

 

[Markfw]

bios to idle = voffset (they may not read the same, different software and bad bios, etc....)

Idle to load = vdroop (same software reading it, it drops due to the load, usually only noticed on quads)

DQ6 = NO vdroop

 

[BonzaiDuck]

Idontcare has it right, I think. I can't imagine an ASUS board with a vDroop of 0.1V. I'd be more inclined to guess about 0.03V.

As I understood it, the vDroop would be idleV - loadV, and you would see it with a utility like Everest Ultimate while running ORTHOS or PRIME95.

Correct me if I'm wrong . . .

[Anyway, y'all got this sorted out, now that I can scan through more of the posts.]

 

[Rhoxed]

I have 3800+ with MSI K8N Neo4 with massive vdroop. At idle it will report 1.54 and at 100% load it will sometimes drop to 1.43 be it not for long its sometimes enough to make system unstable so i need to up vcore a bit because of the droop. the average is probably around 1.47 at 100% load, sometimes jumping to 1.51 back down to 1.43 Would there be any sort of MOD for this board to aid in the vdroop issue?

 

[Duvie]

Originally posted by: Rhoxed
I have 3800+ with MSI K8N Neo4 with massive vdroop. At idle it will report 1.54 and at 100% load it will sometimes drop to 1.43 be it not for long its sometimes enough to make system unstable so i need to up vcore a bit because of the droop. the average is probably around 1.47 at 100% load, sometimes jumping to 1.51 back down to 1.43 Would there be any sort of MOD for this board to aid in the vdroop issue?

Oh cant be....that is an intel feature!!! Those rascals for stealing it!!!

I have a neo4 and it doesn't have near that so it may be a sign of a bad board or issues like idontcare had...

 

[Markfw]

Originally posted by: Duvie

Originally posted by: Rhoxed
I have 3800+ with MSI K8N Neo4 with massive vdroop. At idle it will report 1.54 and at 100% load it will sometimes drop to 1.43 be it not for long its sometimes enough to make system unstable so i need to up vcore a bit because of the droop. the average is probably around 1.47 at 100% load, sometimes jumping to 1.51 back down to 1.43 Would there be any sort of MOD for this board to aid in the vdroop issue?

Oh cant be....that is an intel feature!!! Those rascals for stealing it!!!

I have a neo4 and it doesn't have near that so it may be a sign of a bad board or issues like idontcare had...

Yup.... I am so sick of hearing "well its in the anand article, they can't be wrong, its a feature"

B.S. Its crap motherboards.....

 

[Rhoxed]

Originally posted by: Duvie

Originally posted by: Rhoxed
I have 3800+ with MSI K8N Neo4 with massive vdroop. At idle it will report 1.54 and at 100% load it will sometimes drop to 1.43 be it not for long its sometimes enough to make system unstable so i need to up vcore a bit because of the droop. the average is probably around 1.47 at 100% load, sometimes jumping to 1.51 back down to 1.43 Would there be any sort of MOD for this board to aid in the vdroop issue?

Oh cant be....that is an intel feature!!! Those rascals for stealing it!!!

I have a neo4 and it doesn't have near that so it may be a sign of a bad board or issues like idontcare had...

Quite possibly a bad board, I blew a crappy foxconn out in about 2 months, and reverted back to this board (this is original board bought in early 06). Upgraded from a 430W ThermalTake to a 650W Antec TP3 on the psu side as i added a few more HD's. With stock voltage (1.35) i can run at 100% load with a smaller vdroop obviously.

I'm planning on building another rig soon so im not too worried about the problem anymore. This chip has served me well and will continue to do well as a guest computer. Just figured i would ask around here to see others opinions.