Highest everyday voltage for DLt3C?

Ok, I'm prime testing my kinda disapointing 2100+ tbred B @ 200x11.5 (2300mhz) @ 1.95v. Temp is a supposed 42C load. I am going to add a "third party" temp probe soon though.

So since the defualt voltage is 1.6v most say not to increase long term for more than 15%-20%, should I keep it where it is even though temps are great?

Also, what's the highest voltage I should attempt for a few hours?

depends on the cooling 1.93v is my limit for high end air cooling. water maybe 2.05. using tec's or phase change why not push it. If you can't afford to fry it don't try it.

I have been at 2.0v on the system in the sig for six months now. The 0310upmw needs 2v for prime95 stability at 2.4ghz. The temps are around 45c depending in the humidity.

Do you have a DUT3C and not a DLT3C?

The DUT3C can handle a bit more voltage than the DLT3C.

Sorry, I thought that was in my sig. It's a Dlt3c.

Well the question was about the 2100, but that's useful info about your rig also.

Anyway, I don't think there are any 2100 DLT3C's, that's why I asked.

If that's the case, 1.95v isn't absoluely terrible with good cooling on a DUT3C. It's much safer than on a DLT3C.

DeathONator said:

Ok, I'm prime testing my kinda disapointing 2100+ tbred B @ 200x11.5 (2300mhz) @ 1.95v.

So since the defualt voltage is 1.6v most say not to increase long term for more than 15%-20%, should I keep it where it is even though temps are great?

Also, what's the highest voltage I should attempt for a few hours?

Click to expand...

2.3 is what they're supposed to do man!
Using the term 'disapointing' to describe 2.3 GHz which is the equivalent of a PR rating in the 3000+s is fundamentally flawed.

DLT3C B's (not A's) have been known to do more. Most recent steppings of Barton as well but it is a cause for celebration when they do 2.4, not a validation of expectations.

Epected hit is 2.3 GHz.

Listen to Gautam here:

Gautam said:

Although the percentage increase in each is the same, voltage is a squared factor for heat dissipation. I'm positive that 2v is too much even for a DUT3C. 1.95v may very well be also. The risk doesn't justify the usage of voltages that high...

Click to expand...

Also search for posts by hitechjb, one of the most knowledgable people on the subject.

Its a Dl t3c. I have gone up to the highest my mobo will allow (2.3v I think) for less than 24hrs while first finding a stable/sane Overclock running prime. I turned it down to 2.0v beacause .3v for an extra 100mhz was not worth it. Now at 2.4ghz @ 2v 24/7.

Later....

Nice OC.

Yeah DUT3C I dunno why I said L. I was hoping to hit 2.5ghz considering my cooling. I know it wasn't very likely but I'm an optimissed. Running 2.0v now at 2300mhz, temp is under 44Cish.

Yeah, these just take a ton of voltage compared to the DLT3C's.

BTW, is it a JIUHB or an AIUHB. I bet it's a JIUHB. I figure those are the same as the JIUHB/DUT3C 1700's which hit about 2.2-2.3.

All I know is my dLt3c 1700+ could only hit 1.9vcore+ with the 166 pin mod.....and my 1800+ could hit 1.9+ without any mods.....so if you get a dLtsc that doesn't like the higher vcore try the 133 to 166 pin mod like I did....it mayb not make sense but it worked for me.....fyi I hit 2.46 in windows....2.4 max stable....and my 1800+ hit 2.6ghZ+ I stoped just becuase the I was at my friends house in a 90degree room and I didn't wanna go any higher but I think I could hit 2.6 stable with better ram and temps......so if you can score one of the 1800+ dLt3c.......if not get the 1700+ dLt3c.....2.4 aint bad for a 1.4gHzchip

Yes, it is a rare JIUHB 2100. Weird thing is, it seemed to do 2000mhz at 1.6v perfectly fine... I was getting psyched.

DeathONator said:

Yes, it is a rare JIUHB 2100. Weird thing is, it seemed to do 2000mhz at 1.6v perfectly fine... I was getting psyched.

Click to expand...

I got the exact same thing from newegg not long ago.

2200mhz at 1.775 and 2300 at 1.95v.

My DLT3C JIUHB XP1700 XPMW does around 2400MHZ at just 1.7Volts and that is with 217FSB with no mods of any kind. It also does 2280Mhz at 1.6V. It is on ebay right now in it's last 2 hours if anyone is interested in it. id- darrenregina

I think there not a single number like 1.8 or 1.9 or 2.0V can answer the question. It depends on many factors and situations:

- It depends on a particular chip, chips behave differently 20-30% above nominal voltage (1.5 V). The leakage current of the transistors, heat, the transistors speed distribution and timing delay of the slowest logic circuit paths, ... will determine the max stable voltage and max clock frequency

- It also depends on whether one want to fully experiment with the chip (ok if the chip dies in a month), or want to keep it for 6-12-24 months, ...

- It depends on the overclocking characteristic of the chip at high voltage, i.e. how much frequency gain (MHz) per voltage increase, ..., since every chip slows down due to heat increase at some point, with certain statistical distribution, difficult to predict for a given chip before actual experiment.

- Practically there is no meaning to increase voltage if a chip does not gain in speed. Chips show a diminishing return (on frequency) when it is overclocked to around 10 MHz/C (for Tbred B and Barton).

- Max meaningful voltage and max frequency depend on the type of HSF, the type of cooling used. Poor HSF will slow down the chip (diminishing return as described above) at a lower voltage due to heat and temperature.

- Extreme cooling can let the chips ride to higher voltage and frequency, until hitting the max limit of the chip due to other constraints such as internal heat dissipation, hot spot, leakage current, gate oxide electric field, IR drop, ...

Having said that, for a given CPU, you can generate and plot its overclocking characteristic on voltage, frequency, temperature, ..., from which a better picture of its overclocking behavior, max meaning voltage, limitations and ways to improve can be seen.


Some more details and basics about them can be found in these links:

Relationship of clock, die temperature and Vcore (update)
- What is the active power of a CPU at frequency f and voltage V
- How to estimate CPU static and active power
- Effect of die temperature on CPU clock frequency at a given Vcore
(page 13)

Some numbers to determine max CPU overclocking frequency - Vcore vs temperature,
When do the CPU's slow down? (page 13)
Explanation (page 13)

Max Vcore for Tbred B and Barton (page 5)

How much voltage can be applied to a CPU (page 5)

Overclocking characteristic (voltage, frequency, power, current, temperature) of a Tbred B 1700+ DLT3C with different coolings (C/W) (page 14)

See what I mean about this guy.

... you've got to be careful not to ask him a general question though. Be sure to use limiting factors if you want to get a straight answer :

Assuming, we're talking about
1.) 1700+ or 1800+ DLT3C B, or 2100+ DUT3C B, or 2500+ Barton

2.) We do not wish to get a reduction of highest attainable 12-hour Prime95 Torture Test passable overclock over a period of 6 months to a year (at least).

3.) Average copper heatsink with average air is being used
(Like the cooling in my sig, let's say)


Now try to put some numbers on highest safe permanent Vcore, hitechjb please.

_____________________
2100+ AIUHB 0248 MPMW
256MB Crucial PC2100 + 2x 256MB Kingston (Hynix chip) PC2100 @ 6 3 3 2
[152] FSB x 15 = 2280 MHz @ 1.8 Vcore with memory frequency at 100% [152] @ 2.77 Vdimm
37(min) C to 46(max) C Winter ; 42(min) C to 50(max) C Summer
Epox 8RDA+
Thermalright SK-7 with variable speed 80x25mm YS-Tech FD1281259B-2F
BFG GeForce4 Ti4200 8X 128MB; Antec SX-835II case ; 350W Antec SmartPower SL350

I don't think I can give or have an answer to a pre-determined number for your "highest safe permanent Vcore" under your conditions and definition (?), similarily no definite answer for safety max temperature, be it 50-55-65 C.

As discussed in my earlier post, it really depends on individual chips. One needs to do measurements to determine what is the max meaningful Vcore that can put onto a given CPU. I have attempted to give some way to determine the max meaningful Vcore based on CPU frequency and its temperature (Vcore at 10 MHz/C is the break-even point, 2-3 MHz/C is the limit for Tbred/Barton).

Further, for using HSF such as SK7, SLK800/900, I think the practical, or MHz-economic and stability comes into play sooner than the absolute max voltage that would damage the chip internal (irrevsible breakdown). Since using HSF on air, the chip would not be able to clock meaningfully higher with stability somewhere around 2V, +- 0.2 V depends on a particular chip.

The engineering spec from AMD about absolute rating on Vcore of the various Tbred B/Barton are 2.05, 2.15 and 2.20 V respectively for DLT3C, DUT3C and DKT3C.

For "safety or health" kind of question on CPU/chips, it is difficult to quantify the effect of x mV increase of Vcore on y months of the life expectancy of CPU life. Just like who can answer about doing certain things would prolong or decrease the life expectancy of a particular human being, but certain general trend and statistics may.

Thank you for sharing your thoughts.

People usually have a fraction of your understanding of how voltage and temperature affect overclocking. However for 95% of people who come here only briefly, terms like "One needs to do measurements to determine what is the max meaningful Vcore" and "it is difficult to quantify the effect" simply do not register.

They do not register. You can just feel the blank stare followed by "So can I do permanent 2 volt Vcore or not," they'll say fully expecting a two to three sentence answer.

1. Members report not being able to overclock as high as they used to.
Without exception they did above 1.8 Vcore. They usually have one of the three chips 1700 DLT3C B, 2100+ B or 2500+ Barton.

Is that not enough to say to the 95% of people who come here very briefly and have a very limited understading of overclocking: on average, to be absolutely safe, 1.8 volts is the number for max Vcore, adding but it varies.

instead of just

it varies.

2. Tell us about the degradation of highest attainable overclock due to too high of a Vcore voltage being used too long. Particularily with the three types of CPUs mentioned and with vcore between 1.80 volts and 1.95 volts?

There are some who never experience degredation of overclock, because all chips differ. I've been lucky enough to have two consecutive processors that do not suffer from this at high voltage. (1.9+) So technically a general statement of structure break down is incorrect. I agree with hitechjb1. But to those less knowledgeable and those who do not want to understand certain principles, it is probably prudent to just recommend generalized average limits.