ghulam_phantom

i have a problem with my overclocking

1 - When i increase vCore in the BIOS, the vCore in CPU-Z is always lower and A LOT lower than that in the BIOS,

Current Condition:
• My vCore in BIOS = 1.40
• and that in CPU-Z = ~ 1.296 ~ 1.312 (varies)

2 - I cant stabilize my processor to work on 3.6 GHz..UNTIL i actually let the VCORE to AUTO, then it works but the voltage reading (in bios) on AUTO is 1.42..

• so is 1.42volts normal with 3.6 GHz?
• because whenever i choose the voltage manually, even a LITTLE bellow 1.42, it WILL fail the prime 95 test. Right now im @ 1.400volts and the prime test fails after like a minute of working..please NOTE that only ONE core fails to stop working..the core#0 keeps on working for long..i dont know what that means..

I'll be grateful if you guys reply.. Thanks
PS>my ram is unlinked and FSB(QDR) to MEMORY(DDR) is 1600 : 750 (5-6-6-17)

__________________
Phantom.X

TwoCables

That's indeed quite a big vDrop, but never fear: all that you have to worry about is the voltage that you see in CPU-Z. That's the actual voltage that's going through the CPU. So, if 1.40 yields 1.296 to 1.312V at idle, then that's the actual voltage. But when the voltage droops upon full load, then that's vDroop. If your motherboard has Vdroop Control, then enable it. I have a feeling that it does being that it's a 790i Ultra SLI.

Oh, and by the way: you don't have to use Prime95. You can actually use Orthos instead and be able to experience a better GUI. You see, Orthos is the result of taking Prime95, stripping it of its useless/irrelevant features, and then giving it a much better GUI. But the tests are still exactly the same.

So anyway, now that you know that you have so much more voltage headroom, keep increasing it! It's still fairly low!

__________________

ghulam_phantom

Oh..thats such a relief..

So that means i have nothing to worry about the drooping?

i only need to be looking at the CPU Z reading or the BIOS reading?

And the Orthos is a great application. i tried it and it seems like it ran the MIX test (CPU + RAM) for 20 mins and then one core failed..

+

the HARSH test on the CPU only (the first one) that fails almost instantly..or within 30 secs..

so should i increase the voltage more? from 1.400 to above?

btw Thanks for the reply man

__________________
Phantom.X

TwoCables

It's a pleasure to help when I can!

So yeah, always go by CPU-Z for things like the voltage, and clock speeds and everything else.

And about Orthos: yeah, it's awesome. It's the same prime tests as Prime95, except it just makes the whole testing experience much better. The GUI is better, and it's not cluttered with a bunch of features I'll never use. I can also change the default test selection if I want by editing the prime.ini file.

So anyway, when you refer to your voltage, it's easiest for everyone on Overclock.net to understand when you make sure we know that you're referring to the voltage after vdrop and vdroop (that is, to what CPU-Z says). Or, at least it's easiest for me. But I know that so many people see things like, "my voltage is such-and-such, and so here are my temps and stuff", and even though the person might be referring to the voltage they set in BIOS, most seem to assume that they're referring to the voltage that CPU-Z is displaying only to find out like 3 days later that the actual voltage is lower.

So, if you're saying to raise the core voltage up from 1.400, which voltage are you referring to? Even if you're referring to the actual voltage and not the voltage before vDrop, I still want to say that you can most definitely increase it. But of course, keep an eye on the temps. If they get too high, then you might get some instabilty first.

But, I'll post more later.

__________________

ghulam_phantom

Thats very helpful..

I was actually refering to the BIOS voltage..but anyways, i increased the voltage and valla..its stable now

im glad..and btw..i dont get the concept..to get the best performance the CPU FSB must be equal to the Data rate of the RAM right? i mean ratio must be 1:1 right?

for that i just overclocked my ram from 750 to 800 (DDR2) and voltage upto 1.90 to get a 1:1 ratio..so its better than the last 750mhz?

and i have no knowledge about the timings in the memory TAB in CPU-Z..but i can let you see if its right or wrong..or if any application to test the RAM if its stable..becuz this ram came with a stock data rate of 667 Mhertz

here are the clocks

400 Mhz
1:1 ratio with FSB
5
6
6
18
24
2T

P.S
im still running CPU @ 3.6 ghz = 400 x 9 with 1.312volts(CPU-Z)

__________________
Phantom.X

TwoCables

Hmm. I'm still not 100% sure, but I've found that getting a 1:1 ratio isn't as important with the nForce chipsets since the FSB and memory can be Unlinked. For example, I have excellent performance, and my ratio is 10:9 due to my FSB and memory being Unlinked. But, I've heard that it's a little more important to shoot for a 1:1 ratio with the Intel chipsets. But even that seems to be less important to people these days.
Hmm. I'm not sure yet. :/
Is the memory rated for 667 MHz? Hey wait, may I see a screenshot of the SPD tab in CPU-Z? That would be the easiest for everyone so we can find out everything about the memory without constantly asking questions. hehe

Oh, and here's something really cool that I learned about the nForce chipset: try changing the 2T timing to 1T and see how it goes. It should require a little voltage increase, and it may require you to change the CAS# Latency to 6, which is currently at 5. In other words, it may require you to loosen it by 1 ms from wherever you get it. You see, going to 1T gives a noticeable improvement in the way the general performance feels. It's actually a little bit amazing that such a seemingly small change can make a difference that you can actually feel.

Not only that, but I also learned that the best thing to do with the memory on an nForce chipset is to get the tightest timings possible with the memory at its stock clocks. For some reason, this yields better performance than just overclocking the clocks as high as possible. But we'll deal with that more once the SPD tab is revealed.

__________________

ghulam_phantom

Here it is..factory speed is 667 DDR2

and btw..I'll just wait for your reply if i can use those settings to enhance my RAMs performance like you described

Cant wait to apply those man..that sounds seriously interesting man..hahaha..just see this SPD and tell me if i can use them..and my volts are at 1.9 at the moment..so if i wanna use those settings, when do i need to increase my volts? if it fails to boot the system? or its must to increase it for performance?

__________________
Phantom.X

TwoCables

Yeah, because what I was thinking based on what I've been taught is to use the stock settings listed in that far-right column, and then maybe tweak 2T down to 1T. But again, that should require a voltage increase and then probably even a loosening of the CAS# Latency. But with the performance increase, it's worth the effort.

And then, of course, overclock the crap out of the CPU.

Edit: And I did notice the sig rig edit to reflect the fact that this is the 680i Asus Striker Extreme.

__________________

ghulam_phantom

Absolutely Superb

it worked like a charm @ 1.9 volts and i dont know if its psychological but i really feel the computer working FASTER than before in front of my eyes hahahah

seriously a great tip..+10



so my final questions..

• Im just starting to wonder..i can still increase the voltage and speed of my RAM beyond any limit..so are there really any limits to these overclockings?
Simply asking ""do i need to worry about the TEMPERATURES ONLY?""

• I'm using an ichill series GPU GTS 260 with 192-SP..and i have 52 temperatures @ max load..so do i need to increase the voltage incase i wish to overclock it ALOT! (via riva tuner) (its an irrelevant question but haha..im sure u have knowledge of overclocking and it relates to the above question)

honestly..i should say that im very grateful for your help..

__________________
Phantom.X

TwoCables

Fortunately, that's real performance there! Do a comparison using the old settings with 2T with the new settings with 1T by doing a Cache And Memory Benchmark in Everest Ultimate Edition 4.60 (it's the only version of Everest that I have). I haven't done one myself, but I showed this 1T vs. 2T thing to somebody in chat, and they decided to do a benchmark comparison and found that something was almost twice as fast. But, that doesn't surprise me since I noticed a boost in performance right away!

Anyway, my jaw dropped because I was really glad that it worked. I just sat there going, "no way. It worked!" I just hope it's stable though. But hey - more on that later!
Hmm. I was taught that overclocking the memory clocks with the nForce chipset is kinda pointless because of this 1T thing. And I can see why it's true: I went from 953 MHz, 4-4-4-12-2T @ 2.1V to 800 MHz, 5-4-4-15-1T @ 2.175V (thanks to linpack testing, because it was at 2.150V). When I increased the clock speed, I didn't really notice that much of a difference at all. I even got it to almost 1066 MHz, but I still didn't really feel a difference. But, of course, the difference was felt when I changed 2T to 1T. I coulidn't believe how much of a difference it was, even though it wasn't like HUGE or anything. Still: I was very happy about it!

So, I would say that you now only have to worry about the temperatures. But I guess if the memory clock can still go above stock with the Command Rate at 1T now, then go for it! Although, my memory clock won't go past stock thanks to the 1T setting. Actually, it will, but testing proves that it's not error-free or stable.
Aw, I really wish I knew more about overclocking GPUs. I've tried, but it just seems like my motherboard doesn't like it. I tried overclocking an 8600 GTS as well, but it just didn't like it. I had the same problems that I have with the 9800 GTX+: it just hates being overclocked.

Oh well. :/

__________________