A few E6400/DS3 Overclocking questions

Joined
Feb 10, 2007
Messages
34
Howdy folks,

I've been reading up on this for a while and there are still a few things that I'm not quite clear on so I figured I'd just throw them out here and see if anyone can help:

1. My E6400 @ 3.2GHz at 1.34V is running under full Orthos load at 56-58C, before the AS5 has had time to settle. I assume this is safe, right?

2. I was told to up the FSBv and the MCHv by .1 each to keep it stable in Orthos--is that necessary? More importantly, is that dangerous to my hardware? I have them each upped by .1V currently.

3. Can any of these voltage settings be lowered? I'm so nervous about the increased voltage damaging my components.

4. I'm stable for 8 hours in Orthos with 1.33V on the CPU and +.1V on the FSB and MCH. Is it possible to loosen my memory timings and get stable at stock voltages?

5. Do temps drop over time once a system has had time to "burn-in"? I'm hoping I can get these temps down closer to 55C at full load over time.

6. Are there any settings in the Gigabyte DS3 BIOS that you'd RECOMMEND turning off?

Thanks for ANY answers or speculation you can dish my way!
 
hmm that's actually pretty good. Are you using coretemp for temp measurements?

Also, keeping it at +0.1v, if you're seeing stability, then congrats! That's really not much of an increase at all, although the northbridge heatsink does get hot. If you're worried, either glue a 50mm fan on there blowing down or swap for an aftermarket cooler like the hr-05.

Also, if you ever decide to actually measure vmch, you'll see that if you leave it at "normal" it'll actually scale it up automatically. And I wouldn't be surprised if it's actually +0.3v at the fsb you're running.

As for memory timings, you haven't listed what your vdimm is at, and what default is. Also, what timings are you running right now? I'm assuming it's a 1:1 with fsb. What IC for ram do you have?
 
For RAM I'm running GSkill 2GB DDR6400 at 800MHZ with a 1:1 divider. It's rated at 1.9-2.0V and I've got it set to 2.0V right now. I had it running at 4-4-4-12 but now I'm running it at 5-5-5-12 with stock volts on the FSB, MCH and CPU and Orthos just hit the 4 hours mark stable. Maybe that's what was causing me to error out? (The faster RAM timings). Although, the RAM is rated for 4-4-4-12 at 800MHz so I'm not sure why there would be a problem there.

I'm using TAT and CoreTemp to monitor my temps right now. I have 2 120mm fans, front and back, an Arctic Freezer 7 Pro and a small Antec Flex-Spot fan blowing down on my GPU and Northbridge. It's an Antec Sonata Case, so the airflow isn't great to start but I think I've got it about as good as it's going to get without a ton of noise.
 
Here's a good review showing how memory speed scales with additional voltage.
http://eclipseoc.com/index.php?id=1,72,0,0,1,0

2.2 to 2.3 volts for memory in a properly ventilated case is a good range for 24/7 use without any worries.
Let stability be your guide.

ziddey: How much voltage does CPU-z show when you're running Orthos at 450 MHz X 8? I can run some benches at 3600 MHz but I think I need some better CPU cooling before I can be Orthos stable at that speed.
 
Dave, the reason you're probably passing now might not be due to slacked timings but because you're setting mch to normal. Like I mentioned in my previous post, in the ds3, when you set to normal, it doesn't maintain a baseline voltage. Instead, it'll raise mch voltage by itself with fsb. I can't remember exactly what it was but anywhere up in where I run my fsb, setting vmch to normal results in +0.3v.
'

edit: also with the ds3, get speedfan and look at vcore2. That's actually vdimm. I've tested 2 ds3's and noticed there's differences between set vdimm and actual. For instance, on one, set vdimm is actually 0.1v higher, and the other is about 0.1v lower.
 
Well, I'm now on my 10th hour of Orthos at stock voltages and all I did was switch my memory timings from 4-4-4-12 to 5-5-5-12. I don't know why that would make a difference but it certainly seems to have. If I can get through 24 hours of this at stock voltage I will be the happiest freaking newbie in the world.
 
Got to 12 hours stable before accidentally unplugging my computer. ARGGGGHHHH! Starting over but I'm pretty confident it's stable at this point.
 
Dave,

Sound like you've got a nice stable OC there. Relax and enjoy it. :)

Don't worry so much about damaging your system with too much voltage. You're nowhere near the danger zone. Your CPU temps, too, are quite safe. If you're hitting 58C at full load, you've got absolutely nothing to worry about. Most OCers like to keep their temps at less than 65 (or as low as possible for bragging rights), but as I understand it you're not really in danger of damaging a C2D until you get in the neighborhood of 80C. So you're golden there.

I bet if you bump your vDimm up a notch and give the RAM 2.1v (totally safe), you can probably tighten the timings back to 4-4-4-12 or even lower and stay stable. But in most apps you're not going to notice that much difference (certainly < 10%), so if you just want to leave well enough alone that's cool too.

As for your question #6, yeah, there are a lot of settings I'd recommend turning off in the BIOS. Under "Advanced BIOS features," go ahead and disable the following:

* Limit CPU ID Max to 3 [this is disabled by default anyway]
* CPU Enhanced Halt (C1E)
* CPU EIST Function [definitely disable this, or else you're giving the system permission to lower your CPU multi and voltage depending on load, which is annoying to say the least]
* Virtualization Technology [this is such an esoteric function -- its only purpose is to enable you to run "virtual machines," like running another OS in a window or something... which almost no one does, so it's useless]

I also have "No Execute Memory Protect" and "CPU Thermal Monitor 2" disabled.

From the main BIOS menu, choose "PC Health Status" and take a look at the "Fan Speed Control Method" setting. This is more a matter of personal preference, but I have mine set to Disabled, which makes the CPU fan run at full speed all the time rather than throttling up and down according to temperature/load. That change alone dropped my CPU temps by 6-7C. The tradeoff here is that it's a bit louder.

Hope this helps.
 
Crooked Paul,

Thanks so much for the detailed response.

As far as RAM voltage, I'm only noticing a drop from 7843 to 7780 in PCMark05 when I loosen the timings to 5-5-5-12, so I can't imagine that transfers to much real-world difference. Once I verify Orthos stable at 24 hours I might pop the RAM up to 2.1V and tighten the timings back down to 4-4-4-12 and try again but really, with the amount of time I've spent on this thing, I might not want to bother with it.

As far as temps, I found out late last night that I had the damned Arctic Freezer 7 Pro on backwards! Once I turned it around and moved my spot fan to blow air down on the Northbridge, I have been sitting at 46C to 48C fully loaded in Orthos and 29C to 32C at idle. It didn't make sense for me to have the Arctic Freezer Pro turned the opposite direction of my case fans but I'm a noob at this, as I've amply proven. I figure that once the system burns in a bit and the AS5 settles in, that should drop another couple degrees. I'm totally satisfied with that temp, obviously. :D

I went ahead and turned off everything you suggested in the BIOS--I've heard elsewhere the CPU-limiting items are a pain so I think you're probably 100&#37; correct in suggesting those be turned off.

The most surprising thing in the last 24 hours is that I'm completely stable (well, 12 hours worth at least) at 3.2GHz on completely stock voltage--except for the RAM but that's expected. If I can keep it at stock I would expect to have these components for a long time, right?
 
Been enjoying following this thread. Just put together a DS3 and E6400 and am looking forward to overclocking my first computer.

Apologies for the hijack (I think this has to do with the post above mine though). I'm new to overclocking and I was wondering why I'm noticing decent temperature increases when I only up my fsb. It's my (new) understanding that it's voltage increases that increase the temp, which is why people usually up their fsb as much as possible until they finally need to increase voltage (due to instability).

I think I'ma go ahead and disable the same settings for the moment. The only modifications I've made so far are setting the cpu clock ratio to 8x, the system memory multiplier to 2x, and managed to get through a 3dmax at 3.2ghz with stock voltage, though I know 3dmax is no Orthos ;)

One more quick question. What's the deal with RGB in the bios settings?
 
If you have trouble with Orthos at 400Mhz FSB, try upping it to 401Mhz. It will put you on the next strap up. Many people have had problems on the E6400s with the FSB between 375 and 400. Just bumping to that next step made a world of difference for me. Well, that and loosening my RAM timings.

I'm super-paranoid about upping voltages (probably unnecessarily paranoid) so I'm doing everything I can to keep all but the RAM voltages at stock.
 
Reading more closely what you said, Ziddey, it's really interesting about the MCH automatically going up. So when I had it to +.1 it was locked there, but when it's on normal it's actually going up with the FSB on some level?

Is it okay to leave it on "normal" or am I effectively overvolting my MCH by doing that? Should I lock it back to +.1V?
 
-Ever-:
Temperatures go up as you increase clockspeed. This is normal. If you think about it, it would be really odd if that weren't the case. By running the CPU faster, you're getting it to do more work, right? And since no CPU (or indeed, nothing in the universe of physical objects) is ever 100&#37; efficient, by getting it to do more work you're also getting proportionally more work-byproduct, which is heat.

Of course, increasing the voltage produces a lot more heat. That's why people up the FSB/clockspeed as much as they can before they up voltages: You do the thing that gets you a lot of increased performance and a little extra heat before you do the thing that doesn't get you any increased performance (but probably more stability) and a lot of extra heat.

Hope this makes sense.

I don't understand your RGB question. What is the option called exactly, and what section of the BIOS is it in?
 
Dave:

You are overvolting your MCH by leaving it on Normal and increasing the FSB. But honestly man, this is not cause for alarm. These are enthusiast motherboards. They are built to handle extra voltage within reason, so much so that Gigabyte set them to overvolt automatically under the circumstances where it's appropriate.

Maybe another way to look at it is just simply "if it ain't broke, don't fix it." Your system seems happy and stable. The whole point of the stress-testing you're doing is to figure out whether you've been too ambitious in your OC and you might be risking some system damage. If you pass the stress tests, you don't have much reason to worry.

Also remember that these parts are designed to last a ridiculously long time. Realistically the tradeoff you're looking at here is something like, "I could run it at stock settings and the board will last 10 years, or I can run it at a reasonable overclock and it will last 7 years." Do you really think you'll be using that mobo in 7 years? It's not gonna be a problem.

Ziddey, am I off-base here? I defer to your greater experience.
 
Paul, case design and voltage affect temperature more than just the clock speed. In my experience you can drop the temp by actually dropping the voltage a tad. Your rule of thumb is indeed correct my it is not linear with these motherboards and chips. And idle temp is less affected by it since idle is well, idle.

8446981_1024.ts1171283233000.jpg
 
Is that watercooled, wdn? I'm sitting around 30C idle at 3.2Ghz on air and from what I've read, that's really low.

Also, how did you get Speedfan to recognize the DS3? I can't find a ticket for it in the Speedfan database. Did you just rename the items? And is "motherboard temp" actually the "northbridge temp" by chance? I can't seem to find any program that measures the temp of the northbridge.
 
Is that watercooled, wdn? I'm sitting around 30C idle at 3.2Ghz on air and from what I've read, that's really low.

Also, how did you get Speedfan to recognize the DS3? I can't find a ticket for it in the Speedfan database. Did you just rename the items? And is "motherboard temp" actually the "northbridge temp" by chance? I can't seem to find any program that measures the temp of the northbridge.

Dave: that is air cooled not water cooled -- a Zalman 9700 LED, it is the setup in my signature and I am running it right now. At this moment temps are the same bouncing around 28C +/- as I type this. I am very similar idle temps as you are. Idle temps are also affected by ambient temperature for example if you lived in Death Valley you could not get 28C ambient.

I suggest you go to the SpeedFan web site and get 4.32 beta 8 which has "true digital core temp" readings just like CoreTemp program and TAT. Then go to Configure->Advanced-->Chip IT8712F yes it indeed supports DS3. Even the old version did but the new (beta) is a lot better because it DIRECTLY drives the fan speeds from the CPU core temperature. Yes I rename all the items select it, press the F2 key, type in what you like.

If you are serious about overclocking you should seek to become and expert on SpeedFan first, grasshopper. :D
 
The only problem I have with Speedfan is that it has to stay on 24/7 to control the fans, right? I'm a minimalist when it comes to how many applications I like running in the background.
 
You are correct, if you want to use SpeedFan on "automatic" to thermostatically control your fan speeds, it needs to be running at all times. It does not use much CPU however. If you run it to manually set your fan speeds say to 70% and keep them there, when you exit the program the fans are locked at that speed until you reboot.
 
The motherboard fan control all by itself works fine for most people. I am talking more about the "extreme overclocking" range and the point at which the finer grained control over fan speeds & temperature becomes worth the hassle of setting up SpeedFan;)

A lot of people let their fans run 100% all the time -- nothing wrong with that either.

I like to keep mine real quiet unless the system is running at high throttle it saves wear & tear and saves my hearing.
 
Crooked Paul-

Thanks for the explanation. I dig your teaching style ;)

As far as Gigabyte's R.G.B. goes, it's an option under MB Intelligent Tweaker (bios) and you choose one of three settings, being Auto, Fast, or Turbo. Here's a link to the explanation, and I'm curious if it's legit or if using it instead of manual tweaking is comparable to using Easy Tune vs. respective manual tuning.

http://www.gigabyte.com.tw/FileList/NewTech/old_motherboard_newtech/Tech_20041125_PX_RGB.htm
 
Dave:

You are overvolting your MCH by leaving it on Normal and increasing the FSB. But honestly man, this is not cause for alarm. These are enthusiast motherboards. They are built to handle extra voltage within reason, so much so that Gigabyte set them to overvolt automatically under the circumstances where it's appropriate.

That's exactly what I did, set voltages (except the cpu) to normal, when it came time to start using my new computer, instead of playing around. Giga wrote the bios, made the board, and supply the warranty, so I have no problem trusting their numbers!! My NB and SB temps are fine. If you want to keep learning/playing, then set them yourself. I personally was just spending too much time on trial and error tuning.
 
Crooked Paul-

Thanks for the explanation. I dig your teaching style ;)

As far as Gigabyte's R.G.B. goes, it's an option under MB Intelligent Tweaker (bios) and you choose one of three settings, being Auto, Fast, or Turbo. Here's a link to the explanation, and I'm curious if it's legit or if using it instead of manual tweaking is comparable to using Easy Tune vs. respective manual tuning.

http://www.gigabyte.com.tw/FileList/NewTech/old_motherboard_newtech/Tech_20041125_PX_RGB.htm

Ever, I've fiddled with rgb before. I usually have pcie set to 110 in the bios. When I put rgb to turbo and booted to windows, I noticed crazy clocks for my video card. Stock, it's set to run 500/690. With rgb to turbo, it was setting clocks to roughly 550/760 or so. I guess that's the 10% bump it did in. I'm not sure if rgb to turbo just enforces a 10% overclock or if it's because I'm running pcie at 110%. Eitherway, I overclock my card to 720/880 anyway so rgb seems pretty useless. I'm not sure if it affects any other aspects of video card performance though or just simple overclocking...

Regardless of that, my ds3 bit the dust for the last time. I pulled the trigger on a p5b vanilla earlier today. Can't wait to be out of ds3 hell...
 
Back
Top