• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

E6400 8x Multi Experiments: All Together!

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

rainless

Old Member
Joined
Jul 20, 2006
Since there's been so much talk about how the E6400s overclock I figured I'd make a single thread so that everybody with an E6400 chip could chime in. I wanted this thread to be as specific as possible so:

1. You must have an E6400. This isn't for people to post their E6600 scores or talk about how you should have gotten an E6300 instead.
2. I'm confining this thread to 8x multi. I've already done extensive research into 7x multi and all the benefits of running that, now I'm curious about 8x.
3. NORMAL cooling. By normal cooling I mean standard water (no dry ice, or time machine modified water, or Elvin water from the mountain kings of Scandinavia) or air. (High end air acceptable)
4. 1.5 volts ABSOLUTE MAX.

I was running 2800mhz (400x7) at 2.75 volts and getting a max TAT at 60c load and max orthos at 58c load. So I started my overclock (this is on a DS3) at 1.32v and 375x8, hoping to get 3ghz, but surprizingly CPUz only reported it as 374x8 which put me slightly under 3ghz at 2994. Temps were around 64-66 load. So I dropped the vcore to 1.30 and raised the FSB to 380x8 and now I get 3.04ghz at max 100% TAT temp of 62-63.

I should also note that, on the DS3, at 375x8 I was getting that odd whirring noise I reported elsewhere (seems to be a common DS3 problem. Something about the caps they've chosen...) But at 380x8 I get nothing. I think it's a simple question of frequency and harmonics, which is probably why 375 was showing up as 374 but 380 shows up as 380. I could've also tried 376 I suppose, but if I can do 380... then why not?

I'm pretty sure I couldn't do 400x8 at a temperature that I'd be comfortable with, but I have yet to try 1.275x8, which may very well work. At this point I still believe 400x7 is the ideal because you get more out of your ram, but for the purposes of this test 380x8 seems perfectly nominal.

I'd also like to point out that this was not possible with my last power supply. It simply would not boot at anything over say 350x8. So everyone running into walls there I would suggest you change your power supply before anything else. Even if you THINK your power supply is fine. Change it anyway just to be sure. Either that or get a multimeter and measure the true volts.

Edit: Couple of things I forgot. Idle temp is 43/44 and vcore 1.30 shows up as 1.312 in CPUz.
 
I haven't had time lately to dig in and try to push it but,
400x8 on a DS3 1.375V 100% 24hr orthos stable High 20s idle, low 30s load
425*8 @ 1.45V Seems stable on most stuff, but orthos will crash
450*8 @ 1.45V Sometimes boot and I can Pi 1M but thats it. Crashes quickly.

Hoping to eventually get 450 stable and do suicide runs of 500. Would like to bench at least once at 4ghz.
 
Gabbiani said:
I haven't had time lately to dig in and try to push it but,
400x8 on a DS3 1.375V 100% 24hr orthos stable High 20s idle, low 30s load
425*8 @ 1.45V Seems stable on most stuff, but orthos will crash
450*8 @ 1.45V Sometimes boot and I can Pi 1M but thats it. Crashes quickly.

Hoping to eventually get 450 stable and do suicide runs of 500. Would like to bench at least once at 4ghz.


Hmm... Since your temps are okay, what do you think is holding you back?
 
rainless said:
Hmm... Since your temps are okay, what do you think is holding you back?

The board at this point. I can lower the cpu multi, turn up the mem one and the memory will test well past its rated 800 setting.

Just need to play with voltages and timings and such.

Also the northbridge needs some cooling. Increasing the mch past +.1 doesn't seem to help any.
 
rainless you never mention your own cooling. That should be a requirement of your thread as well.

It's possible gabianni is pushing the limits of his cpu even with not much over ambient temps. Or maybe he could get higher by just ramming some voltage down its throat but that might not be the best thing longterm.

hbb - does your cpu oc-ability change by going to 8x multi and lower fsb? saay...8x440ish (3520)? Obviously for normal operatoin higher fsb is better, I'm just wondering if the final cpu oc doesn't change with higher multi.
 
rainless said:
I was running 2800mhz (400x7) at 2.75 volts and getting a max TAT at 60c load and max orthos at 58c load. So I started my overclock (this is on a DS3) at 1.32v and 375x8, hoping to get 3ghz, but surprizingly CPUz only reported it as 374x8 which put me slightly under 3ghz at 2994.
Software or the DAC chip which senses the clock freq is not very precise. So, there is nothing to worry.


Temps were around 64-66 load. So I dropped the vcore to 1.30 and raised the FSB to 380x8 and now I get 3.04ghz at max 100% TAT temp of 62-63.

I should also note that, on the DS3, at 375x8 I was getting that odd whirring noise I reported elsewhere (seems to be a common DS3 problem. Something about the caps they've chosen...)
This is probably because of the chokes/ferrites/inductors chosen and not the caps. The hum you hear is due to hysteresis losses at higer harmonics.

But at 380x8 I get nothing. I think it's a simple question of frequency and harmonics, which is probably why 375 was showing up as 374 but 380 shows up as 380. I could've also tried 376 I suppose, but if I can do 380... then why not?
In my tests, choosing a lower multiplier overclocks the chipset, credit to OCZTony at Bleeding Edge for the research. (Open up sandra and check the clocks. CPUz reports wrong clocks).

I'm pretty sure I couldn't do 400x8 at a temperature that I'd be comfortable with, but I have yet to try 1.275x8, which may very well work. At this point I still believe 400x7 is the ideal because you get more out of your ram, but for the purposes of this test 380x8 seems perfectly nominal.

I'd also like to point out that this was not possible with my last power supply. It simply would not boot at anything over say 350x8. So everyone running into walls there I would suggest you change your power supply before anything else. Even if you THINK your power supply is fine. Change it anyway just to be sure. Either that or get a multimeter and measure the true volts.
Good point. I'll see if there is any change (P5W-DH) with my results. A Zippy GSM6600 is on the way.
 
Well I just ran 420x8 = 3.36 (same cpu oc) @ 1.35v (same voltage) for about an hour playing sims2/quake4 (Really didn't notice a difference from 7x480). Tried 425x8 and windows crashed at this voltage, just like how I couldn't jump up to 485 on 1.35v. Temps are the same either way 40C idle, 52C load.
 
DumpALump said:
Well I just ran 420x8 = 3.36 (same cpu oc) @ 1.35v (same voltage) for about an hour playing sims2/quake4 (Really didn't notice a difference from 7x480). Tried 425x8 and windows crashed at this voltage, just like how I couldn't jump up to 485 on 1.35v. Temps are the same either way 40C idle, 52C load.

It's interesting to see that you're hitting about the same wall on the P5B Deluxe that I hit on the DS3. Must be the normal human cooling limit...

I'll try to answer everybody's question at once:

Madman: I used to have my cooling in my sig... don't know what happened to it. Maybe I took it out after I blew up those two (actually one) boards. Evercool WC-202: Bottom of the Barrel Edition.

HBB doesn't have an E6400 he has an E6300 and some strange cooling or luck or something :)


SuperNade: When you say "Using a lower multi overclocks the chipset..." Do you mean it somehow makes it faster... (i.e. speeds up transfers and such) or just works it harder? Of course using a lower multi and a faster FSB increases the bus speed, but I don't know how using a lower multi and keeping the SAME fsb could overclock the chipset. What kind of research was done?

I've been editing video all day at 380x8. Solid enough for me. Don't really like to overclock my ram too much is I don't have to, otherwise I could be running at 450x7, but I did notice 380x8 seems to be a lot easier on the processor (even though the temps are higher... don't understand that.)

The temps are higher but the processor doesn't seem to be working as hard. Applications load faster. No whirring noise coming from my chipset and so on. strangeness.
 
You can find the relevent article on the Bleeding Edge forums. The gist of the matter is, you chipset freq scales as :
Code:
 Chipset Freq= (Default Multi)/(User Set Multi) * FSB
This is particularly true for ASUS 965 boards. I'll report in detail one I have finished running a few tests.
 
MadMan007 said:
hbb - does your cpu oc-ability change by going to 8x multi and lower fsb? saay...8x440ish (3520)?

like rainless said, i don't have a 8x multi, because i don't have a e6400 :D
 
DumpALump said:
Well I just ran 420x8 = 3.36 (same cpu oc) @ 1.35v (same voltage) for about an hour playing sims2/quake4 (Really didn't notice a difference from 7x480). Tried 425x8 and windows crashed at this voltage, just like how I couldn't jump up to 485 on 1.35v. Temps are the same either way 40C idle, 52C load.

You should try some cpu benchmarks and maybe some different 3DMark runs to compare. I'm sure the sims doesn't stress your system enough to notice ;) and depending on the graphics settings your vid card might be the bottleneck for Q4.

hUMANbEATbOX said:
like rainless said, i don't have a 8x multi, because i don't have a e6400 :D

lol duh, brainfart. Well then, have you done comparisons with the other multi - 6 I think- you do have? Obviously not at your max oc though. Or maybe don't post it because it will be messing with rainless thread requirements heh.
 
if anyone can run 438*8 for 3.5ghz, i'll show some benches to compare. ;)

although SuperNade is right from what i've read. simply lowering the multi overclocks the northbridge. look in Sandra with 8*266, then look with 7*266. :D
 
Super Nade said:
You can find the relevent article on the Bleeding Edge forums. The gist of the matter is, you chipset freq scales as :
Code:
 Chipset Freq= (Default Multi)/(User Set Multi) * FSB
This is particularly true for ASUS 965 boards. I'll report in detail one I have finished running a few tests.


That's FASCINATING... I can't wait to see your results.
 
I have a few numbers, but I'm waiting to see if swapping out PSU's will make a difference. I will post up a few screens when I get back from school.
 
I'm not quite sure I understand what you mean, I always thought increased FSB simply meant faster operating chipset. Kind of like the chipset is a cpu and it runs at a 1:1 ratio with the FSB. I guess that's what you're saying :confused: I'll be interested to see more.

Aha! I think maybe I understand, in the midst of writing this post...changing the multi is like changing a memory:fsb ratio - you run a certain bus speed but the chips (memory or chipset chips) run at their own higher internal speed on the other side of the bus. I'm used to Intel systems with a locked multi, the 1:1 chipset:fsb statement should hold true for them but not when you can change the multi. So dropping the multi AND running at an even higher fsb to get the same oc sort of double-overclocks the chipset chips: once for the higher FSB and again for the multi ratio.

Got a link to the bleeding edge thread? This is quite interesting I can't wait to see more.
 
That is some interesting and wierd stuff to say the least. Supernade start a new thread for your results we've already pretty well hijacked this one to another topic :( lol
 
Yup... Yup... Used to be my thread... Something about E6400s... I forget... :)
 
Back