View Full Version : I did it, hit 4.0 finally. However, are these voltages stable?
jwei
Jan 5th, 2009, 09:50 PM
Hello back again
Finally picked me up some Patriot Vipers (DDR-1066) 5-5-5-15.
Other specs:
Asus P5Q (Regular)
Intel E8400 C0 w/ OCZ Vendetta 2
I can finally hit 4.0 stable (intelburntest).
However my settings are:
FSB: 445
Vcore 1.4
NB: 1.4
SB: 1.2
Ram @ 1080 or something
Timings set to auto.
Is 1.4 on both the Vcore and NB too high?
Also, after Vdroop, it hits about 1.378-1.865 or so
Perhaps if I manually set the RAM timings it would allow me to lower voltages?
SLee
Jan 5th, 2009, 10:34 PM
1.4v VCore, seems reasonable. I have a E0 E8400 that also required 1.4v to hit 4GHz. You shouldn't need 1.4v for 445 FSB with a P45 based MB though. You also try lowering the memory speed to DDR2-900 first to make sure it's not causing you any problems. Once you have CPU running stably at the lowest voltages possible, then go on to overclocking the memory.
shawn99
Jan 5th, 2009, 11:32 PM
how long does it take to encode a movie from say x264 to WMV-HD
loybond
Jan 6th, 2009, 12:36 AM
Well, ideally, you shouldn't go about 1.38 or so. Have you tried getting it stable at 1.37 or so?
What are your temps with Prime95 small FFTs?
Basically, the popular opinion is that 45 nm CPUs experience degradation above 1.38 or so. Your NB is ok at 1.4, it's not as sensitive, though you shouldn't need 1.4 for that FSB with a dual-core. I need 1.4 on my quad for 471, but that's a different story. What's your CPU PLL? Have you set that to auto? If not, keep it to 1.57, no more.
Vdroop... you have a P5Q, doesn't that have load-line calibration?
jwei
Jan 6th, 2009, 01:06 AM
Well, ideally, you shouldn't go about 1.38 or so. Have you tried getting it stable at 1.37 or so?
What are your temps with Prime95 small FFTs?
Basically, the popular opinion is that 45 nm CPUs experience degradation about 1.38 or so. Your NB is ok at 1.4, it's not as sensitive, though you shouldn't need 1.4 for that FSB with a dual-core. I need 1.4 on my quad for 471, but that's a different story. What's your CPU PLL? Have you set that to auto? If not, keep it to 1.57, no more.
Vdroop... you have a P5Q, doesn't that have load-line calibration?
Currently debating whether to run ORTHOS or p95 over night. I ran intelburntest and passed all 5 nicely. Idling at about 33/37 load (during Burntest...the hottest **** program) around 62/58.
I tweaked settings a little bit....FSB @ 444. Vcore 1.40 still. NB down to 1.39. CPU PLL @ auto right now, should I change it?
It was Vdrop sorry, not droop. Load-line is enabled.
Perhaps if I ran my memory at 800 with tighter timings I could lower Vcore?
Or even if I manually entered better timings in?
Should I attempt to lower
matkun
Jan 6th, 2009, 09:26 AM
Try OCCT: http://www.ocbase.com/perestroika_en/index.php?Download
I've never seen anything heat up or stress a CPU more then it does.
matkun
Jan 6th, 2009, 09:28 AM
Perhaps if I ran my memory at 800 with tighter timings I could lower Vcore?
Or even if I manually entered better timings in?
Should I attempt to lower
Changing VCore has no impact on your memory, so messing with memory settings will not let you lower vcore if your cpu is not stable when you lower it.
B0000rt
Jan 6th, 2009, 10:11 AM
Umm, whats your VDimm?
Have you tried a lower Vcore? I'm debating a E5200 for myself so I won't really be limited by FSB like what it seems is happening with you?
Researched it abit more.
M0 for the lower end, your C0, and the best is E0... Hmm I will wait.
loybond
Jan 6th, 2009, 06:17 PM
Try OCCT: http://www.ocbase.com/perestroika_en/index.php?Download
I've never seen anything heat up or stress a CPU more then it does.
You should try Intel Burn Test then!
loybond
Jan 6th, 2009, 06:20 PM
Currently debating whether to run ORTHOS or p95 over night. I ran intelburntest and passed all 5 nicely. Idling at about 33/37 load (during Burntest...the hottest **** program) around 62/58.
I tweaked settings a little bit....FSB @ 444. Vcore 1.40 still. NB down to 1.39. CPU PLL @ auto right now, should I change it?
It was Vdrop sorry, not droop. Load-line is enabled.
Perhaps if I ran my memory at 800 with tighter timings I could lower Vcore?
Or even if I manually entered better timings in?
Should I attempt to lower
Well, manually setting your CPU PLL will probably allow you to decrease your vcore, if everything else is stable. Playing with gtl references as well, but that's more necessary on quads.
Running at 800 with tighter timings shouldn't have an effect on your CPU's vcore. Plus, tighter timings will probably stress the NB more, so depending on your RAM, you may even need to up the NB voltage for that.
When lowering your cpu vcore, if you're unstable, chances are it is your cpu, but double check. Small FFTs will tell you... if that test runs OK, it means your CPU is doing alright. Large FFTs generally stresses your NB and RAM, so if you can run small and not large, you know what to do.
loybond
Jan 6th, 2009, 06:21 PM
Umm, whats your VDimm?
Have you tried a lower Vcore? I'm debating a E5200 for myself so I won't really be limited by FSB like what it seems is happening with you?
Researched it abit more.
M0 for the lower end, your C0, and the best is E0... Hmm I will wait.
E5200s can be monster overclockers!
OP is definitely not FSB limited at this point, at least from the POV of CPU and mobo. He now has faster RAM, too, so that's not it either. Not every chip will be happy at high speeds. FSB limited would mean that he can't go further with his FSB, but I doubt that. If he's running at 444 right now, he still has a lot of headroom. OP can test this out by trying out 500 x 7 or something like that. I'm fairly certain it'll do it. My E8400 once ran at 533 x 8 or something... took 1.4 to do it (not implying all chips can though, just pointing out that the odds are low that his would top out at 444 vs. 533 or more).
BTW, OP, that reminds me... try a different ratio. Try 8.5 x 471 or 8 x 500. It may not be different, or it might be. And if you find that you still can't drop your vcore below 1.40, I'd recommend finding a speed at 3.8 or 3.9 where you can have a lower vcore. Often times, a chip will do 3.8 or 3.9 at fairly low volts, 1.35 or less, and will need 1.40 to crack 4 GHz. My current Q9550 needs 1.28v or so for 4 GHz, though I have it at 1.30 just to give it some headroom. To get it to run only 200 MHz faster, at 4.2, I need to give it 1.4v!
silentio
Jan 6th, 2009, 06:44 PM
What's the best overclocking Quad right now?
I dont see myself gaining much from a 3.2ghz dual core to 4ghz dual core.
akbar_k
Jan 6th, 2009, 07:46 PM
1.4 on 45nm? You're gonna burn that sucker out in less than 3 months. Do another run of Orthos around then and see if it passes...
B0000rt
Jan 6th, 2009, 07:48 PM
E5200s can be monster overclockers!
Can be, is the operative word here, I checked over at Anand and it seems most people top out at 3.5ish.
jwei
Jan 6th, 2009, 10:15 PM
1.4 on 45nm? You're gonna burn that sucker out in less than 3 months. Do another run of Orthos around then and see if it passes...
For sure? I'm sort of nervous for that to occur as well. However I've heard plently of people running at 1.4 24/7 stable. I'll run Orthos tonight and report back.
board123
Jan 6th, 2009, 11:05 PM
Can be, is the operative word here, I checked over at Anand and it seems most people top out at 3.5ish.
They're low-end garbage bin chips, so a lot of them probably wouldn't overclock too well. E5200 is purely luck of the draw.
B0000rt
Jan 7th, 2009, 09:14 AM
They're low-end garbage bin chips, so a lot of them probably wouldn't overclock too well. E5200 is purely luck of the draw.
Any idea on the E5300s? They're apparently a different stepping, R0 instead of M0..
Could it be that R0 is to M0 as E0 is to C0 for the E8400s?
Anyways, sorry to derail, but I wouldn't run 45nm at that voltage... I was scared to do so with my 65nm at that voltage hah
board123
Jan 7th, 2009, 11:30 AM
Any idea on the E5300s? They're apparently a different stepping, R0 instead of M0..
Could it be that R0 is to M0 as E0 is to C0 for the E8400s?
Anyways, sorry to derail, but I wouldn't run 45nm at that voltage... I was scared to do so with my 65nm at that voltage hah
They're basically the garbage bin chips (again, pun intended) of of E0. There's really no way to tell how they'll fare.
B0000rt
Jan 7th, 2009, 12:02 PM
They're basically the garbage bin chips (again, pun intended) of of E0. There's really no way to tell how they'll fare.
Truth, but since E0s are crazy, one would thing the R0s would be half as good, maybe fare worse than C0s but better than M0s?
Keigotw
Jan 9th, 2009, 12:50 AM
E5200s can be monster overclockers!
OP is definitely not FSB limited at this point, at least from the POV of CPU and mobo. He now has faster RAM, too, so that's not it either. Not every chip will be happy at high speeds. FSB limited would mean that he can't go further with his FSB, but I doubt that. If he's running at 444 right now, he still has a lot of headroom. OP can test this out by trying out 500 x 7 or something like that. I'm fairly certain it'll do it. My E8400 once ran at 533 x 8 or something... took 1.4 to do it (not implying all chips can though, just pointing out that the odds are low that his would top out at 444 vs. 533 or more).
BTW, OP, that reminds me... try a different ratio. Try 8.5 x 471 or 8 x 500. It may not be different, or it might be. And if you find that you still can't drop your vcore below 1.40, I'd recommend finding a speed at 3.8 or 3.9 where you can have a lower vcore. Often times, a chip will do 3.8 or 3.9 at fairly low volts, 1.35 or less, and will need 1.40 to crack 4 GHz. My current Q9550 needs 1.28v or so for 4 GHz, though I have it at 1.30 just to give it some headroom. To get it to run only 200 MHz faster, at 4.2, I need to give it 1.4v!
What motherboard are you using with your Q9550?
kay188
Jan 9th, 2009, 01:13 AM
People have hit higher clocks with an E8400 @ 1.8v.
If i recall, i think something like 4.8Ghz
People were calling them crazy, and it'll go boom and whatever, but it didnt for them.
Yes it's high, and not recommended, but that's JUST an IDEA of what people have done with this CPU.
matkun
Jan 9th, 2009, 09:15 AM
People have hit higher clocks with an E8400 @ 1.8v.
If i recall, i think something like 4.8Ghz
People were calling them crazy, and it'll go boom and whatever, but it didnt for them.
Yes it's high, and not recommended, but that's JUST an IDEA of what people have done with this CPU.
Yes, using liquid nitrogen cooling or some other stupid expensive or loud cooling solution and then they turn it off after their benchmark has run and they have the epeen numbers to post.
Setting something up for daily use that will last a year or two is quite different from a once off record attempt.
mikenogo
Jan 9th, 2009, 11:28 AM
What? When e8400 first came out, i overclocked it easily to 4.2ghz stable 24/7
leaving my computer all day long without any problems at all, of course this is using crucial ballistic 1066mhz pc2 8500.
I got the good pair of rams for ballistics before they went total garbage, same with e8400 good stepping :)