Well, at least with your Gulftown the 2684s don't take as long as what most of us have to contend with. I don't ever see TPFs lower than 50 minutes with the 2684 and it's usually higher than that.
![[H]ard|Forum](/styles/hardforum/xenforo/logo_dark.png){kind=logo}
New bigadv project - 2685
- Thread starter Nathan_P
- Start date
- Joined
- Apr 25, 2001
- Messages
- 16,148
I get 32 min tpf on 2684s with 2 480s folding.
These Gulftowns are just damn near supercomputers when you start overclocking them past 4ghz. However, you really are brave though as i'm not sure I would ever consider running a vcore of 1.42 even with good water cooling.
Remember when I told you that I would sneakernet you my next 2684? Perhaps Stanford took care of that for me, LOL!
tjmagneto
[H]ard DCOTM x2
- Joined
- Aug 6, 2008
- Messages
- 3,299
sense those things are unlocked the vcore you posted and just uping the multiplyer should be almost tweakless @4ghz soo long as your temps are good. and manually setting the memory voltage to 1.65 not as fast as a 200 bus speed but should be very stable.
Yeah at the vcore of 1.33@4.1ghz, I believe my temps shot into the 80's under the Intel burn test and upper 70's with Prime95. I only ran 2 passes with the burn test so I didn't leave it there very long since 2 passes isn't long enough to determine stability. My goal is to make sure the temps never go higher than the upper 60's at most. Right now, they are hovering around 65-67 degrees at stock voltage running at 3.63ghz.
I wasn't too impressed with the H50, but this new H70 is really intriguing.
Kwisatz Haderach
Gawd
- Joined
- Jul 6, 2010
- Messages
- 644
After getting the new version of CPUZ, I found I do not have llc enabled. It was reading my board wrong. I am running 4ghz with a multi of 30 at 1.25 volts, with a vdroop down to 1.244 volts. Using an H50, my average core temp while folding is 66C. I hit 75C on P95 with this setting. I am pretty happy with it.
I came home from work and found I picked up a 2682 this evening so far, I have a TFP of 23:17 on this project.
I came home from work and found I picked up a 2682 this evening so far, I have a TFP of 23:17 on this project.
- Joined
- Oct 23, 2007
- Messages
- 6,105
I think you're being unrealistic (and moreso overly cautious when you don't have to be), but that's just my personal opinion (and clearly your hw on the line, not mine).
FWIW, 60s, 70s, or even 80s with high ambient is not going to harm any i7 out there; they are built to handle much worse.
Also keep in mind - IBT/LinX/Linpack etc., you're going to see temps higher than you will running f@h at full load. I can hit upper 70s in LinX and see no higher than 71 while running f@h on the same system; that's what LinX is all about....that extra notch of stress and pushing temps another notch.
Obviously you run yours however you want, I'm just saying....any amount of volts above stock, even on the best air cooling, with HT on and a minimum of a 4c/8t, you're going to hit 70. Those all in one water coolers may put you lower, but you're still pushing the same juice through the same chip, so in essence you're still doing the same amount of damage, you're just a few degrees cooler.
Personally IMHO I think those Corsair coolers are really overrated. If I wanted a water system, I'd do a water system, the whole 9. I don't buy a sealed all in one kit. I'd buy rads, blocks, tubing, pump, res, I'd do my own bleeding, leak testing, etc. However I don't because I'd rather run air. I may run a couple degrees hotter than someone running an H50 or H70, but I'm still far within safe range and that's all I care about.
~just my $0.02
Last edited:
I think you're being unrealistic (and moreso overly cautious when you don't have to be), but that's just my personal opinion (and clearly your hw on the line, not mine).
FWIW, 60s, 70s, or even 80s with high ambient is not going to harm any i7 out there; they are built to handle much worse.
Also keep in mind - IBT/LinX/Linpack etc., you're going to see temps higher than you will running f@h at full load. I can hit upper 70s in LinX and see no higher than 71 while running f@h on the same system; that's what LinX is all about....that extra notch of stress and pushing temps another notch.
Obviously you run yours however you want, I'm just saying....any amount of volts above stock, even on the best air cooling, with HT on and a minimum of a 4c/8t, you're going to hit 70. Those all in one water coolers may put you lower, but you're still pushing the same juice through the same chip, so in essence you're still doing the same amount of damage, you're just a few degrees cooler.
Personally IMHO I think those Corsair coolers are really overrated. If I wanted a water system, I'd do a water system, the whole 9. I don't buy a sealed all in one kit. I'd buy rads, blocks, tubing, pump, res, I'd do my own bleeding, leak testing, etc. However I don't because I'd rather run air. I may run a couple degrees hotter than someone running an H50 or H70, but I'm still far within safe range and that's all I care about.
~just my $0.02
I suspect my cautious nature with cpu temps stems from the fact that i've owned mostly AMD's and those have a max safe temperature of 62 degrees usually except for the Regors which I believe were safe upto 72. For now, I guess i'll stay cautious but perhaps i'll overcome my fear of high cpu temps soon.
On a side note, I just got done trying to see if I can get 3.8ghz at stock voltage. I am currently running at 3.63ghz at stock vcore. This was a no go as it crashed on the second pass with the Intel burn test.
Yep, voltage is what matters most, not heat. Processors are designed for high temperature operation and as long as you have a good cooler of whatever make, it is highly unlikely that heat-related damage will occur. What will probably happen under excess heat is the system will crash before the CPU suffers any serious harm.
That's my sentiments on the whole liquid vs. air debate in a nutshell.
- Joined
- Oct 23, 2007
- Messages
- 6,105
I had to train myself all over again with the i7's just because before now, the max I'd seen was low 60s in Prime95 on my q9450....seeing 70s scared me initially to the point I didn't want to bother with putting my cpu through those kinds of temps (almost thinking I was punishing it). After doing some more thinking though and seeing a lot more screen caps and whatnot, I realized that it's fairly common (especially with HT on) to see upper 60s and lower 70s on an i7 at 3.8+....that's just the nature of the beast I suppose.
Then I had to re-train myself on volts too, that 1.3 was ok to use and wasn't going to kill it. Or slightly over 1.3 even. (my one is running at 1.325, the other is 1.3.) Obviously now the Intel spec goes to 1.4 so I'm far under spec, but again - I initially got worried about pushing/punishing too far.
I just have to say WOW! Memory speed makes a HUGE difference when working on bigadv work units. Folding2 has a 2685 now that it has finished the 2682. Previously with the ram running at 1333mhz CL7, I had a TPF that was almost dead on at 26 min. After my rig was fixed, it is now running ram at 1600mhz CL8 and the TPF's are 24min 30 seconds. That is a full 90 second decrease in the TPF just from running the ram at a faster speed, even if I am running more conservative timing now! My overclock is still the same before and after@3.6ghz.
- Joined
- Apr 7, 2006
- Messages
- 10,477
You are running a different WU too.
- Joined
- Oct 23, 2007
- Messages
- 6,105
IME it does make a slight difference but pure brute ghz still makes the biggest difference. I noticed a greater difference (drop) in TPF going from 3.8 -> 4 than I did in running a higher bclk and getting the ram to run a little higher. Obviously like nitro said it varies WU to WU but especially now since ram use in the native client is only around the 2gb mark, it's not hitting the ram as hard as it used to when it was in the vm and using 5gb. (again IME at least)