Sucessful Sandybridge build @ 4.7. Going for 5ghz on a 2500k?

adamt19

n00b
Joined
Mar 29, 2011
Messages
30
I've "seen" it with the 2600's (online), but is it happening with the 2500s?

Hi everyone - first thread for me on [H]. I used to build PC's every couple of years when I was a kid traveling around playing CS, but went off to school, got into Macs for my web design career.. and my gaming PC was left collecting dust.

But recently - after playing the Crysis 2 demo on XBL, I caught the bug again [:)]

Here's what I put together this week, after a bunch of reading and research (and by that I mean that I decided I wanted a PC -- AND ordered all the parts, within a 4 hour period.)

I'm prepared for all of you to tell me I bought all the wrong shit. :p Anyway:

Mobo:
Asus P67 TUF Sabretooth

Cpu
Intel Core i5 2500K - Stock fan

Mem
4gb Corsair XMS3 (I know, I cheap'd out. More on this below)

Psu
XFX 750w Black Edition

Vid
XFX HD 6950

HD
1x 300GB WD V.Raptor (storage)
1x 120GB OCZ Vertex 2 (OS/games)

Monitors:
My old 20" Apple Cinema
another 23" Asus LED (2ms)

Case:
my old Antec P180

.. and a $40 asus cd drive.

Put it all together.. fired up Crysis 2, had my "wow, my computer can play this game finally" moment.. and thought I was done spending money.

But the bug wasn't finished with me. I started OCing my system.. realized 4.7 on the stock cooler was do-able, and realized a 150-point gain in 3Dmark11 between OCing my CPU/Vid card.. along with a noticeable framerate increase.

Began wondering what else could be had for minimal additional investment.

Then I went to work yesterday morning and Newegg (as they always do) put me over the edge with an email promo. So, I ordered 16Gb of CL9 Ripjaws yesterday (meh, $130), and it arrived today - along with the upgraded CPU fan that I decided to throw in.

I went with the Scythe Mugen 2 rev. b. And uh, this thing is really subtle, let me tell you.

Here are some pics:
189030_555274997889_59801023_32124581_5791133_n.jpg


youve all seen it, no doubt
IMG_1447.jpg


all set to go
198430_555415985349_59801023_32126824_182401_n.jpg


ran with that for a couple days, then ordered the new stuff, and here's more pics:
real-subtle.jpg


real-subtle-2.jpg


real-subtle-3.jpg


now some questions -

What's passes these days for "Ok I can run this 24/7?" A certain intel burn test arrangement? Hours of prime 95?

I ask because 4.7 was "achievable" on the stock fan @ 1.31v, with everything I threw at it BESIDES the intel burn test. The Intel burn test gave me BSOD and temps would jump up to over 80c. I didn't let that happen for very long, of course.

Is this test just abnormally harsh? Or, should this test really be ALL I'm using to judge the true stability of my system?

I've noticed with this new fan set-up, which actually blows hot air off the heatsink directly into another case-exhaust fan, that my temps are -20c across the board. I'm super happy with this cooler, ESPECIALLY for $40! It's massive, a bit tricky to install (had my computer-building neighbor from upstairs lend a hand), but the performance/$ is right on-par with the reviews I read, where it came in just behind $100 coolers.

Anyway - I want to know if I anyone out there is seeing 5ghz (non water) on the 2500's?
 
Last edited:
Nice stuff there dude. 5GHz with the 2500K is not that uncommon with good cooling, including air coolers. I use Prime95 blend and AIDA64's stability test to check max voltages, temps and stability. Some people say to use IBT but I am personally wary of it because the author writes hacks for computer games and calls himself AgentGOD (lol), and all it does is wrap linpack anyways, so if you want to use it might as well use linpack.
 
what temp am I aiming to stay under at max load? 75c?

I can boot with a 50multi, at 1.35.. but it eventually climbs to 80c (and I kill it immediately).

Am I letting it get too hot, at 80, during testing?

Im forcing PLL overvoltage to enabled, and set LLC to ultra high. Other than that - just the multi and vcore. I missing anything else? I did find that thread from Jose @ Asus. Will have to dig more into it later.

Will try lower voltages and more testing tomorrow, its 2:30 and I'm just happy to see 5ghz for the first time ever :)

ill-take-it.PNG
 
85C is probably a good stopping point for stress testing if you want to push the envelope with air cooling. It will never get that high in real world usage. Even if you reach 98C, all that will happen is the CPU will throttle down automatically. And 120C (or thereabouts) is when the CPU will actually shut off to prevent damage. You have to really do something stupid like run with no thermal paste or no heatsink to damage modern Intel CPUs.
 
seriously?

this... is excellent news. Can't wait to come back home and find the sweet spot with this rig. Thx for the ifno
 
I didn't touch TDP at all - now I'm reading to up the max TDP to 200w, if possible in the bios. Not sure if that was available as a hard setting in my (asus) EMI bios..

also reading that I should hard-set the 1600mhz ram speed, otherwise it's going to run at 1333?

The game done changed? This is weird.
 
Isn't a vcore 1.35 rather low for a 5ghz boot?

I feel like I shouldn't have been seeing 80+ temps with that voltage. Wondering if there's something else I'm doing wrong. Perhaps the mounting of that cooler? It was tricky but I thought pressure was good..
 
something is wrong i achive 8.97. at 4.9ghz



EDIT after looking closely i see his memory speed is only 1333 or so. memory speed defiantly holding him back.
I think so.He did kick the memory to higher frequency later but didn't show any Cinebench result.:D
 
Isn't a vcore 1.35 rather low for a 5ghz boot?

I feel like I shouldn't have been seeing 80+ temps with that voltage. Wondering if there's something else I'm doing wrong. Perhaps the mounting of that cooler? It was tricky but I thought pressure was good..

Did you enable the pll override
 
I did tonight, yes - and I'm no longer seeing 80 degree temps.

1.35v 100x49 idletemp:[34] Loadtemp:[70C]

1.35v 100x50 idletemp:[36C] Loadtemp:[72C] (then BSOD after only 30 seconds of prime)

1.37 100x50 idletemp:[37C] Loadtemp:[74C] (then BSOD after 10 minutes of prime)

1.375 100x50 idletemp:[39C] Loadtemp:[74C] (then BSOD after 3 minutes of prime)

Did I hit teh wall? :/

Had to go back to work and play with it more later
 
You're only at 1.375, lots of room to play with. Mine takes 1.46+ for 4.8GHz... and still BSODs after 9 hours. Of course I have a different board (that is way past due for a BIOS update).
 
Like I said above, under stress testing 85C is about the most I would go. Mine tops out around 75C at 1.46V with a liquid cooler, so I feel I could push even further but I'd rather wait and see what a BIOS update does to voltages/stability on my board. Real world temps should be more around 50-60C in normal loads like games.
 
temps seem a little high i hit 68C max under prime95 with 2500k and cosair A70 at low fan. i did do some lapping an using shin estu. my cooler is not as good as yours. my max ambient is 25C.

is your Vcore what CPUZ is reporting under load or bios ?

edit: i am at 1.408 max with cpuz @ 4.9, idle is 31-33
 
Like I said above, under stress testing 85C is about the most I would go. Mine tops out around 75C at 1.46V with a liquid cooler, so I feel I could push even further but I'd rather wait and see what a BIOS update does to voltages/stability on my board. Real world temps should be more around 50-60C in normal loads like games.

oops my bad I didn't follow the usernames - you guys need avy's in here, everything blends :)


temps seem a little high i hit 68C max under prime95 with 2500k and cosair A70 at low fan. i did do some lapping an using shin estu. my cooler is not as good as yours. my max ambient is 25C.

is your Vcore what CPUZ is reporting under load or bios ?

edit: i am at 1.408 max with cpuz.

68C with which settings?
edit: OK on the voltage, but what multiplier? Or does your multi not matter? In other words, is the voltage alone responsible for the temps, regardless of your multiplier? Or is it a result of the combination?

The voltage ratings I posted were bios settings - so it's dropping under load but not by much.

I used the therm. that came with the Mugen2. Made sure to clean the old stuff off with WD40 then Alcohol. I didn't lap the cooler though.. used about pea-size drop of the therm.
 
Last edited:
oops my bad I didn't follow the usernames - you guys need avy's in here, everything blends :)

Lol, NP. Avatars are overrated anyways, they just distract from the discussion :D

68C with which settings?
edit: OK on the voltage, but what multiplier? Or does your multi not matter? In other words, is the voltage alone responsible for the temps, regardless of your multiplier? Or is it a result of the combination?

Voltage is definitely NOT the only contributor to heat. In fact voltage is nothing without wattage (power draw), and increasing the multiplier is what increases wattage and thus heat.

I used the therm. that came with the Mugen2. Made sure to clean the old stuff off with WD40 then Alcohol. I didn't lap the cooler though.. used about pea-size drop of the therm.

I'm just going to continue posting this everywhere I see someone use the term "pea-sized drop" in reference to thermal interface material:
TIM_Before_Spread.jpg


:D
 
HAHAHA

ok, you got me - I didn't use a pea-sized drop I suppose. It was closer to what is used in that picture.

I see some people spread the thermal paste on the bottom of their heat sink using a plastic card.. instead of dropping it in the center. Shouldn't make a difference should it?
 
I prefer to not spread it and let the action of installing/twisting the heatsink slightly to spread it out. Less chance of user error, contamination, introducing air pockets, etc. It also depends on the TIM and the heatsink. The entire CPU heatspreader does not need to be covered, the important area is above the cores themselves. Google "best thermal paste application methods" and click the first link to get to the guide that picture came from. For some reason [H] blocks it so I can't link to it directly. http://www.google.com/search?btnG=1&pws=0&q=Best+Thermal+Paste+Application+Methods+
 
Last edited:
That has got to be the coolest motherboard I've ever seen. I love the look of that thing. Good job on the overclock, and I love your hardware set up
 
Thanks! I really fell for it too. Something about hiding all the fragile hardware of a motherboard appeals to me. Maybe it's the fact that I'm always worried about bending capacitors, or something.. but the "thermal armor" (whether it actually helps cooling is up for debate) is sexy, I agree.

Tonight I made some changes..

Moved a fan from the top of the case to the front. Flipped it around and mounted it where my hard drives were because I needed an intake fan. Moved the HD's to the lower bay.. getting power over there was tricky but I found the-only-way-possible, eventually.

01.jpg


02.jpg


03.jpg


04.jpg


05.jpg


06.jpg


I've been doing light work on the machine with voltage at 1.285 and a multi of 48. But after getting into Crysis Warhead for 10 minutes, I got BSOD..and figured I needed some more voltage.

So I bumped up to 1.3 and have been playing for about 2 hours.

I quit the game, and came back to a WARNING about the following two components being pretty hot. Please let me know if these are cause for real concern, and how I may lower the temps at these areas:

PCI-E: 72C
USB3: 62C.

I wasn't using USB3 until tonight. I just happened to plug my mouse in there tonight.. Not sure if that matters..

I'm gonna go into bios and make sure my voltages are default aside from Vcore. I had raised some of the other voltages earlier - when trying to support a 5ghz OC, but I think I've reset them all by now. Going to double check - but assuming they're all at stock voltage - should I be worried about these temps at these areas??
 
actually.. is there any fan I can even fit next to my CPU cooler?

real-subtle.jpg


..b.c. I'm not thinking there is?? The video card is right there, too..
 
Even if you can,you shouldn't mount it perpendicularly with the first fan since it'd interfere with the airflow from the later.
 
good point

anything wrong with this arrangement (cpu fan + case fan)? They're blowing in the same direction
real-subtle-2.jpg


would flipping the cpu fan around, and mounting it on the other side be a bad idea? (same direction as the case fan, but too close together..right?)
 
CPU fan and case fan must blow in the same direction (toward the rear of your case).

If you only use 1 fan,let it blow air through the heatsink.Letting it suck air through the same heatsink would be a bad choice.
 
CPU fan and case fan must blow in the same direction (toward the rear of your case).

If you only use 1 fan,let it blow air through the heatsink.Letting it suck air through the same heatsink would be a bad choice.

So then I have it about as good as I can get - so long as I run this cpu cooler, and in this case.

I suppose I could buy another fan to replace the one I stole from the top of the case. But I've dropped down to 4.7/1.29v and I think I'll stay there anyway
 
Another way to mount your fan would be mounting it right above your video card.It'd blow the air upward to your top fan.This solution is better when you have top fans in your case.

That will require you to rotate the HSF 90 degrees to the right side.
 
I could certainly move the back fan up to the top, and move the HSF to the 'under' position, blowing through the heatsink, upward.

I suppose that would actually be the ideal arrangement (if I'm going to leave it at one case fan in that corner) - thanks!
 
Only do it if your top fan has better air flow than the rear fan (higher CFM).Sorry,i forgot to mention this.
 
Well I moved my top fan to the front of my case, for intake.

So I only have the rear fan - but could move it to the top.. and move my CPU fan into position so that they're both blowing toward the top of my case.

IF that's going to do anything for me vs. the current arrangment
 
If that's the situation then you should keep the current setup.No need for changing until you got some high CFM fan for the top position.
 
The fans in the case are Antec Tri-Cool 120mm, which are manually adjusted L-M-H, up to 2000rpm/79cfm

time to look into stronger fans. These use molex connectors and aren't mobo-controlled. Time to dump 'em. I guess.

I want one seriously strong fan for intake at the front, and perhaps two less expensive fans for the rear-rear and rear-top.

Suggestions welcome! but you guys have been more than helpful already - thx
 
Back
Top