The Official OC (OverClocking) Thread!

Is there a big performance gap between those 2 since the Zalman costs like 3 times as much as the Arctic?...Holy the Arctic Freezer 7 Pro is soo expensive there...its only $28 up here :S http://www.ncix.com/products/index.php?sku=17225&vpn=FREEZER-7-PRO&manufacture=Arctic%20Cooling
Im guessing that makes it a really good buy

 

i use the Zalman but word is from alot of members (JR) in here that the Arctic Freezer 7 Pro will out preform the Zalman so i wouldn't be afraid to go with that one at all. JR says it will run about 5 or 6 degrees cooler than the Zalman

 

hmm if its true that it cools better I guess it would be a no brainer since its 1/3 the price aswell, its just the kid inside me still wants to buy the big shiny expensive one lol.

 

you can't go wrong with either one

 

I've owned a Freezer 7 Pro and an older Zalman cooler, and I can say both are excellent coolers.

 

i cant give any side by side comparisons as i have only used the freezer pro, what i can do is tell you how mine performs.

right now its sat in a 26c case, idling at 32c with the fan running at 1200rpm which is making less noise than my graphics card does. if i let it spin up to full speed it can drop my idle temp to 30c but i am able to hear it then. i have to fully load both cores for a prolonged time for it to get up 40c

this is on a e4300 running at 3.2ghz on 1.44v

if i am correct russ did some side by side tests with both the zalman and the freezer 7 pro in the same case, mobo, cpu and im sure he found the 7 pro to be the better cooler, but i think he will share his results when he sees this question.

 

Can't comment on the freezer 7 pro, never had one but my 9500 keeps my E6600 at 31 C idle and I can't hear it. It hasn't disapointed me. I too think Russ had better results with the freezer 7 pro though. Not sure what zalman he was comparing to. 9500, 7700, or 7000.

 

Mort81
I think you've just seen a classic example of the difference in performance given by crossing the strap point on the fsb.

I think that the 1333 settings come in at 401 on the fsb as opposed to the 1066 settings at 400 fsb.

This is where people need to get above 450 nearer to a 500 fsb if they are going to cross the 400 fsb mark, otherwise it's not worth bothering.

It takes some serious upping of CPU speed to negate the effects of the strap points.

Not sure what superpi tests you did, but to get a real indication then you need to do the whole 32m, this will show the difference in speed better.

 

BigDK,

My only concern with running the fsb at 400x8 as oppossed to 355x9 is premature failure of my cpu due to the higher fsb wth x8 multiplier. I will gladly live with a little lower memory bandwidth if a higher multiplier and lower fsb will dramatically improve the life of my cpu. I'm going to OC it regardless. I'm just not sure which way to go.
What do you think?

Yeah, I redid it and there was an improvement.

 

My E6600 has been running at 3.6 or above since the first week I got it, that was at the beggining of Sept last year.

It's not happily running at nearly 3.8 and has been for over a month now.

I've also pushed it to over 4.0 Ghz at points for benching, and still it goes on.

You never can say 100% that you won't get a chip that blows, but you're more likely to blow the mobo before the cpu, if you keep the heat down.

The only people I've seen report that they've blown anything, are those that don't have a clue and do something wrong.

The board will easily take 1600fsb QP speed fsb, and then its just how far you want to push the cpu, if you want to use a x7 multi then 3.2 is still a good OC, the fsb will still be using the 1066 strap and the memory running at DDR800 with correct timings.
Then you can keep the voltage down oh the CPU to 1.4x or whatever yours needs.
Mine is quite power hungry compared to some, thats just the way it goes, but then again I'd rather it was that wat, than not a good OCer.

 

Are you saying the fsb strap for 1066 cpu's is better under 400. Maybe I misunderstood. My benchmark results were better at 400x8 than they were at 355x9. The memory bandwidth showed the most improvement because I was able to run it at 800 and a 1:1 fsb:dram ratio.

I'm still able to run the vCore voltage under 1.4 volts.

 

memory at 355x9 888mhz



memory at 400x8 800mhz

 

ASUS has set the NB strap so that the 1333Mhz FSB strap does not come into effect until you get to 401Mhz FSB (1604) why 'I don't know' they just did.

NB FSB straps are
1066 MHz FSB and under,
1333 MHz FSB and under,
1600 MHz FSB and under, etc... etc...

When you go from the 1066 MHz FSB strap to the 1333 MHz FSB strap, the north bridge's internal latencies loosen to allow for greater stability, and again when you hit the next strap.

The results will be better for 400 than 355 as you are still using the 1066 strap but with a higher clock freq.
That 1066 strap remains until you hit 401, then the effects of the strap negate the extra CPU freq until you get over a certain point in the CPU freq.
I.e. what I said about then getting over the 450 FSB mark, giving you the extra raw speed to combat the strap effect.

Anything you add to the freq below the 401 will just add to performance, but this additional freq needs to be considered along side the memory freq.

If you are using DDR800, as I always recommend for OC'ing this CPU, then you are within the reliable settings that the memory can handle, right up to 400 FSB, anything over this 400 FSB, with lesser memory, then not only are you pushing the capabilities of the memory, you are adding the lower latencies of the FSB strap, the memory etc..., and then if not using a 1:1 divider, you add the increased instability of that to the equation.

You then end up with what looks like a better performing PC, i.e., higher clock and memory, but actually poorer performance and less reliability.

Sorry if this is not that clear, but understanding it, and explaining it are two different things, the latter I sometimes fail to do as clearly as I thought.

To sum it up though:
I would go for the 400FSB, which is high as you can go on the FSB without hitting the 1333 strap.
This gives you the highest possible FSB speed, the capacity to run the DDR800 memory at full speed, with the optimal timings, and still the capacity to drop the CPU multiplier down to a level you're happy with.
Even x6 will get the CPU running at default speed, and voltage, but get the memory up to the correct speed.
Anything above the x6 is purely down to required OC to make you feel
That then comes down to what the board is capable of, and we know the P5WDH is easily capable of giving the E6600 what it needs to perform, and then cooling.
If you want to keep the cooling down and power consumption down, then you're looking at 2.4 2.8 3.2 or 3.6, with 2.8 and 3.2 being the ones to look at going for.

Heat was the reason I changed to water-cooling, as I knew the board and CPU could handle the setup, but the heat was a problem, now I've got the heat under control, I'm able to go higher.

I can get my FSB up to 470 because a) the board can do it, b) the CPU can handle it, I got a very good OCer and c) the memory is able to cope.

I think that should explain, but again I may have left something out, as I' normally thinking about 2 or 3 things at the same time.

Currently getting my Xbox 360 backup skills up to scratch:
I can see a new DJ case being required very soon

 

Thanks, I understand the OC'in fundamentals. I was suprised by the improvement in memory bandwidth between 355x9 running 4:5 and 400x8 running 1:1

 

i havent yet changed any of the memory settings but from 3.3ghz (367 x 9) to 3.2ghz (400 x 8) i have seen gains of around 10% in the read, write, copy and latency tests in everest, and a very little drop in the cpu tests.

what can i use to find out the nb voltage?

 

@Mort81
Sorry if I was teaching you to suck eggs there.
I was trying to answer your last question, as it seemed you had an issue with the reasoning behind the bandwidth differences, I was trying to explain why they exist.

Maybe I misunderstood what you were asking?

Regardless of the CPU you're using, its the motherboard NB strap being set at 1066, once you get over 400 you're no longer using the 1066 strap, you're using the 1333.

Most people seem to have no clue at all about FSB straps, and as it's never mentioned in the OC'ing thread, I pressumed people here could do with it to being explained!

@marsey99
You can either use the BIOS to show you, but this is obvioulsy subject to voltage drop, or use something like Everest to show what it is, however reliable that is.
Or you can use a multimeter, but for that you need to know where to take the reading, I know where to take it on the P5WDH but not for your board, you'd need to look it up.

Maybe VR Zone will have it listed.

 

everest doesnt fully support my mobo yet and i cant seem to find it in my bios other than the auto setting (plus the allowed manual settings).

not to worry.

i only recently learned about the fsb/nb strap and that was when i posted the link a few pages back. i found it quite interesting, if a bit heavy going, as internal latency was somthing i had never thought about before.

from my limited understanding of it the people who get the 1333 fsb cpus will have to be very carefull in choice of mobo if they want any head room for ocing. as it seems that the envolope is at its limits for that at stock speed right now.

will edit to add that link again in a min.

that link again

 

BigDK,

Please, no apology necessary. Let me be the one to apologize to you. Yes I do know the basic fundamentals about OC'ing but I'm still learning more all the time. I wasn't aware of the fsb straps. I admire and respect your knowledge. I hope I haven't offended you.

I'm just trying to achieve the best and safest OC possible on air. I'm looking to stay between 3.2ghz and 3.4ghz and the vCore voltage no higher than 1.4 volts. I appreciate and respect any advice you are willing to provide.

3.2ghz+ I know where to set the fsb but I'm not completely sure when it comes to memory frequency. Anything over 900 (450x2) the memory requires more voltage to remain stable. I'm not sure if I should up the memory voltage and run it at fsb:dram ratio 4:5 or lower it under 800 (400x2) and run it 1:1 maintaining 2.1 volts.

 

BigDK,

What would happen if you used DDR2900/PC2 7200? It more than likely would run on the 1066 strap, but could you even clock it to stock speeds. If it would and the buss was at 1066, would you get better performance because of the higher memory speed and the higher FSB?

Happy Computering,
theone

 

RE: esoteric1,

About the CPU coolers, you certainly go wrong with anything from Zalman on "2" major performance angles. Cool & Quiet". The 9700 is an excelent cool for this but a tad large but if it fits, use it if you like it. The Artic Freezer 7 Pro (for Intel) is a bit louder the the Artic Freezer 64 version for AMD. The fan is a lot smaller and thus needs to move much faster to be as effective. They both utilize the newer 4 pin fan connector for AC digital control via mobo & CPU. I like the Thermalright XP 120 for its low profile compact design and ability to utilize many choices of add on fans of many colors & such. It does have a large footprint but if it fits, it works great. I use the copper version with low RPM 3-wire Papts 12db idle/29db load and like it. THe 9700 has a great tunnel desigh and works great in cases with large 120 fan in rear or tunneled ATX case. I use the 9700-NT in a SilverStone TJ06 tunneled canse with great results. I haven't had a chance to test the Artic for LGA775 but the 64 for AMD.