i7 2600K overclocking

[hunter45]

Hi all,

I'm going to be getting an Antec Kuhler H20 620 next week:biggrin: but I'm getting very tempted to try something on stock cooler:sneaky:. I live in NZ and i have heard the chips we get here and in Australia aren't all that good in OC'ing but is it OK if i crank up the settings for 5.0GHz and then see if it gets to the login screen, or is that even too much to get it to TJmax on the stock cooler, or is it something else that will restrict it?

motherboard : Gigabyte - P67a - UD3r - B3

 

[tweakboy]

Your thread heading should have been "OCing to 5ghz with stock fan" ?

Anyhow if you have the K version it will OC to 4ghz guaranteed and even 4.5Ghz , But your not guaranteed you can get it to 5Ghz stable. Also your motherboard matters and PSU be quality. gl

 

[Puppies04]

Going much higher than 4 - 4.2 on the stock cooler isn't a very good idea.

 

[mrjoltcola]

Going much higher than 4 - 4.2 on the stock cooler isn't a very good idea.

+1

4.1 is about max for stock fan unless you get a cherry chip. I have tried 4.2 on stock 2600K cooler and short load (short videos or photoshop work even) will push it past Tcase (72.6C). Don't even think about 50x on the stock cooler. It will throttle down in a heartbeat at much lower clocks than that.

As to 5Ghz in general, only a minority will go to 50x without major voltage (1.500v+). Some chips won't get there period. Nothing wrong with trying, but I personally like to get an idea of how it handles 46x before trying 50x, just to see how the voltage & temperatures look before getting crazy. At 50x temps get hot in an instant.

The thing to keep in mind, you won't get an accurate picture of how it will clock with the stock cooler because the chip wont perform the same hot at 95C as it will cool at 65C. Keep the chip cool to give it a good chance to clock to 50x in the first place. Keeping it cool is nearly as important as voltage for keeping it stable.

 

[BonzaiDuck]

Hi all,

I'm going to be getting an Antec Kuhler H20 620 next week:biggrin: but I'm getting very tempted to try something on stock cooler:sneaky:. I live in NZ and i have heard the chips we get here and in Australia aren't all that good in OC'ing but is it OK if i crank up the settings for 5.0GHz and then see if it gets to the login screen, or is that even too much to get it to TJmax on the stock cooler, or is it something else that will restrict it?

motherboard : Gigabyte - P67a - UD3r - B3

Sad to see you have a P67 motherboard. Here's why.

You can overclock (with stock cooler or better) to the levels recommended by our colleagues in this thread. But a "fast" computer doth not on CPU-speed solely depend. Your system is a hierarchy of different "memories:" CPU registers, L1, L2 and L3 caches; RAM; non-volatile storage -- hard disks. Open up the worst bottlenecks, and you don't need to strain pushing the CPU to its limit -- whatever motherboard you choose.

But the successor to the P67 chipset was the Z68, and I am a "True Believer." For ~$100, you can get an SATA-III SSD with 60GB size -- I recommend the Patriot Pyro. If you use the Z68's ISRT SSD-caching/HDD-acceleration feature, you can hook up a midrange SATA-II HDD to an SATA-II port, and make it the "accelerated" HDD for the SSD caching. This may well give you up to 400% of the HDD's native speed.

You seem too eager to push your Sandy Bridge to 5.0 Ghz. Some caution is in order. I use a top-end heatpipe cooler. While my TCASE temperature seems to reach just over 60C at death-defying stress-testing, the high-end of the vCore voltage setting (when the CPU is still in "turbo" but unloaded) reaches 1.36V. That's where I stop, and I can get to about 4.7 Ghz.

So . . . follow the advice of others here who note the limitations with the stock Intel cooler. You really don't want to volt your processor much over 1.35V, and you don't want the TCASE temperature to exceed 73C. Further, if the temperature gets close to that spec (72.6C) while you are OC'ing, you should be sure you had disabled the adaptive thermal monitoring (TM1? or TM2?) in the BIOS -- to provent throttling during any stress-tests. Then re-enable it after your OC settings are validated as "rock-stable."

 

[Ratman6161]

Sad to see you have a P67 motherboard. Here's why....But the successor to the P67 chipset was the Z68, and I am a "True Believer." For ~$100, you can get an SATA-III SSD with 60GB size -- I recommend the Patriot Pyro. ...If you use the Z68's ISRT SSD-caching/HDD-acceleration feature, you can hook up a midrange SATA-II HDD to an SATA-II port, and make it the "accelerated" HDD for the SSD caching. This may well give you up to 400% of the HDD's native speed.

Sorry, but really there is little to no difference in performance or overclocking between the P67 and the Z68. On SSD caching, many many reviews have repeatedly shown the gains to be had relatively minor. If you have a 60GB SATA III SSD you would be far better off loading your OS on the SSD and data and some non-critical programs on the hard drive than you would be using that SSD for caching. And if you do that the Z68 doesn't really buy you much benefit over the P67.

For people who are buying a new motherboard, by all means you might as well go with the Z68 as there is little to no price premium, but since the original poster already has a P67 board, it would make absolutely no sense to go out and buy a new motherboard just to get Z68.

 

[LiuKangBakinPie]

Look for the Hyper 212. Bang for buck cooler and going for cheap

 

[BonzaiDuck]

Sorry, but really there is little to no difference in performance or overclocking between the P67 and the Z68. On SSD caching, many many reviews have repeatedly shown the gains to be had relatively minor. If you have a 60GB SATA III SSD you would be far better off loading your OS on the SSD and data and some non-critical programs on the hard drive than you would be using that SSD for caching. And if you do that the Z68 doesn't really buy you much benefit over the P67.

For people who are buying a new motherboard, by all means you might as well go with the Z68 as there is little to no price premium, but since the original poster already has a P67 board, it would make absolutely no sense to go out and buy a new motherboard just to get Z68.

I agree about the over-clocking, but not on your remarks about ISRT. Of course, this latter feature would affect the "storage" or "hard-disk" element of the Windows Performance Index -- and it does. With an SATA-III SSD on an SATA-III port to accelerate any SATA-II or -III HDD on an SATA-II or III port, the end result should be "accelerated" HDD performance that is about 80% of the SSD's standalone performance, and "as much as" a 400% improvement over the HDD standalone performance.

Of course you get the best performance from an SSD alone, but you pay ~$2.00 per GB, and a 60GB SSD will limit the amount of programs and data you can store on a single volume. You will then face a choice of putting more programs and data on another SSD (@ $2/GB), or using an HDD under a different drive label -- which will be much slower for what is stored on it. You can also buy several SSD's and configure in RAID0, but you are still paying $2/GB. By comparison, the price-per-GB for a $60 1TB drive would be about 6 cents or $0.06/GB.

By contrast, with the performance improvements I mentioned, you might pay $100 for an SATA-III 60GB Pyro and about $60 for a Samsung F3 1TB SATA-II drive -- for a price-per-GB of $0.16/GB. Given the performance (approximately 80% of an SATA-III SSD's standalone speeds) -- ISRT is a pretty good choice among options.

BUT == I agree as to the OP's choices having already acquired the P67 board. In that case, he'll spend more on SSD(s) in the mix to get equivalent performance.

 

[mrjoltcola]

Two problems with your argument for SRT

1) You assume that everyone in the equation needs a 1TB platter drive; many don't. Most users are served just fine with a 120GB OS drive, and an el cheapo backup platter.
2) Your comparison is wrong because of the tradeoffs: Your SRT setup has no redundancy or backup. A 2-drive SRT setup still needs a third drive to meet minimum requirements for backup and recovery. Whereas an SSD + separate data/backup drive allows full system disk backups with 2 total disks, and the backups are very fast. So raise that price comparison to include a third drive in the SRT model. People who do real work with their PC don't run without backups.

I can't justify SRT and its complexity, I predict that it will become an obsolete technology pretty soon. Business systems, including servers, commonly run RAID-1 OS disks, and nobody I know is interested in adding SRT in this setting.

 

[rbk123]

Not to mention there are 80/96/128 GB SSD's that can be had for $1/gb if you are patient and monitor the Hot Deals section. Those are more than large enough to hold the OS and your application binaries.

 

[peonyu]

Stock cooler cant do 5ghz [maybe 1 or 2 ppl have cpu's that can do it]. I have a Noctua heatsink and to do 5ghz I have to run 1.38v for it to be completely stable and it hits 80c under P95 load...Stock cooler would fry at that voltage. I think I have a good chip though, other people need 1.45v or more for 5ghz. That of course adds even more heat if you get a chip like that.

My motherboard is the "budget" Z68 Extreme 3 gen 3...If it can do 5 ghz with no issues, others can to.

 

[BonzaiDuck]

Not to mention there are 80/96/128 GB SSD's that can be had for $1/gb if you are patient and monitor the Hot Deals section. Those are more than large enough to hold the OS and your application binaries.

Two problems with your argument for SRT

1) You assume that everyone in the equation needs a 1TB platter drive; many don't. Most users are served just fine with a 120GB OS drive, and an el cheapo backup platter.
2) Your comparison is wrong because of the tradeoffs: Your SRT setup has no redundancy or backup. A 2-drive SRT setup still needs a third drive to meet minimum requirements for backup and recovery. Whereas an SSD + separate data/backup drive allows full system disk backups with 2 total disks, and the backups are very fast. So raise that price comparison to include a third drive in the SRT model. People who do real work with their PC don't run without backups.

I can't justify SRT and its complexity, I predict that it will become an obsolete technology pretty soon. Business systems, including servers, commonly run RAID-1 OS disks, and nobody I know is interested in adding SRT in this setting.

I can only say that -- despite the arcane use of a "RAID0" for the SSD itself (and not including the accelerated drive -- a misconception some have] -- it isn't much in the way of complexity. The only thing I've mentioned in some posts is the need to "un-hinge" the accelerated drive temporarily when making installs from large DVD install discs. The "un-hinging" and "re-hinging" is a couple mouse clicks in either direction. And the problem I thought I'd experienced may have been due to something else, or it may be ironed out with BIOS and IRST-software updates.

Personally, I don't have a TB drive in regular use as the accelerated HDD, but may stick one in there when I clone the current 600GB unit.

I'd had a RAID5 -- a full terabyte volume with 4 HDDs and a hardware controller -- on the machine this Z68 system replaces. Now I have two single HDDs -- one for media-capture [MPGs, Media Center MSDVR's, etc] and a buffer for my TV-capture card. The other is the system/boot disk -- which is accelerated. Backup is done with a WHS server (regular backups) and a hot-swap drive bay for cloning the system/boot drive. But since the SSD only consumes about 0.7W of power, I've cut my disk-storage power-consumption in half.

I can't say myself where this will lead. But I've been using an ISRT configuration for this system since June. Haven't even needed to re-initialize the Pyro SSD caching drive at all yet. It's fast and energy efficient. I've got the backup problem covered, as you can see.

Also, if for instance someone wanted a RAID1 setup for reliability/integrity (part of the backup issue), it can be cached with ISRT. A RAID0 can be cached with ISRT, but there you have the vulnerability (requiring backup) that has been mentioned.

Back "on message" -- opening bottlenecks in your system will obviously make it faster overall. Of course, I remain a dedicated over-clocker . . .

 

[Diogenes2]

.............
2) Your comparison is wrong because of the tradeoffs: Your SRT setup has no redundancy or backup. A 2-drive SRT setup still needs a third drive to meet minimum requirements for backup and recovery. Whereas an SSD + separate data/backup drive allows full system disk backups with 2 total disks, and the backups are very fast.

..

What is this data/backup drive thing?

Where is the data backed up to ?

 

[BonzaiDuck]

What is this data/backup drive thing?

Where is the data backed up to ?

I'd encourage Jolt's response here, but in meantime, I'm wondering if he hasn't fallen victim to the myth that somehow the accelerated HDD is part of the "RAID0" configuration with the SSD. The RAID0 is exclusive only to the SSD. The accelerated HDD remains independent, despite being accelerated by the RAID0-SSD-caching. If you lose the caching SSD, and assuming you set up ISRT in "Enhanced" mode ( and not "Maximized"), then loss of the cache or SSD doesn't leave the HDD damaged or unreadable or in a crisis hoping for repair. So the risk of data loss should be identical to the risk of using a single, simple hard disk.

Your own questions seem a bit cryptic, but I'll give it a try. We all want to preserve our data, our files -- our entire boot/system disk. At least if you make regular backups of data/files, losing your boot drive would allow you to replace it, reinstall Windows and your programs, and load up your important data. Cloning allows you to duplicate the entire boot/system disk-volume with a result that can simply be swapped in and used if the first disk fails. We can back up in our LAN -- if we have one, and we can back up FROM our WHS server -- if we have one. We can back up the data files on the WHS server to a hot-swap drive there.

When RAID0 with two HDD's or more became faddish and we were buying hardware controllers with more features, you faced the possibility that one drive going bad meant loss of the entire RAID0 volume. RAID5 solves that, by offering speed advantages closer to RAID0 performance plus enhanced reliability if one drive goes bad. Supposedly with RAID6, two drives could fail but you could still rebuild your boot-drive or data volume.

I think we're getting off-topic here -- way off topic -- and I'm a primary culprit for bring this up in the first place. My point was that a computer system is an entire hierarchy of fast-expensive devices at the top and slow-high-volume, less-expensive devices at the bottom. Opening up bottlenecks of any kind affects overall performance, whether you're running your i7-2600K @ 3.8Ghz in "Turbo" or 4.6 Ghz in "Turbo."

 

[mrjoltcola]

Responding to last 2 comments - No, I didn't fall victim to any myths, I know how SRT works; I just didn't make my point very well I guess, so let me try again.

Everytime someone makes the case for SRT, cost is the focus, or the number one advantage listed. But they ignore that SRT is really comparable to a single disk configuration, not an SSD primary + larger platter data/backup configuration.

I run all of my computers (PC and servers) with a minimum of 2 "failure groups",
meaning 2 separate devices will always hold a copy of my data, if not more. I run scheduled backups, nightly on most PCs, to backup from the primary disk(s) to a larger slow disk, usually a 600GB - 2TB internal platter.

So my point is, above we were comparing:
1) SRT - SSD cache and a single primary disk
2) An SSD primary disk and a large platter disk

That is apples to oranges; why? Because with option (1), SRT, you have a single disk, in essence, so it is a single failure group. By using the SSD as a cache, it is no longer a seperate copy of data, and I can't backup my data without adding a third disk (which is the second failure group) to the equation. So my point is if you are going to compare SRT's price 'savings', then you need to remember what you are sacrificing in total capability / reliability, or add the cost of the third disk to make it an apples to apples comparison to option (2) the non-SRT SSD + big platter setup.

 

[Dadofamunky]

Sad to see you have a P67 motherboard. Here's why.

-1

Spare us all.

And on the actual POST, OCing on the stock cooler is a really bad idea.

 

[Diogenes2]

....

That is apples to oranges; why? Because with option (1), SRT, you have a single disk, in essence, so it is a single failure group. By using the SSD as a cache, it is no longer a seperate copy of data, and I can't backup my data without adding a third disk (which is the second failure group) to the equation. So my point is if you are going to compare SRT's price 'savings', then you need to remember what you are sacrificing in total capability / reliability, or add the cost of the third disk to make it an apples to apples comparison to option (2) the non-SRT SSD + big platter setup.

You still haven't clarified how you eliminate a third disk in option 2 without sacrificing reliability .

What do you use to back-up your big platter ?

 

[BonzaiDuck]

-1

Spare us all.

And on the actual POST, OCing on the stock cooler is a really bad idea.

Your displeasure is noted. The OP was focused on using his acquired equipment to OC (over-enthusiastically to 5.0Ghz), and I went on a rant about chipsets.

If I OC on a stock (Intel) cooler, I only see how far I can go with the "stock" or "normal" voltage, so it's a minor tweak. I've done that with three machines -- all alive and kicking. For instance, an E6700 Wolfdale bumped up from 3.2 to 3.6 with the same voltage setting and stock cooler. A 12% over-clock with a stock cooler.

Back to JoltCola's argument. There are myriad options for backup, and "mainstream" users mostly buy computers with single hard disks (or an SSD if they want to pay for it). But compared to my four-HDD RAID5 array, with the ISRT-cached HDD you can backup with a hot-swap drive, and for the most part, power consumption is equivalent to maybe one drive, because you don't need to run the backup disk night and day. With RAID1, you can have ISRT with the redundancy, and you've now doubled the HDD power-consumption. But you have performance closer to that of SSD.

If we wanted to pay for it, we could buy two or more SSDs for either RAID0 or RAID5, forego any "caching," and with RAID0 get read-speeds closer to 800 MB/s (with the right choice of SSDs). With ISRT, you'd get at most 400 Mb/s. But you'd never notice the difference. Either way, we want a backup solution. Nobody disputes that.

I suppose -- or agree -- that the price per gigabyte needs to include the cost of a backup solution. Not a really big difference though, if drives can be had for $60 each.

However we want to play tug-of-war over this, I'm satisfied that my data is as safe now as it was under the RAID5 machine. The ISRT solution is maybe twice the speed of the RAID5 in reads and writes. There are no hourglass moments in Windows. I spent maybe $95 on the caching SSD.

It works for me. . . . If someone didn't want to worry about power-consumption or equipment cost (per GB), go for it! RAID0 two of those SSDs and add an HDD into the mix! Throw in a second HDD and RAID1 your backup!! But with the performance I'm getting out of my Pyro and HDD combination, I've got better ways to spend the money.

The new machine uses fully 100W less power than the old one -- at idle -- I measured the difference. It seems lightspeeds faster. So for me -- "mission accomplished."