Guide: Overclocking AMD And Intel CPUs On A Budget

[Crashman]

[citation][nom]dirtmountain[/nom]Gathering components, overclocking the CPUs, running the tests, writing the article, dealing with editors and staff all take time. It's not like he did this all over the course of a week or got the components yesterday.[/citation]

Actually, it went something like this:

Middle Management "Another editor has committed you and I to have this article published in ten days"

Editor "And how is that our problem"

Middle Management "Let me talk to upper management."

A few days and a weekend later,

Middle Management "It's our problem because he made comitments to outside companies."

Editor "It can't happen in 5 days"

Middle Management "I'll get the parts, you make it happen as quickly as possible"

After THAT, it took a week

 

[egowhip69]

[citation][nom]Crashman[/nom]Actually, it went something like this:Middle Management "Another editor has committed you and I to have this article published in ten days"Editor "And how is that our problem"Middle Management "Let me talk to upper management."A few days and a weekend later,Middle Management "It's our problem because he made comitments to outside companies."Editor "It can't happen in 5 days"Middle Management "I'll get the parts, you make it happen as quickly as possible"After THAT, it took a week[/citation]

HA! That sounds exactly like the process I go through every time there is a new "project" here at work.

 

[lowguppy]

I've noticed in recent GPU and CPU shootouts that most benched games were CPU limited when using high end graphics solutions. As a budget system builder, I'm interested to see what the maximum useful graphics solution is for more budget oriented CPU options like these.

 

[Ramar]

I still prefer buying a good ol' E8400. I know it's a hundred dollar price premium, but surely the extra [4?] MB of cache help quite a bit. I've had mine running at 3.6 on slightly lower than stock voltages for almost a year now. Rarely gets above 45C. I'd gladly test its performance in the upper ranges, but I'm limited by my crappy DDR2-800. =[

 

[pauldh]

[citation][nom]4ghz_or_gtfo[/nom]It's nice to see a Tom's article that doesn't intentionally low-ball their AMD OC, I'm sick of seeing 3.6ghz PhenomII OCs when we all know they can do better. [/citation]
You got 6 positive votes, so others agree they like to see these high OC's. Can't fault that sentiment as we like seeing it too. Just keep in mind, results will vary chip to chip. Tips straight from AMD were to expect 3.65-3.9GHz for a PII on air cooling, all depending on the voltage scaling of that individual chip. So please be cautious on your assumptions that a site is intentionally (or even unintentionally) limiting an OC because you don't see 3.8GHz.

Example: One of the PII’s I received topped out below 3.7GHz and did not like additional voltage despite temps being under 50C. Popping in a different sample added 150MHz. To some extent, it's luck of the draw. Same goes for Intel.

Also remember not all user OC claims floating around out there are to be trusted. I’m sure many are suicides, or at least not completely stable. Nobody wants a sub-par OC, but it happens. Tom's (or any other site worth reading) isn't going to use/publish an OC that doesn't first pass a fair amount of stability testing.

 

[darkpower45]

when i was overclocking my Q8200 i was able to reach 3.5Ghz with a 2000Mhz FSB. i dont know why they couldnt get higher clocks from the CPU. maybe its because i got my Q8200 right when it came out, and now they Q8200s went down hill? who knows.

 

[Crashman]

[citation][nom]darkpower45[/nom]when i was overclocking my Q8200 i was able to reach 3.5Ghz with a 2000Mhz FSB. i dont know why they couldnt get higher clocks from the CPU. maybe its because i got my Q8200 right when it came out, and now they Q8200s went down hill? who knows.[/citation]

500MHz FSB is almost impossible with ANY Core 2 Quad, even a high-end Q9650 with a dropped multiplier. Duo's can do it, but for the most part quads can't.

 

[Guest]

@pauldh: The general sentiment is that they almost always(todays 8200 being the first exception I can recall) manage to get about 4ghz out of whatever Intel CPU they're using, coupled with the fact that I own a 940BE and can overclock it to 3.4 on the stock heatsink at stock voltage, 24/7 stable. Surely additional voltage and a real enthusiast heatsink are worth more than 200mhz???? Hell, since AMD came out with ACC, the original Phenom can hit 3.5ghz, to say that switching to 45nm + all the other process/architecture improvements are only worth 100mhz is naive at best.

 

[Guest]

and something else, pauldh: What makes it so incredibly suspicious is when a Tom's article manages a whopping 3.54ghz OC out of a Phenom II, which then loses half of the benchmarks by a very small margin, whereas with a standard high-end OC of 3.8(normal max OC for PhII, Core2 and i7, anything higher is exceptional) would've shifted many of the close victories to AMD. To anybody who is not completely gullible, it really makes it look like they sat down and figured out how to make it lose...

 

[skora]

You got 6 positive votes, so others agree they like to see these high OC's. Can't fault that sentiment as we like seeing it too. Just keep in mind, results will vary chip to chip. Tips straight from AMD were to expect 3.65-3.9GHz for a PII on air cooling, all depending on the voltage scaling of that individual chip. So please be cautious on your assumptions that a site is intentionally (or even unintentionally) limiting an OC because you don't see 3.8GHz.

Example: One of the PII’s I received topped out below 3.7GHz and did not like additional voltage despite temps being under 50C. Popping in a different sample added 150MHz. To some extent, it's luck of the draw. Same goes for Intel.

Also remember not all user OC claims floating around out there are to be trusted. I’m sure many are suicides, or at least not completely stable. Nobody wants a sub-par OC, but it happens. Tom's (or any other site worth reading) isn't going to use/publish an OC that doesn't first pass a fair amount of stability testing.

I wish more people would accept this. I get a little miff'd when I hear, "The E5200 can do 4.0ghz easy." That kind of rhetoric is what makes the web dangerous for FACT finding. No, I'd say the 4.0 ghz is the exception and most are in the 3.6-3.8 range. Getting to 4.0 is not easy. I get laughed at for only getting to 3.6, but it must be my penance for not living right.

For every person gloating they hit a high number, there has to be many more that aren't interested in that high of a clock that don't speak up about it. Expand the cpu speeds. Less than 1% are above 3.7 ghz. http://store.steampowered.com/hwsurvey/

 

[Crashman]

[citation][nom]5ghz_or_gtfo[/nom]Hell, since AMD came out with ACC, the original Phenom can hit 3.5ghz, to say that switching to 45nm + all the other process/architecture improvements are only worth 100mhz is naive at best.[/citation]

Only if you think it's naive to believe the truth. Personally, I think it's naive to belive all the BS that keeps showing up in these threads.

 

[noob2222]

But someone said they saw it, it must be true (WOW a pig just landed on my house).

 

[pauldh]

[citation][nom]6ghz_or_gtfo[/nom]and something else, pauldh: What makes it so incredibly suspicious is when a Tom's article manages a whopping 3.54ghz OC out of a Phenom II, which then loses half of the benchmarks by a very small margin, whereas with a standard high-end OC of 3.8(normal max OC for PhII, Core2 and i7, anything higher is exceptional) would've shifted many of the close victories to AMD. To anybody who is not completely gullible, it really makes it look like they sat down and figured out how to make it lose...[/citation]
I don't recall a PII maxing out at 3.54GHz? Do you have a link as I'd assume there was a reason for stopping there. Ex: Look at the last SBM and you see all Intel OC's were mega limited by cooling as small Micro ATX was the theme.

Are you suggesting that Phenom II's and Core 2 Duo's have the same max OC? Limiting an E8600 to 3.8GHz is like limiting a PII to 3.2-3.4GHz.

 

[Guest]

Hi Tomhardware. It's be nice to have reviews and methods come as fast. We are thankful of your site for the articles. I'd like to save you time and I'd like to share on my overclocking methods. It's the easiest to know the maximum and minimum voltages at load. It's also the quickest way to find out stability (no need for waiting for an hour or so. It could only take seconds, to 5 minutes to less than 30 to 30 minutes): http://www.tipidpc.com/viewtopic.php?tid=189355

 

[Guest]

Crashman: WTF are you on about? Did you not hit a respectable OC of 3.8something here?

Pauldh: I don't keep bookmarks of these things for posterity, and searching google with the query "phenom II 940 3.6 ghz site:tomshardware.com" is a complete clusterfuck, but I am refering specifically to the initial review when Phenom II came out and was pitted against the i7 920.

 

[pauldh]

[citation][nom]5ghz_or_gtfo[/nom]@pauldh: The general sentiment is that they almost always(todays 8200 being the first exception I can recall) manage to get about 4ghz out of whatever Intel CPU they're using, coupled with the fact that I own a 940BE and can overclock it to 3.4 on the stock heatsink at stock voltage, 24/7 stable. Surely additional voltage and a real enthusiast heatsink are worth more than 200mhz???? Hell, since AMD came out with ACC, the original Phenom can hit 3.5ghz, to say that switching to 45nm + all the other process/architecture improvements are only worth 100mhz is naive at best.[/citation]
Most Intel processors we would chose to use in a build can approach 4.0GHz. But, again check the last SBM as none came close to that.

Your 940BE is acting the same as my sample and could likely approach/hit 3.8GHz stability with decent air cooling. But, that doesn't mean another sample used by another editor won't require a voltage bump to reach 3.4GHz and would also top out far less than that. It varies chip to chip.

Not sure the point here, but very few 65nm Phenom's can reach 3.5GHz. Even with ACC, most still top out below 3.4GHz.

 

[Guest]

another thing, Crashman: What's this bit here?:

"Middle Management "It's our problem because he made comitments to outside companies."

Why would this article have commitments to outside companies? What outside companies? Kind of pisses on the notion of independent journalism and hardware reviews, the fact that outside companies will commit you to such things is suspect. Would you care to elaborate on what you meant?

 

[pauldh]

[citation][nom]skora[/nom]I wish more people would accept this. I get a little miff'd when I hear, "The E5200 can do 4.0ghz easy." That kind of rhetoric is what makes the web dangerous for FACT finding. No, I'd say the 4.0 ghz is the exception and most are in the 3.6-3.8 range. Getting to 4.0 is not easy. I get laughed at for only getting to 3.6, but it must be my penance for not living right.For every person gloating they hit a high number, there has to be many more that aren't interested in that high of a clock that don't speak up about it. Expand the cpu speeds. Less than 1% are above 3.7 ghz. http://store.steampowered.com/hwsurvey/[/citation]
Actually I don't agree with you here. At 3.6GHz I doubt the E5200 is the wall you are hitting. What mobo are you using? What voltage have you tried? E5200's need more voltage than E8x00's. Often it's the mobo being the cause of such limit though. Or it can be a bootstrap issue. Of the numerous E5200's I have OC'ed now, the worst one still had total stability at 3.9GHz, and the best one did 4.183Ghz at that same voltage.

 

[pauldh]

[citation][nom]7ghz_or_stfu[/nom]Crashman: WTF are you on about? Did you not hit a respectable OC of 3.8something here?Pauldh: I don't keep bookmarks of these things for posterity, and searching google with the query "phenom II 940 3.6 ghz site:tomshardware.com" is a complete clusterfuck, but I am refering specifically to the initial review when Phenom II came out and was pitted against the i7 920.[/citation]
Thomas reached two very respectable OC's on these PII's. Above average for sure. Ask him what he has hit with C2D though.

So you were singling out one specific chip that was OC'ed to 3.64 GHz and ignoring at least 4-5 other Tom's articles where different PII's went far higher (roughly 3.8GHz or beyond). Buy a handful of the same model and you yourself will see a range of max OC's reached with them.

 

[Crashman]

[citation][nom]8ghz_or_lmao[/nom]another thing, Crashman: What's this bit here?:"Middle Management "It's our problem because he made comitments to outside companies."Why would this article have commitments to outside companies? What outside companies? Kind of pisses on the notion of independent journalism and hardware reviews, the fact that outside companies will commit you to such things is suspect. Would you care to elaborate on what you meant?[/citation]

The best time to advertise a large case, cooler, or high speed RAM is when an overclocking guide is on the front page. That's just one example of targetted advertising. It doesn't affect the outcome of the article (winner/losers, etc), but can affect when the article comes out (publishing date) by forcing editors to stand behind their commitment date.

 

[Guest]

i think this is a good start... i would like to see this with more cpu's (like one every other week or something) and a chart that puts all the results head to head.... that way people in the market can see more than just four cpu's... im building right now and i know i have far more than 4 options. next i would like to see from the AMD side the 720x3 or the 810x4 because those are the true value parts

 

[daeros]

Wow. I can't believe how poorly that Q8200 clocked. I have an OEM pull from a Gateway tower on a P45XE, and without getting crazy on voltages, I got to FSB1828 (457MHz) for a clockspeed of 3.2GHz. On stock volts I had no problem running at FSB1600 (400MHz) for a clock of 2.8GHz.

 

[fabolous]

[citation][nom]Crashman[/nom]500MHz FSB is almost impossible with ANY Core 2 Quad, even a high-end Q9650 with a dropped multiplier. Duo's can do it, but for the most part quads can't.[/citation]

I agree with this statement, for the most part. My roommate however must have gotten a stud, because his Q8400 has been running at 500MHz FSB 24/7 since last November. He does have some quality DDR2 1066 as well as a huge Tuniq Tower cooling it though.

Even so, it can't touch my Phenom II 940 @ 3.8

 

[donutey]

I'm not understanding why you've had so much trouble overclocking a Q8200. I'm using a P45 board as well, with a Sunbeamtech CCC ($25).

My highest stable overclock *without changing any voltages* is 3.2Ghz.
The highest I've ever validated as stable is 3.4Ghz:

http://www.overclock.net/5850797-post509.html

Going much further is difficult, since you're pushing near 500Mhz FSB, which is hard for a P45 with any CPU.

This is a normal Q8200, NOT a Q8200S. All the Q8200 overclocks I've seen all get above 3Ghz, I'm not sure what the issue is with yours.

 

[donutey]

[citation][nom]Crashman[/nom]500MHz FSB is almost impossible with ANY Core 2 Quad, even a high-end Q9650 with a dropped multiplier. Duo's can do it, but for the most part quads can't.[/citation]

I've done it on a normal Q8200, with 1.3vCore and 1.3v Northbridge.
The highest I've proven stable is 486FSB:

http://www.overclock.net/5850797-post509.html