Overclocking: Dual- vs. Quad-Core CPUs

[caamsa]

Whos the moron who got the prices for them on this article?

On page 2 they show prices for the different chips and all their prices suck. Are they buying them at BestBuy or what?

When in doubt, use NewEgg prices. They are a little more indicative of prices people who care to read this article will actually be paying.

I would have also loved to see them put the E6850 against the Q6600 because currently those 2 chips are going for the EXACT same prices on NewEgg. I think that would be a better shootout since the cost is the same and it begs the age old question. For the same price is it better to have a faster dual core or a slower quad core.

3.00 x 2 = 6.00 Ghz total power
2.40 x 4 = 9.60 Ghz total power

You would think the choice would be obvious, but in alot of cases your only using 1 Core anyway.

Here you go dude.....just follow the link and your wish is my command..........

http://www.xbitlabs.com/articles/cpu/display/core2quad-q6600.html

 

[SpinachEater]

Who has a Zalman fan cooler? Are they pretty much top of the line or do they get more efficient than that? The design looks pretty b@d@ass none the less.

 

[cnumartyr]

For a quad.. I consider it Thermalright Ultra 120X or nothing for OCing. Pair it with anything over a 60 CFM fan and it's pretty hard to beat.

I use a 110 CFM fan lol.

 

[SpinachEater]

Hey cool...no pun intended...I found a review of them. About the same price as the Zalman too on newegg.

http://www.xbitlabs.com/articles/coolers/display/thermalright-ultra120ex.html

 

[cnumartyr]

Cheaper from SVC and you could also buy them prelapped.

 

[ArcticFlipper]

There is some info that you need to know of before you all run out and go buy new Q6600 CPU's.

We've recently just completed a very similar test between the Q6600 and the E6850, on various mobo's.

Our findings might be of some help. The test bed was setup with the same RAM, Corsair Dominator 1066MHz, Corsair Dominator 800MHz, Kingston HyperX SLI 1066Mhz, Kingston HyperX 800Mhz and finally with Corsair Dominator 1250MHz.

Mobos used was the Asus Striker Extreme, Asus P5N32 -E SLI, Asus P5N just get a feel for what the 680I and 650 Nvidia capabilities are.

And for the Intel chipset options we used Asus Blitz Extreme, Asus P5K, Asus P5B Premium

The interresting and most important part we found is that the intel chipset will give you the best performance on both CPU's when compared against the Nvidia boards.

BUT and this is a big BUT. The intel chipsets we used would not boot with the Corsair Dominator 1250MHz ram in SLI mode.

The Nvida chipsets gave us the best Graphics and RAM performance.

The Q6600 would only clock up to around 2.8GHz on the 680i chipset, there after with a lot of tweaking we did manage to ge it over 3.0 GHz mark, but the board did not last long at all.

The Q6600 on the Intel chipset we managed to clock stable up to 3.44 with the 965 chipset without any problems at all. The P35 chipset gave us 3.80 for the same CPU.

To make a long story short, what we found was that the best setups so far for us was the Nvidia chipset with a E6850 CPU, and the Q6600 with a Blitz extreme board.

There are such small differences in performance when testing the entire package, not just CPU or RAM or GFX. But the major downfall for us was the fact that no P35 board at this stage supports Nvidia SLI, only ATI crossfire.

I surely didn't spend serious money on SLI cards not to be able to use them.
Oh and let's not forget about the Corsair Dominator 1250MHz RAM we have here.

In the end my personal choice is the Asus P5N32-e SLI board with an E6850 cpu with the Dominator 1250Mhz RAM and 2 WD Raptor-X HDD's.
I know this is allready way out the pricelist of the same test done by Tom's Hardware, but if you are here looking at this you most prob will be spending much more money than what they did on this test.

From all the test we ran there was some very interresting findings, especially when it comes to RAM.... Sometimes it's worth your while to invest in value RAM rather than the more expensive Dominator or HyperX RAM. The HyperX was the worst RAM from the Bunch we tested, even slower and with less OC capabilities than their Value RAM options.

 

[night_wolf_in]

what suprises me that. i overclocked my Q6600 to 3Ghz with Voltage of 1.26875v, While the article they used 1.3175 (or 1.31070) refer to page http://www.tomshardware.com/2007/11/08/dual_vs_quad/page15.html n my overclock was stable, n im using p35-DS3L. (tested with prime95 for more than 8 hrs.)

 

[dopbaggins]

Great write-up.

I realize the reviewer used an Engineering Sample Q6600, and stated their 3.3ghz O/C required a 1.43 v-core to pass Prime95.

I don't know if the it was their engineering sample Q6600 or the budget (yet nice) motherboard that required the extra boost to the V-core for stability, However I would like to add that my Q6600 at 3.4 ghz (100mhz faster) requires only 1.38 v-core to reach 24hr prime stability.

I am using an Asus P5K deluxe motherboard. (200$ vs 100$ ?)

maybe the extra v-core necessary to get their O/C stable was a combination of the E.S. Q6600 AND their more entry-level OC motherboard.

input ?

 

[smithgotsurf]

Great write-up.

I realize the reviewer used an Engineering Sample Q6600, and stated their 3.3ghz O/C required a 1.43 v-core to pass Prime95.

I don't know if the it was their engineering sample Q6600 or the budget (yet nice) motherboard that required the extra boost to the V-core for stability, However I would like to add that my Q6600 at 3.4 ghz (100mhz faster) requires only 1.38 v-core to reach 24hr prime stability.

I am using an Asus P5K deluxe motherboard. (200$ vs 100$ ?)

maybe the extra v-core necessary to get their O/C stable was a combination of the E.S. Q6600 AND their more entry-level OC motherboard.

input ?

dopbaggins,

at work, we have three computers using the msi p35 neo2 and two of those motherboards are using the xeon 3220 (same thing as the Q6600) and i have those machines overclocked to 3.6 ghz and the v-core is around 1.43.

i'm extremely pleased with these mb. i think they are a great bang for the buck.

 

[gomerpile]

Ok, what is it with Tom's lately. On page 21 they review the test setup which is clearly different then the setup described in the rest of the article.

AMD Platform AM2
(Nvidia Nforce 5) Asus M2N32-SLI Deluxe, Rev.1.03G
Nvidia nForce5, BIOS: 1001 (03/13/2007)
Intel Platform S775
(Intel P35) Gigabyte P35C-DS3R, Rev. 1.0
Intel P35, BIOS: F2o (05/11/2007)
Intel Platform S775
(Intel X38) Gigabyte GA-X38-DQ6, Rev. 1.0
Intel X38, BIOS: F4 (09/19/2007)
Memory 2x 1GB A-Data DDR2-1066+ Vitesta Extreme Edition
DVD-ROM Samsung SH-D163A , SATA150
Graphics Card Foxconn Nvidia GeForce 8800 GTX
GPU: 575 MHz
Shader: 1350 MHz
Video Memory: 786 MB DDR4 (900 MHz, 384 Bit)
Sound Card Creative Labs Sound Blaster X-Fi XtremeGamer
Power Supply Zalman, ATX 2.01, 510 Watt

^^^ What is that garbage? Its from page 21 of this article and it has in it a board for AMD.

off the topic but oohno's not a cat again, where's the hound

 

[dark41]

I was very disappointed with this article, the results, their choice of components, the recommendations, and conclusions. But its obvious to me that because the components weren't the best options, the conclusion is flawed.

I've got 3 E6750s running F@H 24/7 at 3.8GHz with 1.425v and its been 100F degrees in this room all week (Gotta love Australia summers). The systems have been running F@H non-stop for several weeks now, no reboots, no disruptions. 2 of them have been running this way for roughly 6 months, and we've only had the 3rd for a month.

With 1.475v all 3 CPUs will run 4.0GHz stable, and up to 4.15GHz not so stable. I couldn't get past 3.6GHz on our Go stepping Q6600 unstable, and 3.4GHz stable. So we sold it and bought another E6750. I've actually built and tested about 100 E6750s now and every one seems to run very similarly. I also built a system for a customer with a Q6600 after we sold ours that stress tested for 24 hrs with no problems at 3.6GHz. If nothing else, this has proven to me that E6750s are more consistant than Q6600s for OC results. I suppose with 4 cores the odds are twice as high that one core will not overclock as well. Every system we've built has been fired up at 3.6GHz with the stock Intel fan at 1.425v and stress tested at 100% for 24 hours for stability prior to being sold.

We have 3 different CPU fans (Gigabyte G-Power Pro, G-Power Pro BL, and Thermaltake BT with Smart Fan II) and all seem to do a better job than THG's recommended Zalman for the same price since we have no problem achieving 4.0GHz stably on each of our systems.

We use the Gigabyte P35-DS3P boards. The DS3P is listed on newegg for the $145 (randomly selected price limit) that the article chose and has more options than the DS3 or MSI Neo. I can't speak to the overclocking benefits over the DS3 and MSI as we've never used them. But the DS3P will overclock as well as my P35-DQ6, up to at least 515FSB with the multiplier on 8x. Since I prefer to run our memory at 1:1 I haven't tried to see how far they'll actually go.

We use Crucial Ballistix PC2-800 RAM. It's cheaper than the Geil RAM on newegg and performs quite a bit better. All 3 2x1g sets of Ballistix run at 4-4-4-12 timings up to 960MHz and 2.1v and are warrantied up to 2.2v. Relaxing the timings to 5-5-5-15 and staying at 2.1v all of these sets will do 1200MHz+ stably. They range from 1210MHz to 1235MHz.

So the systems I've listed stay within the article's price range and perform substantially better than theirs.

http://www.ultramaxcc.com.au/images/des.jpg

I guess time will tell if the quad core is utilized by more apps and games, but currently nothing we run with the exception of encoding/decoding (which we do only occasionally) was faster on the quad than the C2D. The quad is not only more expensive, but it costs more to run, runs hotter, and is slower overall for everything we do, even with AVG scanning in the backround. Our F@H WUs finish faster on the C2D too. I still think the quad is a waste of money unless the bulk of what you do will utilize the quads and I've seen no evidence that multi-tasking is faster on a quad.

 

[bdubslawman]

Wow, what a great article. And very timely for me personally.
The fact that my current CPU, the E4300, was always at the bottom of that list, sure makes me want to take the plunge and go to one of their older brothers. Well, and the fact that my E4300 seems to run really hot does not help.
Now, what do I do? the e6750 $199, the q6600 $273, in real canadian dollars ;-)
Add to that another $62 over 1 year in hydro, and now we are talking big diff is $$$.
So I am thinking the e6750 for now, and then maybe 1 year from now, look at what is around in terms of CPUs and the apps that run on them.
What do you think?

How hot does your E4300 run? Is what stepping is it? I'd try a re-seating or new heatsink if you trust the settings you are getting.

I own a E6750 & a E4400 runs 50-51 degrees C on benchmark tests with the rebate Cooler Master that has been nearly free or close to it After Rebate for months. If you can deal with your system or find a cooler solution to running it (I assume you must be OC'ing it right)....then why not wait for a Penryn? Dual or Quad?

Penryn CPU's:
These are tray prices (per 1000) yes the price will initially sell for more as it did with the 333Mhz processors E6#50's release but the prices settled to within $10's of the tray price on the E6750 ($183) within 2 1/2 weeks as Microcenter and a few other stores began selling for prices @ $193.
Dual Core:
---E8500 - 3.16Ghz - 9.5 multiplier - 333 Base system clock - 6MB cache - $266 - Jan 08
---E8400 - 3.00Ghz - 9.0 multiplier - 333 Base system clock - 6MB cache - $183 - Jan 08
---E8300 - 2.83Ghz - 8.5 multiplier - 333 Base system clock - 6MB cache - $### - Jan 08
---E8200 - 2.66Ghz - 8.0 multiplier - 333 Base system clock - 6MB cache - $163 - Jan 08
Quad Core:
QX9650 - 3.00Ghz -unlockedmulti - 333 Base system clock - 12MB cache - $999 - Nov 12, 2007
---Q9550 - 2.83Ghz - 8.5 multiplier - 333 Base system clock - 12MB cache - $530 - Jan 08
---Q9450 - 2.66Ghz - 8.0 multiplier - 333 Base system clock - 12MB cache - $316 - Jan 08
---Q9330 - 2.50Ghz - 7.5 multiplier - 333 Base system clock --- 6MB cache - $266 - Jan 08

I don't Trust Tom's Hardwares information that I have quoted below:

You may ask "And what about Penryn?" Well, Intel only just released the new processor family based on a 45-nm production process. For now, only the flagship model costing $1,464 is available. Smaller and less-expensive CPUs are still a ways off. According to Intel, it may be March 2008 before we see smaller Penryn-based CPUs hit store shelves. Thus, the Penryn family is not really much of an option for the cost-conscious buyer for the time being. ....

I'd like to know where this information is actually coming from. Insider or simply Intel spokesperson...because Intel of course doesn't want people to stop buying the current lines...waiting in anticipation for Penryn. That would impact 4th quarter sales far too much.
Sites that have been SPOT on for the last three Intel releases or price drops give the above dates & tray prices (HKPEC, Techarp)... Jan 8th is a Hell of a lot sooner than "MARCH" even if the prices don't drop until sometime in February the prices are LESS than current E6#50 line while you get the benefits of SSE4, less power, therefore less heat, bigger cache (although...?)


One last note that waiting for the Penryn Quad cores if you desire a quad core will offer a step towards better efficiency. From an environmental standpoint there is little downside. Better on power & faster clock, more efficient code than what is currently available in a Q6600. Other than having to wait a few months... Zero Downside...unless your MOBO is not capable of 333 FSB.

Oh yeah, I disagree with the conclusion once again of TH. IF you heavily need to use Photoshop or transcoding video software constantly then Q6600 even now is the way to go if you can't wait for the 45nm's. TH intimating that the games due out over X-mas 2007 season will validate the Quad core purchase for those who are looking for things to do with the extra quads is misleading. At this point, games will not be written that can only be well played/enjoyed by the only the small few (but growing) legion of gamers who now occupy the quad core market! Game developers sell to the greatest market. That is as ridiculous a notion as the introduction of DX10 would bring about the rapid demise of DX 9 in cards or games.
Reminder beyond all its other many benefits, when transcoding programs were written to take advantage of SSE4 instructions in the new Penryn's the increase in one benchmark (using dual core was 81%)...that's well worth the wait for dual or quad core if your programs can/will be written using that kind of code.