Core 2 Overclock: No Appreciable Difference!

[Ladyhawk]

Here are a couple of e-mails I wrote to a friend during my overclocking experience. Perhaps some of you would like to weigh in?

I researched everything pretty well before I started overclocking. I was hoping to get a 100% overclock, but even after setting the core voltage to the maximum allowed for this chip (1.5v), I couldn’t get a POST, so I went with 3.3 GHz. Some sites said I could pump up the vcore to above 1.5, but I decided not to. If you have to push your chip, the overclock will probably be unstable and shorten the life of the processor.

The original specs on the chip (Core 2 Duo E6600):

FSB: 266 (quad pumped)
CPU Frequency: 2.4 GHz
Multiplier: 9

The multiplier on this particular chip is locked, so the only way to overclock is to raise the front side bus. I tried 300, 333, 366 and 400. I could put a wacky number in there, but I’d prefer to run my RAM in synchronous mode. 2 X 366 = 732, so I am running the RAM at DDR2-732. Because the RAM is rated at DDR2-800, I can probably tighten the timings without raising the voltage. I’ll have to run memory tests to make sure the hardware can handle it.

So, I’ve decided to stick with a front side bus of 366. 366 X 9 = 3.3 GHz. I have the vcore at 1.46 and I’m running a CPU torture test. If it makes it through the night with no errors, I can move on to tightening up the RAM timings and restoring the energy-saving CPU features (temporarily disabled) in BIOS. If not, I’ll raise the vcore slightly and torture test again. My CPU cores are running at a stable 59 – 60 degrees Celsius under load, which is about 5 degrees below the high temp recommended by overclockers. I’ll check them again when the ambient temperatures in the apartment are warmer.

Before overclocking I ran some benchmarks on my games. It will be interesting to see if there’s real-world improvement. If not, what’s the point of overclocking?

My torture test ran all night at 1.4750 vcore with no errors, so I ran the same benchmarks I did before and there was no appreciable difference between 2.4 GHz and 3.3 GHz! Here are some possible reasons:

*I hadn’t tightened up the RAM timings. This is unlikely because the DDR2 RAM speed seems not to matter with Core 2 Duo systems.

*The higher-than-usual voltage created some kind of CPU instability. This is unlikely because the torture test ran for 11 hours with no errors.

*Background apps killed the overclock. This is unlikely because I ran the same background apps in both sets of tests.

*The bottleneck in my system is the graphics card. I think this is the most likely reason.

I’m going to put everything back to normal and if/when I get another graphics card, I’ll try again. I may ask around and see what others think. Sometimes an overclock does well on standard benchmarks, but not so well on real-world applications. I used Fraps to record frames-per-second in actual play while using the background apps I will be employing on a day-to-day basis. Crysis, the game hardest on a gaming rig, was as slow at 3.3 GHz as it was at 2.4 GHz: barely playable at 1024 X 768 with everything on “medium.”

This was educational, if nothing else. I think I’ll ask the folks at Hard|Forum what they think.

I used Fraps to benchmark Crysis, FEAR, Unreal Tournament 3 and Team Fortress 2. Where applicable I ran the benchmark from the same save game for four minutes. I ran each benchmark at least three (sometimes four) times. Except for one benchmark (which I threw out), everything was within 2-3 fps, INCLUDING the overclocked benchmarks.

So, what do you think?

System Specs:

ASUS P5W DH Deluxe
E6600 Core 2 Duo with Arctic Freezer 7 Pro HSF
2 X 1 GB G.Skill RAM DDR2-800 (5-5-5-15) (Dual Channel, slots 1 and 3)
2 X 1 GB A Data RAM DDR2-800 (5-5-5-18) (Dual Channel, slots 2 and 4)
Sapphire X1950 XT with 512 GB RAM (yes, this is the one I had to wait six weeks for during that hellish RMA, but it works)
Windows XP Professional (don't want Vista yet)

I used Prime 95 (dual core version) for the torture test and Core Temp to check the core temperatures. I used Fraps to test real-world frame rates with all the background apps I need running, most of which are necessary to run my gaming hardware and software: GWS Fragpedal II software, Logitech G7 software, nostromo speedpad n52 software, Pinnacle profiling software, Game Commander 2, ATi Tray Tools and Power Mixer. I also left my antivirus and backup software running (AVG and Mozy) because I generally don't disable them to play games.

So, is my video card the bottleneck?

 

[hstuehmeyer2000]

yes, a new video card would make your games go faster, your x1950 is only so fast

 

[NKDietrich]

Even with a quad SLI setup you wouldn't see much difference going from a 2.4GHz to 3.3GHz Core 2. Even with the best GPUs, you're GPU limited at high resolution 99% of the time, and the 1% are games that are already running over 100fps.

 

[kush]

Where you would see the easily discernable differences would be in video/audio encoding (MT aware apps of course), and perhaps the best reason to overclock (responsibly) to help out your F@H numbers! << hint hint (check the distributed computing forum if you're interested in joining the [H]orde, assuming you aren't already).

Anyway, another vote on the video card. You didn't mention what res you're running at, nor any eye candy levels; but I'd be willing to be it's the card holding you back regardless. For the older games, your current card is probably fine within reason, but w/ Crysis, and likely some of the bigger upcoming titles, you'll be best served upgrading. Just north of $200 (AR) will get you a G92 GTS - GT's are a bit less at ~$180 or so. ATI, while the 38x0's are pretty nice for their price points (or at least were before the bottom dropped on G92's), they still suffer from lackluster performance that the previous generation did (esp w/ AA/AF, and at higher resolutions), and R700 isn't due until at least July (I say later even).

 

[TheRapture]

My E6750 at stock speeds (2.66ghz) shows no noticeable gameplay difference in Crysis...TF2....CoD4....etc., than it does clocked at 3.2ghz.....I game at 1920x1080.

But I am an [H] member, and leaving things like cpu and gpu at stock speeds is just not right

 

[MaxTheLimit]

Another vote for the GPU being the bottleneck. I might also mention that this is another reason why I think synthetic benchmarks are beyond their usefulness. You will see big increases in point from a CPU overclock, but marginal results in gaming.

Encoding is why I overclock, along with a couple of other tasks that are highly CPU dependent. If you game at fairly low resolutions to the point the GPU isn't so much the bottleneck you may see more results...but again that kind of defeats the point.

 

[aethelwulf]

You were hoping to get a 100% OC? I think 4.8Ghz on that processor is asking a bit much. Also, the FSB is 266 (NOT quad pumped).

Your GPU is probably holding you back, but a lot of games really aren't that CPU dependant these days anyway, and you're going to see diminishing returns OCing most modern processors in games. I could get by with my Quad at stock clocks, but what's the fun in that?

 

[skadebo]

Have you tested lower voltages? Also, use OCCT for stress testing, it'll catch error quicker.

 

[Ladyhawk]

You were hoping to get a 100% OC?

That isn't what I meant, but it's 1:40 a.m. and my brain isn't telling me how to re-phrase it. Come on, brain, or I'll stab you with a Q-tip.

I wanted to go from 2.4 to 3.4, so I wanted to increase it by 1 GHz, not by 100%. My bad.

I think 4.8Ghz on that processor is asking a bit much. Also, the FSB is 266 (NOT quad pumped).

My bad. I thought about double-checking that and didn't.

Y

our GPU is probably holding you back, but a lot of games really aren't that CPU dependant these days anyway, and you're going to see diminishing returns OCing most modern processors in games. I could get by with my Quad at stock clocks, but what's the fun in that?

Thanks...I shouldn't be trying to respond. It's frickin' 2 in the morning and I can't do simple math. It's like a 29-30% o/c, right? Just call me braindead zombie Ladyhawk.

 

[Ladyhawk]

Thanks for the input. I guess my next upgrade is going to be a vid card, but it will probably be awhile.

Have a good one!!

 

[kaiweiler]

I'm sure there is an appreciable difference from stock to overclocked, just not in your gaming benchmarks.
Try a benchmark application that is not so GPU dependant and you will see the difference.
PCPitstop has a nice benchmarking tool online, check it out.

 

[KranZ]

There's always the pride knowing that you're getting more Hz out of the CPU than you paid for!

 

[Ziggy!talon]

the overclock will still help alot in things like video encoding....

i never game so i just run a PCI-E x16 7600GT 256MB, but i run a presler dual core at 3.6GHz, while the graphics card is a huge bottleneck and i cant play the newest games, i still get great video encoding speeds...especially with the 3 drives i have in RAID:0

 

[psyshack]

XP won't use all that ram so find your best out of the 4 sticks and use just 2 gig. Drop the cpu multi to 8 and then ramp the ram up. look into NB voltage and get the ram voltage up a tad if needed.

There is more than one way to skin a cat.

 

[RAMBO29]

If you are happy with what you have,don't bother.

But for CPU intensive things you will definitely see difference.