What should i do?

hi everyone im new to the board here and need some advice. please note im new to overlocking and what i know comes from the guide here.
Must say its the best guide iv read.

I am building a new pc and wanted to go with the e4300 and oc it with water. my question is, should i go with ddr2 800 ram and if i have to underclock it to get a 1:1 ratio or could i lower the mutiplyer and go from there. Or should i just get some ddr2 667 ram. I need some advice on how to get best performance and this is a budget pc.

Thanks, Paranoia

 

also with the price drops and the e6320 should i just spend extra few bucks.

 

My e4300 overclocks very well with dirt cheap 667 ram on stock air cooling, so unless you plan on lowering the multiplier (Better off with a e6320 then anyway) stick with the cheaper stuff.

oh yeah, welcome to the board!

 

also will the 4mb cache do alot for me to were its worth the extra $ or should i just go with the e4300 and also what is your clock at. thanks again

edit: will the 4mb cache help in games is what i meen

 

The 4mb of cache will help you out performance wise all around. By how much just depends on the application.

My e4300 has been stable at 2.8Ghz on stock air cooling. I did drop the multiplier though since there was a FSB hole in my motherboard. Any FSB speeds between 1200 and 1333 wouldn't work. I had it stable at 3.3Ghz for a few hours, but it would frequently crash as my memory couldn't take the high clock speed (DDR2 800 would have come in handy here).

 

well i planed on gettin ddr2 800 and with the msi mobo im gettin i read about the hole that people are havin problems with. so considering ill have water do you think with the ram ill be gettin i can get in the 3Ghz area. and also since im a gamer should i just get the 4mb cache chip. Thanks again!

 

get the one with 4mb cache. also, if you're serious about this OC stuff, get memory rated higher than your mobo runs stock. that way you have some head room when you start to raise the FSB. sometimes you can get some very nice memory that OCs well, like I have (crucial ballistix) or some other brand that has a performance line of memory (crucial ballistix, corsair XMS2, corsiar dominator...there are lotsa good brands. patriot extreme is another)

 

E4300 is the true Allendale Core and some haven't been scaling as high as the early Conroe core CPU's were. It may be worth your while to go with a 6420 which will ultimately give you a 266FSB / 333FSB for 1066MHz / 1333MHz FSB Quad Pumped while the Allendale (4300) only gives you a 200FSB (800MHZ ~ 1000MHZ). Yes the 4MB cache which will help with all app's as was stated, however; Gaming performance is invariably dedpendent on your graphics card especially with PCIe. This is the reason cards now carry 1GB of the fastest memory made (next to CPU cache) and have GPU's with as many transistors as some CPUs.

In so far as your CPU that's important but perhaps nothing is more impartant then the platform which suppots it. Motherboards will ultimately determine how fast your chip will overclock by their FSB speed available in BIOS dictated by the North Bridge. Nvidia chipsets are known for their Gaming sterngths and the Nforce still uses the DASP feature which cats as an additional pre-fetch. The NVIDIA 680i or 590 chipseta are both excellent but if it's true FSb speed you want then the 965 chipset sclaes beyond 500MHz FSB, the 975 chipset to 400MHz FSB and the NVIDIA varies around 325FSB.

I have the E6400 8x266FSB = 2136MHZ and I can easily reach 8x400FSB = 3203MHZ on my P5W DH and I hope to get much higher on a Gigabyte P65 DS3 I have to review.

I cannot tell you how smart it is to go H20. As a part time reviewer I used to get the best heatsinks on the market for testing and none came close to the temps from a quality H20 system. I was introduced to H20 cooling about 3-years ago and never used air-cooling again except to test something. It's fun because there are so many different waterblocks and water-cooling systems to choose from. I absolutley recommend going with German waterblock/system maker's who are the best at ths. Some would say Australian or USA and while I love Danger Den and Switech I've tested a lot of kits and the European blocks by fare are the best. They don't use the bulky cumbersom 1/2" tubes instead they use 8mm/10mm tubes whuich are much easier to mount and save space. Thye also cool just as well even though they essentially use less water volume. That's what we call effciency.

Anyway setting that your kit up will make or brake you. Just like eveything else in PC world you never cool below ambeint room temp unless you go active with phase change or chilled H20 eaitherway it requires a compressor and coolant.

So the best you can do is to be sure your radiator is mounted externall to the case drrawing ambeint air in. many peopple starp the radiator inside the case which then places that device in the internal case temp zone. It's always going to be much hotter inside your case no matter how great the ventillation. If you decice to mount a radiator inside a case the best place would be at the front bottom drawing cool air into the cooling fins and then into the case. Having more exhaust fans then intake will create a vaccum and so long as your intake fans are filtered there's no reaosn to obsess over dust.

As for memory IMHO DDR800 is way too slow for the processor your getting. Forget about 1:1 ratio why would you want to run your CPU slower to run 1:1 with a slower memory? I would suggest getting high speed memory such as PC2-8500, PC2-9200 etc.

 

right on, another OC genius for the forums welcome to the site

I second that about H20 cooling, i love it and never want to go back lol

 

I'm confused , I'm running my 4300 at a 1500 FSB right now and had no problems running it at 1333 without any other changes made to the bios.

 

I never said 4300's weren't excellent overclocker's however, some people are now seeing the scalability dropping. Your chip actually represents the majority but I've heard many people only getting 325FSb speeds. Remember every chip you get which overclock's is a gift depending on how your overclocking which I explain below. 4300 is a great Purist overclocking chip but the performance difference between it and 4MB higher speed Conroe core leaves no room for doubt. Yes a 4300 "could" and do great and probably will but we're not gauranteed this. At least with the 6420 your guaranteed certain performance.

We cannot always expect such exceptionally high Overclock's. And when people don't get these "golden" chips they begin going nuts with Vcore. There's a lot of C2D overclocker's at Xtremesys although this was/is my home site (actually Major Geeks was the first site I joined about 7 1/2 yrs ago under another username Socrates something) I don't visit Xtremesys alot because they've lost the PC-Enthusiast hobbyist atttitude which was to help others not belittle others. Anyway here's a thread showing the diff between 4300 and 6400 http://www.xtremesystems.org/forums/showthread.php?t=130771

Once you increase Vcore beyond default you increase heat, with heat you increase the potential for gate current leakage and you ultimately begin irreversably damaging a percentgage of the millions of tiny switches which are at the core of your processor. These tiny electronic switches which make up every CPU are known as cmos transistors (once nmos). The nature of such a small device is that it's far too small to be a "mechanical" switch. So in order to work as fast as they do and remain as small as they are cmos switches are essentially capacitance switches. That is they always hold a small charge and they have two states: on/off (hence the term binary). These switches are very fragile and voltages have to be precise in order for them to work. This is why it's never as simple as 0.0V = off and 1.320V = on (Logic Low = off | Logic High = on).

With each die-shrink and smaller gate-width processes (45nm etc) each CPU increases it's number of transistors. And with this increase in cmos switches while the die-size shrinks, so too does their proximity to one another. The entire theory behind SOI technology is to prevent the unwanted results which occur with every shrinks such as current leakage and static current dischage into the surrounding silicon.

Every time we as Overclocker's increase Vcore we also increase the aforementioned and ultimately damage our CPU. The worst part is you won't see this damage immediately, it will be a much more subtle and therefore frustrating process where you have the occaisional error, the occaisional system hang, re-bott etc. There are many redundant systems (cmos switches) in a CPU but ultimately they die.

Lets disucss how they work. As voltage flow to the transitors they either see a Logic High (ON) or Logic Low (OFF). In a perfect world this process would look like a voltage sqaure wave. Since this occurs at a frequency the squarewave always looks like a sineuwave. Below we have the ideal situation creating on/off states;

http://www.modthebox.com/articles/overclock103/overclock1.jpg

In reality the transistors see the voltage wave below during operation (binary processing);

http://www.modthebox.com/articles/overclock103/overclock2.jpg

The rate at which this sinuwave flows is known as frequency. In CPU's there's no strictly defined point (voltage) at which a logic high (on) or logic low (off) occurs. This is why there are small circles around the top and bottom of each wave. It's at these points a gate will open or close,. If there were no tolerances your CPU would constantly "mis-fire" or recognize a Logic High for a Logic Low.

Now here's what happens when we overclock (increase frequency) on default voltage and the sinuwave falls short of the Logic High / Logic Low.

http://www.modthebox.com/articles/overclock103/overclock3.jpg

Now lets see what increasing Vcore does while leaving the frequency at default;

http://www.modthebox.com/articles/overclock103/overclock4.jpg

As you can see a new "artificial" Logic High has been created however all we've done is to increase heat and heat (voltage) is the bane of the CPU maker's exsitence. This is why so many chips have failed and why the Core Duo (which is nothing more then a PIII and Dothan combined) has been so successful. Lower voltages = lower heat = higher speeds = Dothan, Yonah and the Desktop version C2D) Shortter pippelines give Conroe a huigh advaatge over it's Presler predecessors which where blast furnaces in comparison to Conroe. Presler's execution pipelines were so deep you needed a Spelunker (as I've said many times before) to explore their depths.

The scenario above doesn't show an overclocked processor, it merely shows what happens when voltge is raised to an artifical Logic High which the processor will conform to, the problem is it's running hotter and therefore slowly cooking itself to death which is actually how most CPU's die is used until they fail. Now in the chart below lets increase frequency.

http://www.modthebox.com/articles/overclock103/overclock5.jpg

Predictably the sinuwave changes and the artifial Logic Highs are not reached, however; the slightly lower original Vcore Logic Highs are now within tolerance. While this does describe accruately what happens, to be honest I have my doubts. Some people see this as "overclocking" I've rebadged it OverVolting and it's far from the Purist method. Basically what I'm saying is that overclocking can be done several ways and there are methods much safer then manipulating (increasing) Vcore. If there's one law in micro-electronics which should be on your mind continually, it's this: HEAT is the Enemy / Voltage = HEAT.

The charts above were borrowed from ModtheBox which published one of the best overclocking articles ever written. It's to them we owe this re-cap. I learned more readin White Papers on cmos transistors then anything else. Understanding HOW a CPU works helped me to understand everything else.

 

I see what you were getting at now. For some reason I first thought you meant that the chips couldn't get any higher than the rated FSB speeds. What you say is true. One might end up with a lemon chip and overclocking results will not be that great. It's a chance you take, but if you're on a budget, I'd recommend the 4300 since you can use dirt cheap, and slower, memory because of the higher multiplier. However, if you're not strapped for cash, a 6XXX processor is great as well.

 

Your totally correct, I mean ultimately the 4300 is the way to go for a budget overclocker and even overclocker's with the money to spare love them because a few have hit 100% overclocks.

All I was saying was that people are now claiming recent purchases haven't been getting those really great overclocks. If it's a bugdet system then the 4300 is the best advice and even if it's not budget as I said above it's still a great chip
Oh by the way all that writing about overclocking was just something that was on my mind so I thought hey why not. I am not trying to sound like some expert which I am not, nor am I trying to offend anyone.

I posted here because I was looking for something on Google and I landed on this thread somehow. This is the very first site I ever posted at. My username wasn't liquid3D then and tjhat was over 7-years ago. Sinvce then I've studied every white paper and read everything I could get my hands on from basic physics to Wafer Fabrication. I studied Computer Science back in 1988 so I've been arounbd PC's a while and really love the DIY hobby. Anyway as I said I would never claim to know everything about anything especially computer science or overclocking. My philosophy is that the only way I continue to learn is to remain teachable.

I respect your reply very much because I respect the humility in it, this is an admirable quality and one more people in Forums everywhere would do well to learn.