hi guys, im new on building computers and need your help.
my question is:
1. how do you match processor fsb against memory fsb.
ie. if i have a P4 D805 with fsb of 533mhz or a P4 D930 with fsb of 800mhz of a E6300 with an fsb of 1066mhz; what type of memory steup &/or speed should i get so i wont get a bottle neck? the minimum and the ideal setup?
2. can anyone explain what Synchronous & Asynchronous mean? what is the difference? and which one is better/faster?
3. i read that dual channel doubles the fsb in theory. does this apply in real world? or does is just increase the fsb a few percent more only?
1. Idealy you want the ram and processor fsb frequencies to be the same (1:1 DRAM:FSB) On the socket 478 and LGA775 Intel CPU's (up to the Pentium D 900 series), the real FSB frequency is 200Mhz (800Mhz effective clock). On motherboards which use DDR1 ram up to DDR400, the DRAM:FSB divider (aka the RAM divider) would be 1:1 because both the FSB and the RAM operate at 200Mhz. With the Core 2 processors just out from Intel, the FSB is now 266Mhz (1066Mhz effective clock). To achieve a 1:1 ratio, you would need to use either DDR2-533, or to achieve a 2:1 ratio (DRAM:FSB) you would need DDR2-1066 (which is rediculously expensive). To be honest though, most people don't really care about synchronous operations with a Core 2 chip though. They are so fast right now, it doesn't really matter. For anyone that cares, using DDR2-800 with a Core 2 chip would result in a 2:3 ratio, and DDR2-667 would result in a 4:5 ratio.
2. When refering to dual channel operation, synchronous is when you have 1 (or 2) ram sticks occupying each ram channel which are of the same capacity (for example: 1 512MB stick in each channel, or 2 512MB sticks in each channel. the same goes for 1GB sticks). Asynchronous dual channel occurs when you have any number (except 1) of ram sticks installed which equate to differing sizes in each channel. Example... if one had 1 512MB stick in the first channel, and 1 256MB stick in the second channel (BIG NOTE: Asynchronous dual channel operations can only occur on a motherboard with an Intel chipset which supports Intel's Flex Memory Technology. Attempting to run an Asynchronous dual channel setup on any other chipset will result in single channel operation). Asynchronous DC setups lose a bit of available memory bandwidth, therefore Synchronous dual channel is faster.
3. No, dual chanel does not double the FSB. What it does do is double the amount of memory bandwidth available. For example, 1 stick of DDR400 provides 3.2GB/s of memory bandwidth. 2 sticks of DDR400 in dual channel (this is a Synchronous setup, because the chipsets which support Flex Memory only support DDR2) would provide 6.4GB/s of memory bandwidth, thus the rating PC3200, or PC2700 on DDR1 ram sticks. The same goes for DDR2....
quote "Following the 4:5 (2:3 unessasary) rule for optimizing Netburst-DDR2, that equates to DDR2-500. So yeh, you'll only need DDR2-533 and due to the characteristics of P4, timing isn't so important so you can buy a cheap pair of 2x1GB DDR2-533 3-4-3-12 with the D805. "
I am looking at a Asus P5B Deluxe WiFi and it says that the memory is clockable in 1Mhz increments from 533 to 1066Mhz. Since I also want to OC a D 805. Cannot I simply OC the CPU as far as I can get it, then use whatever memory ie. DDR2-800 and clock that as far as it will go. Asyncroniously of course? Is this not true and/or will I get any performance out of this method?
There is something about this board having a fixed ratio for memory? can you explain how that is detrimental to OC'ing when is sayd you can set the memory FSB in 1Mhz incraments?
At least support your statement with some factual reasons why you think its crap. That Chip can run reliably at 4Ghz and that puts it right up there with the Pentium EE HT chips and amd X2 4800's and higher. And thats a fact proven time and time again. Even Toms Hardware based a couple articles on this fact. And since it is less than $100 ....Well I guess I need you to back up that statement, or I'll just assume your talking out of your rear end!
Look at Neweggs reviews of it. OF the 270 peeps that revieved it 97% of them gave it a 4 or highter. Of those 270 reviewers, only 7 people gave it a 3 or less eggs. And the ones that didnt like it, said it was either too hot or a dead CPU. Heat is know issue easily remedied witha really good HS and you really cant review a DEAD cpu cause its dead. After relpaceing it, then give your opinion on it, dont give it a one cause its dead! Far as I can Tell the D805 is gods gift to gamers on a budget.
The Newegg reviews are complete sh*t dude.... I never trust any of them for any reason.
Tom's benchmarks did show FX-57 and X2 4800+ like performance, but in order to achieve that performance, you have to get quality liquid cooling setups (which effectively double the price of the processor). Then you have to deal with the hit in the wallet in terms of power requirements. This thing at high clock speeds puts out more heat and uses more power than the P4 Socket 478 Prescott's did. The 533 stock FSB really hurts this chip. The 915 I linked to before could probably beat this chip with a slight overclock (therefore not needing expensive cooling, and not putting your electic bill through the roof). Hell, a low end Core 2 system could probably beat this while underclocked.
Don't waste money on inferior chips, especially those that can perform with everyone else, but need massive amounts of money thrown at them to do it.
You seem very knowledgable so if you dont mind me extracting your wisdome...I had planned on buying a CNPS9500 led but have heard that the 7700-cu's and some scythe model are actually better as well as cheaper. Can you verify this and has anyone done any reviews lately comparing them all togather? Id like to see the results.
I got a CNPS9500 and was absolutely amazed by it's performance. The P4 630 (OC'ed to 3.4) in the system in my sig had it's idle temp dropped by 5C, and load temp dropped by 15C. Idle/Load temps used to be 32C/55C, and are now 27C/40C. All of the temps I just listed are with the 630 OC'ed to 3.4. The 7700 MIGHT outperform the 9500, but it depends on what chip they are going on to. I believe the 7700 is louder though. The Scythe Ninja is also a good cooler. I don't remember how it performs, but I do know it is definitly cheaper and a bit quieter as well.
Are you HIGH?!! The Scythe Ninja and other Scythe didnt cool any better! The Ninja only pulled a 59 deg C. The CNPS9500 led pulled a 60 deg C. One degree!! we are talking one degree. And for one degree of cooling Im going to buy a HS and fan that is 30 percent larger??? I dont think soo. They are quieter yes But im not building a silent PC. Im not knocking scythe at all its just that you sound off like they are WAY better and ..well...they're not. A couple of them performed just a hair, slight hair at that, better. Just ALOT bigger. Plus the 9500 looks way kewler!
Just my thoughts on it. I dont mean to come off arguementative. But after saying it wont handle OC'a D 805 you post a link that show quite clearly that it is just as good as any.
I was suprized buy the Freezer pro 64! I didnt think it was that good. Too bad its amd
EDIT: IF they made a 120mm version of the Cnps9500 I'm pretty sure it would put down everything else!!
i'm a newbie here so can you kindly explain what "4:5 rule for Netburst" is?
when you talk about the P4 D930 reaching 4.0 ghz and requiring DDR2-667 using the netburst rule. this is about overclocking requirements.
what about when using them on stock speeds? what kind ram do i need on the P4 D805, P4 D930 and E6300? and is this on single channel setup or dual channel? (since i am not planning to overclock my system yet. i have no idea how to do that yet, safely).
Note the different RPM/noise fan being used on the coolers. 45dB vs. 35dB is a noticible difference
But 45dB of noise at 50cm (2 inches) is not going to be 45 dB when the case is closed and tucked under my desk 4 feet away from me. With that in mind, I dont expect Ill hear that thing over my case fans at all.
No 92mm can ever match a 120mm HSF under the same condition (CFM, noise, room temp...etc.)
But it just did.
the 92mm CNPS9500 had the exact same temp as the Shogun and Katana. Was only one degree behind the Ninja and the Mine.
I doesnt matter what CPU they are on. Its a relative comparison to each other. So if you put them all on a much higher wattage CPU, the temps go up relatively the same for all. That one degree difference isnt going to turn in to 15 degree difference is it? Maybe it is, Im not an expert in thermaldynamics.
Dont get me wrong, I get what you are saying. Im just saying that the superior design of the 9500 allows it to work as well as larger 120mm HS's. As I said before, if they made a 120mm version of the 9500 who do you think it would do against the rest? I think it would be FAR superior because of its design.
See Im switching from agp ddr400 to pci-e and DDR2 so its a new mobo, ram, cpu, and HS all in one shot. Thats a big chunk of money all at once. I wont skimp on a mobo cause I want to use it for a while. So I have no choice to go REALLY cheap on CPU, and ram. when I recoup some money I will definately take your guys advice and buy a better CPU. Then Better ram.
EDIT: We are kinda hijacking this guys thread. Lets make another one cause I have a couple more questions and what I need is convincing. Ill start a thread in the cooler section and copy paste thei thread as a starter. Ill call it 9500 vs scythe.
hey generals i got stuck at the same problem as you too. I want to know the minimum speed at which the memory bus must run for no bottlenecks to occur. Most people are saying that so long as the FSB speed and memory bus speed are the same there will be no bottlenecks. If you calculate it this way, you are going to spend extra money. See:
On P4 805 D, FSB533, system clock 133MHz, the FSB bandwidth = 4.26Gb/s.
To achieve the same bandwidth for the memory bus, you could....
a) Run a DDR 533 ram on a single slot (bandwidth = 4.264Gb/s)
b) Run two DDR 266 ram on dual channel (bandwidth = 4.26Gb/s)
dual channel doubles bandwidth, but its only for intel processors as AMD does not have it.
You could run RAM at higher clock speeds than required (which is what most people do), but doing so wont bring in significant performance and will increase your electricity bill by a tiny tiny bit.
So, if i were to...
Run two DDR 533 ram on dual channel (bandwidth = 8.528Gb/s), I would be having exess and redundant bandwidth.
I think the answer is so long as the FSB bandwidth and Memory Bus bandwidth are the same, your PC will have good performance.
People out there, please tell me whether you agree that we should look at bandwidth to check bottlenecks.
E6300 is quite different, it's like creating an art piece.
It needs specificly Gigabyte 965P-DS3/DS4/DS6 with the new beta BIOS, DDR2-800 4-4-4-12 and optionally, pencil mod to the mobo if 500Mhz+ FSB is to be reached giving the E6300 @3.5Ghz.
Overclocking the E6300 is concerntrated at mobo and RAM work (giving them more voltage) to elevate FSB as high as possible to insane never-heard-of clockspeeds.
what about stock E6300 speeds? would DDR2-533 on dual channel be ideal?
I am having a tad bit of trouble with my Patriot 512x2 Dual Channel DDR400 on my Biostar 6100-M9, now, i didnt exactly understand what u meant by "look at the colors on the slots" and somehow, my POST keeps recognizing it as Single Channel 1 gig, also, someone mentioned that DDR400 and a processor both run 200 MHZ, but its supposed to be DDR400=400mhz, another thing i have noticed also is that in my handbook for my MB, it says that i am supposed to have 2 peices of RAM next to eachother (I have 4 DIMMs with alternating White and Blue) i need help, thx alot in advance