If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. |
|
|
Thread Tools | Display Modes |
#1
|
|||
|
|||
Two DIMMs, three differences. Which is better?
So I need memrory for a computer that has only 256K!
A Dell Dimension 4700, it can take four 1-gig DIMMs and Cruscial recommends one of the two below, but says the other is compatible too. Everything about them including the price is the same, except what I list on lines 2 vs.3 below. Which is better and why do they recommend the other one? :-) I know I could buy either one and probably not notice the difference, but I would like to understand this. Both a 2GB kit (1GBx2) Unbuffered NON-ECC 1.8V 128Meg x 64 DDR2 One is: PC2-5300 CL=5 DDR2-667 Part #: CT908120 The other is: PC2-6400 CL=6 DDR2-800 Part #: CT1449212 If you want to see the url itself, this is it: http://www.crucial.com/store/listpar...04700%20Series) I'd also note that again the listings say nothing about parity. Or do they? Perhaps 128Meg x 64 means even parity since it's 64 and not 72 ?? OTOH, if I'm confused and that's not how you figure out parity: Now when I'm using the Crucial or Kingston part number finder, I guess they don't have to say, but Crucial describes all those other values. Why is parity left out Thanks a lot. |
#2
|
|||
|
|||
Two DIMMs, three differences. Which is better?
micky wrote:
So I need memrory for a computer that has only 256K! A Dell Dimension 4700, it can take four 1-gig DIMMs and Cruscial recommends one of the two below, but says the other is compatible too. Everything about them including the price is the same, except what I list on lines 2 vs.3 below. Which is better and why do they recommend the other one? :-) I know I could buy either one and probably not notice the difference, but I would like to understand this. Both a 2GB kit (1GBx2) Unbuffered NON-ECC 1.8V 128Meg x 64 DDR2 One is: PC2-5300 CL=5 DDR2-667 Part #: CT908120 The other is: PC2-6400 CL=6 DDR2-800 Part #: CT1449212 If you want to see the url itself, this is it: http://www.crucial.com/store/listpar...04700%20Series) I'd also note that again the listings say nothing about parity. Or do they? Perhaps 128Meg x 64 means even parity since it's 64 and not 72 ?? OTOH, if I'm confused and that's not how you figure out parity: Now when I'm using the Crucial or Kingston part number finder, I guess they don't have to say, but Crucial describes all those other values. Why is parity left out Thanks a lot. ECC is a 72 bit wide array, non-ECC is a 64 bit wide array. The listing for your computer, lists only non-ECC, implying there is no ECC support in the computer. Intel tends to support ECC, only on a limited number of desktop chipsets, while on their server chipsets, such support would be a given (always present). You don't generally run servers without ECC, because the people buying the servers, know better. Servers have humongous memory stores, and detecting errors there is important. There are at least three kinds of protection in existence. Parity, ECC, ChipKill. To save time, there are some half-assed articles here. http://en.wikipedia.org/wiki/Parity_bit http://en.wikipedia.org/wiki/ECC_memory http://en.wikipedia.org/wiki/Chipkill (for 128 bit wide memory arrays, x4 chips) The reason the DDR2 PC2-6400 4-4-4-12 Unbuffered NON-ECC DDR2-800 2.0V 128Meg x 64 Part #: CT834311 is not a recommended part, is because it meets timing when the DIMM voltage is "2.0 volts". Your computer may lack an adjustable Vdimm setting. A memory product with a 1.8V voltage rating at stated timing, is a better choice for a computer without extensive BIOS controls. The 1.8V product is more likely to "just work". Since you claim to only have 256MB in the computer right now, I would remove that stick, and install a 2x1GB kit in dual channel mode. That may operate the memory interface, in as fast a mode as possible (not that you'd notice or anything). Keep the 256MB stick handy - don't sell it. If you ever have trouble with your Crucial product, and need warranty service, that 256MB stick can be used while it is being dealt with. Keep the 256MB stick in an antistatic bag. When the new memory arrives, download memtest86+ from memtest.org and prepare a floppy or CD with the software. Run a memory test first, to make sure the memory is good, before booting into Windows for the first time. If the memory is bad, you can corrupt Windows while using the bad RAM. A simple test, is better than nothing, before your first Windows boot. This tool is a boot CD or boot floppy, and no OS runs in the computer while this program is doing its thing. Make sure the boot order is set in the BIOS, to access this test media first. http://www.memtest.org (downloads, half way down the page) The Crucial site claims your motherboard chipset is 915G. This is a document from Intel, with the various memory configuration options available. You want Dual Channel Symmetric Memory Mode, as shown in Figure 3. With your new 2x1GB kit, you place a 1GB stick in each channel. I think what the diagram is communicating, is the DIMMs don't have to be "directly across" from one another, unlike on a first generation S939 system. You just need to install a 1GB DIMM, on each channel, one per channel. http://download.intel.com/design/chi...s/30167003.pdf If you insist on installing and keeping the single 256MB stick too, that puts you in Figure 2 Asymmetric Mode. And that's why I was recommending the removal of the 256MB DIMM, while the nice matched set is present. Paul |
#3
|
|||
|
|||
Two DIMMs, three differences. Which is better?
On Sun, 10 Jul 2011 16:00:08 -0400, Paul wrote:
micky wrote: So I need memrory for a computer that has only 256K! A Dell Dimension 4700, it can take four 1-gig DIMMs and Cruscial recommends one of the two below, but says the other is compatible too. Everything about them including the price is the same, except what I list on lines 2 vs.3 below. Which is better and why do they recommend the other one? :-) I know I could buy either one and probably not notice the difference, but I would like to understand this. Both a 2GB kit (1GBx2) Unbuffered NON-ECC 1.8V 128Meg x 64 DDR2 One is: PC2-5300 CL=5 DDR2-667 Part #: CT908120 The other is: PC2-6400 CL=6 DDR2-800 Part #: CT1449212 So my question is, how come Crucial recommends the first of these, when the second one is faster? Is it because CL=6 is not as good for this compute?. Oh, yeah. I didn't pay enough attention to CL before and I figured higher was better, but lower is better and that must be why they recoomend the one t hat is CL=5, even if the other number(s) is slower. ?? More, below. If you want to see the url itself, this is it: http://www.crucial.com/store/listpar...04700%20Series) I'd also note that again the listings say nothing about parity. Or do they? Perhaps 128Meg x 64 means even parity since it's 64 and not 72 ?? OTOH, if I'm confused and that's not how you figure out parity: Now when I'm using the Crucial or Kingston part number finder, I guess they don't have to say, but Crucial describes all those other values. Why is parity left out Thanks a lot. ECC is a 72 bit wide array, non-ECC is a 64 bit wide array. The listing for your computer, lists only non-ECC, implying there is no ECC support in the computer. Intel tends to support ECC, only on a limited number of desktop chipsets, while on their server chipsets, such support would be a given (always present). You don't generally run servers without ECC, because the people buying the servers, know better. FTR this is not the computer for which I ordered the server cpu. This is just an out of the box Dell that a friend bought with very little memory. Servers have humongous memory stores, and detecting errors there is important. There are at least three kinds of protection in existence. Parity, ECC, ChipKill. To save time, there are some half-assed articles here. http://en.wikipedia.org/wiki/Parity_bit http://en.wikipedia.org/wiki/ECC_memory http://en.wikipedia.org/wiki/Chipkill (for 128 bit wide memory arrays, x4 chips) The reason the DDR2 PC2-6400 4-4-4-12 Unbuffered NON-ECC DDR2-800 2.0V 128Meg x 64 Part #: CT834311 is not a recommended part, is because it meets timing when the DIMM voltage is "2.0 volts". Your computer may lack an adjustable Vdimm setting. A memory product with a 1.8V voltage rating at stated timing, is a better choice for a computer without extensive BIOS controls. The 1.8V product is more likely to "just work". Since you claim to only have 256MB in the computer right now, I would remove that stick, and install a 2x1GB kit in dual channel Thats' what I plan to do. I only mentioned the 256 so I wouldn't look like a wastrel, spending money willy-niilly on memory where it wasn't needed. mode. That may operate the memory interface, in as fast a mode as possible (not that you'd notice or anything). Keep the 256MB stick handy - don't sell it. If you ever have trouble with your Crucial product, and need warranty service, that 256MB stick can be used while it is being dealt with. Keep the 256MB stick in an antistatic bag. When the new memory arrives, download memtest86+ from memtest.org and prepare a floppy or CD with the software. Run a memory test first, to make sure the memory is good, before booting into Windows for the first time. If the memory is bad, you can corrupt Windows while using the bad RAM. Good to know. I'll do that. Thanks a lot. Micky A simple test, is better than nothing, before your first Windows boot. This tool is a boot CD or boot floppy, and no OS runs in the computer while this program is doing its thing. Make sure the boot order is set in the BIOS, to access this test media first. http://www.memtest.org (downloads, half way down the page) The Crucial site claims your motherboard chipset is 915G. This is a document from Intel, with the various memory configuration options available. You want Dual Channel Symmetric Memory Mode, as shown in Figure 3. With your new 2x1GB kit, you place a 1GB stick in each channel. I think what the diagram is communicating, is the DIMMs don't have to be "directly across" from one another, unlike on a first generation S939 system. You just need to install a 1GB DIMM, on each channel, one per channel. http://download.intel.com/design/chi...s/30167003.pdf If you insist on installing and keeping the single 256MB stick too, that puts you in Figure 2 Asymmetric Mode. And that's why I was recommending the removal of the 256MB DIMM, while the nice matched set is present. Paul |
#4
|
|||
|
|||
Two DIMMs, three differences. Which is better?
micky wrote:
On Sun, 10 Jul 2011 16:00:08 -0400, Paul wrote: micky wrote: So I need memrory for a computer that has only 256K! A Dell Dimension 4700, it can take four 1-gig DIMMs and Cruscial recommends one of the two below, but says the other is compatible too. Everything about them including the price is the same, except what I list on lines 2 vs.3 below. Which is better and why do they recommend the other one? :-) I know I could buy either one and probably not notice the difference, but I would like to understand this. Both a 2GB kit (1GBx2) Unbuffered NON-ECC 1.8V 128Meg x 64 DDR2 One is: PC2-5300 CL=5 DDR2-667 Part #: CT908120 The other is: PC2-6400 CL=6 DDR2-800 Part #: CT1449212 So my question is, how come Crucial recommends the first of these, when the second one is faster? Is it because CL=6 is not as good for this compute?. Oh, yeah. I didn't pay enough attention to CL before and I figured higher was better, but lower is better and that must be why they recoomend the one t hat is CL=5, even if the other number(s) is slower. ?? Lower CAS latency (CL value) is better. CAS5 is faster than CAS6. It's basically a measure of the time needed, for the memory to look up the data. The address is presented, the memory array is accessed, and your results come back. Reducing that time, improves the speed you can make "random accesses" to the memory. The CAS5 memory, would allow more random accesses per second, than the CAS6 would. It affects the overall cycle time to some extent. CAS is measured in units of "clock ticks". There is usually an associated clock speed spec (like the DDR2-667 thing), which allows a person to work out the number of nanoseconds a "clock tick" is equal to. You multiply the CAS number, by the clock period in nanoseconds, to obtain the time penalty for looking up a result inside the memory. The chipset on the motherboard, is the other limiting factor. There have been some Intel chipsets, where you can buy very fast RAM (low CL), but the CAS setting can't be dialed low enough on the chipset to take advantage of it. The reason this happens, is the chipset design isn't ready to accept data, when the RAM offers it. Not a big deal, and mostly only an issue for people who spent hundreds of dollars for that "special DIMM". They're supposed to check the available info on the chipset first, to make sure they can actually use the special properties of the DIMM. And minimum CAS value, isn't stated in official chipset docs - it tends to be discovered the hard way, on overclocker sites. The chances of you needing to worry about this, are very very slim :-) If they're both the same price (CAS5 and CAS6), buy your CAS5 and enjoy. Paul |
#5
|
|||
|
|||
Two DIMMs, three differences. Which is better?
micky wrote: On Sun, 10 Jul 2011 16:00:08 -0400, Paul wrote: Which is better and why do they recommend the other one? :-) I know I could buy either one and probably not notice the difference, but I would like to understand this. Both a 2GB kit (1GBx2) Unbuffered NON-ECC 1.8V 128Meg x 64 DDR2 One is: PC2-5300 CL=5 DDR2-667 Part #: CT908120 The other is: PC2-6400 CL=6 DDR2-800 Part #: CT1449212 So my question is, how come Crucial recommends the first of these, when the second one is faster? The reason the DDR2 PC2-6400 4-4-4-12 Unbuffered NON-ECC DDR2-800 2.0V 128Meg x 64 Part #: CT834311 is not a recommended part, is because it meets timing when the DIMM voltage is "2.0 volts". Your computer may lack an adjustable Vdimm setting. A memory product with a 1.8V voltage rating at stated timing, is a better choice for a computer without extensive BIOS controls. The 1.8V product is more likely to "just work". What Paul said (always). Try to get memory modules made from chips marked with the actual chip manufacturer's logo or full part number because any other modules are made either from junk chips (failed chip manufacturer's testing), chips that weren't fully tested (sold as whole wafers to other companies that cut up and package them, like Kingston), or chips being run at faster than recommended speeds. If the voltage rating is higher than standard, which would be 1.8V for DDR2, it means the chips failed testing at standard voltage. Heatsinks are another bad sign because they're more likely to be purely decorative than functional (old RAMbus and some Samsung DDR3 modules being exceptions), and it's cheaper to slap on a pretty heatsink than to use higher quality chips. Crucial generally sells only top quality modules with chips marked by their actual makers and standard voltage ratings, but their Ballistix line is the exception, and it's their only line with heatsinks and higher voltage ratings. Ballistix also has the highest failure rate among all Crucial modules. Actual chip manufacturers include Micron ("M" with orbiting elipse, but watch out for the Superman "S", indicating they're recycled chips from Spectek), Hynix, Samsung (SEC), Elpida, ProMOS, PowerChip (tilted bow tie), and Nanya/Inotera. Whatever you buy, be sure to test it thoroughly overnight with MemTest86, MemTest86+ (both use the same test methods, but results often differ), and Gold Memory. Don't accept even a single bad bit. |
#6
|
|||
|
|||
Two DIMMs, three differences. Which is better?
On Sun, 10 Jul 2011 15:05:59 -0700 (PDT), "larry moe 'n curly"
wrote: micky wrote: On Sun, 10 Jul 2011 16:00:08 -0400, Paul wrote: Which is better and why do they recommend the other one? :-) I know I could buy either one and probably not notice the difference, but I would like to understand this. Both a 2GB kit (1GBx2) Unbuffered NON-ECC 1.8V 128Meg x 64 DDR2 One is: PC2-5300 CL=5 DDR2-667 Part #: CT908120 The other is: PC2-6400 CL=6 DDR2-800 Part #: CT1449212 So my question is, how come Crucial recommends the first of these, when the second one is faster? The reason the DDR2 PC2-6400 4-4-4-12 Unbuffered NON-ECC DDR2-800 2.0V 128Meg x 64 Part #: CT834311 is not a recommended part, is because it meets timing when the DIMM voltage is "2.0 volts". Your computer may lack an adjustable Vdimm setting. A memory product with a 1.8V voltage rating at stated timing, is a better choice for a computer without extensive BIOS controls. The 1.8V product is more likely to "just work". What Paul said (always). Try to get memory modules made from chips marked with the actual chip manufacturer's logo or full part number because any other modules are made either from junk chips (failed chip manufacturer's testing), chips that weren't fully tested (sold as whole wafers to other companies that cut up and package them, like Kingston), or chips being run at faster than recommended speeds. If the voltage rating is higher than standard, which would be 1.8V for DDR2, it means the chips failed testing at standard voltage. Heatsinks are another bad sign because they're more likely to be purely decorative than functional (old RAMbus and some Samsung DDR3 modules being exceptions), and it's cheaper to slap on a pretty heatsink than to use higher quality chips. Crucial generally sells only top quality modules with chips marked by their actual makers and standard voltage ratings, but their Ballistix line is the exception, and it's their only line with heatsinks and higher voltage ratings. Ballistix also has the highest failure rate among all Crucial modules. Actual chip manufacturers include Micron ("M" with orbiting elipse, but watch out for the Superman "S", indicating they're recycled chips from Spectek), Hynix, Samsung (SEC), Elpida, ProMOS, PowerChip (tilted bow tie), and Nanya/Inotera. Thank you, and Paul, for all this valuable information. Whatever you buy, be sure to test it thoroughly overnight with MemTest86, MemTest86+ (both use the same test methods, but results often differ), and Gold Memory. Don't accept even a single bad bit. Okay. I've used memtest before and I willl do as you say. I ended up buying straight from Crucial, they're not much more than the cheapest places, plus they have eveyrthing in stock, and one advantage was I woudl be dealing with them and not some guy who wouldn't have anymore if one was bad. So I will accept no bit before its time . |
#7
|
|||
|
|||
Two DIMMs, three differences. Which is better?
micky wrote: Whatever you buy, be sure to test it thoroughly overnight with MemTest86, MemTest86+ (both use the same test methods, but results often differ), and Gold Memory. Don't accept even a single bad bit. Okay. I've used memtest before and I willl do as you say. Don't rely only on Memtest, which is a Windows based program that doesn't catch as many errors as MemTest86 or Gold Memory. Here are only reviews I can find of memory diagnostics, and they're ten years old: http://www.realworldtech.com/page.cf...WT052001232443 http://www.realworldtech.com/page.cf...WT120901222920 I ended up buying straight from Crucial, they're not much more than the cheapest places, plus they have eveyrthing in stock, and one advantage was I would be dealing with them and not some guy who wouldn't have anymore if one was bad. So I will accept no bit before its time . Good choice. I've never had a dud Crucial module, although I never bought one with a heatsink on it. A few times I received modules with Samsung chips on them, and the modules themselves were made by Samsung. They were rated 6-6-6-18 but could be run at faster timings than my OCZ modules rated for 5-5-5-15, not that I normally overclock. |
#8
|
|||
|
|||
Two DIMMs, three differences. Which is better?
larry moe 'n curly wrote:
micky wrote: Whatever you buy, be sure to test it thoroughly overnight with MemTest86, MemTest86+ (both use the same test methods, but results often differ), and Gold Memory. Don't accept even a single bad bit. Okay. I've used memtest before and I willl do as you say. Don't rely only on Memtest, which is a Windows based program that doesn't catch as many errors as MemTest86 or Gold Memory. Here are only reviews I can find of memory diagnostics, and they're ten years old: http://www.realworldtech.com/page.cf...WT052001232443 http://www.realworldtech.com/page.cf...WT120901222920 I ended up buying straight from Crucial, they're not much more than the cheapest places, plus they have eveyrthing in stock, and one advantage was I would be dealing with them and not some guy who wouldn't have anymore if one was bad. So I will accept no bit before its time . Good choice. I've never had a dud Crucial module, although I never bought one with a heatsink on it. A few times I received modules with Samsung chips on them, and the modules themselves were made by Samsung. They were rated 6-6-6-18 but could be run at faster timings than my OCZ modules rated for 5-5-5-15, not that I normally overclock. Crucial isn't "flawless" in the memory department. Maybe back in the DDR or SDRAM days they were OK, but something happened in the move to DDR2 and DDR3. In fact, *before* purchasing memory, I recommend a review of memory products in the Newegg product pages. Taking some sample products (like DDR2, if you're in the market for DDR2), examine the major brand names and see how the reviewers treat them, how many DOAs and so on. I've used that guidance, and actually ended up changing which manufacturer I would buy from. I didn't buy the modules from Newegg, but I did use their reviewer content as an indication of product quality. (If I buy from Newegg, I'd end up paying a brokerage fee at the border. Only if the warehouse is on my side of the Canadian border, can I do business with a company.) Small sample size, but my Kingston purchases *all* still work, and the same can't be said for Crucial. I've had horrible results with "generic" memory, but you and I both know why that is. I don't buy generic any more, but some of my older computers still have that stuff inside, which is why I'm still uncovering failures to this day :-( I bought eight generic modules, installed three in a computer, and two recently failed, leaving me with two groups of three modules to install. Now, if I'd bought brand name, I'd have been able to purchase four modules for the same money, three to install in the computer, and one spare. Am I "still ahead" ? Who knows. Time will tell. It means doing periodic memory testing, as a price for my "cheapness". I have earlier generic module purchases, where *all* the modules failed. Perfection... Time limited offer, means you'll only get 1.5 years worth of usage from your purchase :-) That's what generic UTT gets you. (Article on UTT... So people know what it is. And where the parts for "generic" come from.) http://www.simmtester.com/page/news/...ws.asp?num=124 Paul |
#9
|
|||
|
|||
Two DIMMs, three differences. Which is better?
Paul wrote: larry moe 'n curly wrote: I've never had a dud Crucial module, although I never bought one with a heatsink on it. Crucial isn't "flawless" in the memory department. Maybe back in the DDR or SDRAM days they were OK, but something happened in the move to DDR2 and DDR3. Did the bad modules have generic chips, or were they identifiable as a major brand, like Micron or Samsung? I've heard of just one bad no- heatsink module from Crucial, for a printer, and it had Samsung instead of Micron chips. I don't know if it was a matter of the SPD having the wrong parameters (or if it even had an SPD). In fact, *before* purchasing memory, I recommend a review of memory products in the Newegg product pages. Taking some sample products (like DDR2, if you're in the market for DDR2), examine the major brand names and see how the reviewers treat them, how many DOAs and so on. I've used that guidance, and actually ended up changing which manufacturer I would buy from. I didn't buy the modules from Newegg, but I did use their reviewer content as an indication of product quality. (If I buy from Newegg, I'd end up paying a brokerage fee at the border. Only if the warehouse is on my side of the Canadian border, can I do business with a company.) Small sample size, but my Kingston purchases *all* still work, and the same can't be said for Crucial. I've had good luck with DDR2 Kingston, but my failure rate with their DDR was around 20-30%, with 8 out of 12-13 of their 512MB PC3200 modules failing (fairly consistently according to the markings on the chips and countries of assembly -- USA & Dxxxxxx chips gave 0-5 errors, China & Bxxxxxx hundreds of errors, Taiwan & blank chips thousands). OTOH Kingston's warranty replacements are very good if you speak to higher level tech support and detail the problems experienced with multiple motherboards -- "We'll get you something that WILL work." OTOH until about 3 years ago, I never got a bad Corsair module, but now about 25% have shown errors. I've had horrible results with "generic" memory, but you and I both know why that is. I don't buy generic any more, but some of my older computers still have that stuff inside, which is why I'm still uncovering failures to this day :-( I bought eight generic modules, installed three in a computer, and two recently failed, leaving me with two groups of three modules to install. Now, if I'd bought brand name, I'd have been able to purchase four modules for the same money, three to install in the computer, and one spare. Am I "still ahead" ? Who knows. Time will tell. It means doing periodic memory testing, as a price for my "cheapness". I have earlier generic module purchases, where *all* the modules failed. Perfection... Time limited offer, means you'll only get 1.5 years worth of usage from your purchase :-) That's what generic UTT gets you. (Article on UTT... So people know what it is. And where the parts for "generic" come from.) http://www.simmtester.com/page/news/...ws.asp?num=124 Paul |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Thread Starter | Forum | Replies | Last Post |
Two DIMMs, three differences. Which is better? | micky | General | 3 | July 10th 11 10:11 PM |
4 DIMMS vs 2 | Dennis[_7_] | Homebuilt PC's | 2 | November 28th 09 12:07 PM |
4 DIMMS vs 2 | Paul | Homebuilt PC's | 1 | November 28th 09 07:47 AM |
4 DIMMS vs 2 | Pen | Homebuilt PC's | 0 | November 27th 09 11:11 PM |
Unstable with 2 Dimms | Fitz | Homebuilt PC's | 1 | December 12th 03 11:39 PM |