RAM, which one to get?
I've been looking at RAM prices, and they're rather cheap ATM. I need to run VPC so I can take work home with me. I have a PC too, but I'd like to be able to move info around and access my outlook without any hassles. If it's slow, that's not a big concern, I will move docs into my OSX and work on them there, then post things back up for the group when I'm done.
Anyway, I figure bumping the RAM from 256 to 640 on my 12" PB867 should give a nice performance boost.
Is it really crucial to buy the most expensive RAM? Or is it all pretty much the same crap? Also, would I gain any extra measure of reliability from buying DDR333 and running it at the 266 speed that my PB uses. has anyone run into problems running RAM at a lower than advertised speed?
Some of the 333 RAM is cheaper than the equivalent 266 size, from the same manufacturer. It could just be a quirk of our local supplier???
Anyway, I figure bumping the RAM from 256 to 640 on my 12" PB867 should give a nice performance boost.
Is it really crucial to buy the most expensive RAM? Or is it all pretty much the same crap? Also, would I gain any extra measure of reliability from buying DDR333 and running it at the 266 speed that my PB uses. has anyone run into problems running RAM at a lower than advertised speed?
Some of the 333 RAM is cheaper than the equivalent 266 size, from the same manufacturer. It could just be a quirk of our local supplier???
Comments
Don't pay too much attention to SPD timings like CAS latency and all that. I've personally never cared to notice the difference. But if you don't plan on using the RAM in other machines, it might be better just to get the PC2100 with the faster timings than the PC2700 with the slower, because the Mac won't let you set them manually like a PC's BIOS typicall allows. If the PC2700 is cheaper, just get that.
I don't know much about DDR SO-DIMMs, so I'd probably just get the Crucial stuff.
Some people buy PC2700 for this reason, paying MORE for it, so if you can get it for less it's a no brainer.
Top Tier and Generic Memory
How testing influences which name appears on your module.
Sure, testing is important. But does it really matter what will happen to your memory at 200° F when it's a comfortable 70° F in your house? Actually�it does.
Even though it's 70° F in your house, the chips inside your system get a lot hotter. One common cause is a spike in power that reaches your PC (a good reason to use a surge protector). Also, any normal computer use will cause the parts inside your PC to heat up. Basically, your computer may see conditions that are more extreme than what you see in front of the monitor. For this reason, major DRAM manufacturers use what's called guardbands in their testing.
It's a pretty simple concept really, kind of like insurance. A guardband helps to ensure that a chip will work properly, even under many abnormal operating conditions. For example, let's say a chip needs to meet a particular parameter between 10 and 20 to be acceptable. Rather than set the test to fail anything less than 10 or greater than 20, the test is set to fail anything less than 11 or greater than 19. Yes, more failures occur this way, but these additional failed parts are close enough to the limits to imply some sort of process variation. It's not worth the risk to the chip manufacturer to ship these parts to customers under their brand name.
Top Tier Memory
After going through this extensive series of tests, you can be confident that the passing parts will work for a long time.
Generic Memory
So what happens to the parts that failed testing? Some get thrown out, but many are re-tested and graded to lesser speeds and/or conditions (i.e. without guardbands). Others aren't even re-tested (even though they might not have gone through burn-in). Yes, these lower-grade modules will probably work today, but they're more prone to marginal performance and failures, especially over time. DRAM manufacturers often sell this memory to third party vendors who then re-label them with their own brand name. Third party vendors are often required to remove the original DRAM manufacturer's name from the chips. This memory is typically referred to as "generic memory."
Unfortunately, substandard memory can affect you in more ways than you may realize. It can cause problems ranging from annoying error messages to a complete system crash. When these failures occur, critical data that's stored in memory is typically lost or overwritten. And it's not always obvious that memory is the cause of your problem.
Let's say you're building a system, you assemble it completely then turn on the power and nothing happens. You may not suspect the memory at first. Most people will double check all connections, maybe swap a power supply with a different machine and spend hours trying to solve the problem, when in fact it was a failure due to bad memory.
Or, your system may seem to work fine at first then mysteriously crashes later. Most of us have experienced a "blue screen" at one point in our computer lives and we all know how unreliable PCs can be. However, it's not always easy to tell what caused the blue screen. Was it a bad cell in the memory? Was it an intermittent failure in the memory? Was it something else?
Low-grade memory is also more prone to compatibility problems. For example, your computer might only recognize a portion of the memory, or even worse, it won't see the memory at all. Or maybe you upgrade your PC only to discover that the new generic module won't work with the existing memory, so you're stuck using just one module or the other.
Marginal memory is certainly not the only cause of computing problems. But it can contribute to them. Why do you think major memory manufacturers like Micron refuse to ship these lower grade parts? It's because they don't want their name on it.
The Assembly Process
How poor assembly procedures affect you.
Module assembly can have a considerable effect on how the end product works. Is it automated? Or are people soldering on parts by hand in a warehouse? You probably won't be able to tell if the parts went through proper material storage, screen print, and reflow processes during assembly just by looking at them. However, these are some of the practices that play key roles in the short and long term reliability of your memory module.
During the reflow process, (where the DRAM components are permanently attached to the PCBs) the boards go through a heat cycle of a few hundred degrees. If they are heated too quickly or remain at an elevated temperature too long, they can be significantly damaged. However, this damage may not be visible. If the parts weren't properly stored, any large amount of moisture that was trapped inside tends to expand and cause a failure almost immediately. Additionally, the reflow process can sometimes affect the module just enough so that it performs properly initially but degrades over time. This can result in memory that doesn't work by the time the end user gets it, or memory that fails intermittently.
Module Test
You would think all memory assemblers would test every module they build.
Some assemblers test only a sample of parts and call the entire lot good based on an "acceptable defect level." These companies argue that it's cheaper to get a certain amount of returns than to detect and/or prevent them. This may be true for their bottom lines, but if you're in the small percentage of end users that ends up with a failed part, you probably won't think it was a good idea. At Micron, we test each and every memory module we manufacture.
There are several different methods used to assess memory modules. Many of them take modules through a series of tests, including checking for opens and shorts, leakage, verifying refresh rates and running pattern tests. All of these tests are typically performed at high speed to ensure functionality.
As an additional safeguard, many assemblers have invested in automatic handlers to manage their modules. This prevents an operator from accidentally placing a failing module into a good tray. The yield at this step should be greater than 99%.
On top of these production tests, reputable assemblers will also pull additional samples for outgoing quality checks and to qualify the modules on different motherboards. As processors and memory speeds become faster and faster, it becomes increasingly critical for memory modules to do more than just meet standard specifications. They must also be proven to work in specific systems.
Unfortunately, many assemblers consider comprehensive testing an unwarranted expense. Some of the equipment is extremely expensive and requires a great deal of engineering support. On the other hand, it can be very easy for an assembler to buy an inexpensive hand held tester, just to check for opens and shorts and perform no additional testing.
http://www.crucial.com/library/quality_page4.asp
Originally posted by futuremac
Low-grade memory is also more prone to compatibility problems. For example, your computer might only recognize a portion of the memory, or even worse, it won't see the memory at all. Or maybe you upgrade your PC only to discover that the new generic module won't work with the existing memory, so you're stuck using just one module or the other.
guys this happened to me.
i went online last year and was shopping for memory. and i couldn't understand why two sticks of memory both 512 mb, same specs, were so far apart in price. one was 79.00 and the other was like 159.00 so i bought the 79.00 stick, put it in my comp and got a black screen, i took it out and put it in a diff slot and it booted up, but within 5 min the comp rebooted itself, and gave me a bluescreen. and it kept doing this. so i took the new memory out (i had 3 sticks, 2 old and the one new stick) and my comp was back to normal. so it doesn't always pay to get the cheapest.
however...
i know some people who buy cheap generic memory and have no problems...
Originally posted by progmac
i've been buying generic memory for 7 years ( ~ 15 sticks prolly) and have never had a single issue. i'd get generic, but then again, i'm cheap
thats what i mean progmac its odd, it seems so hit-and-miss. i had to return several sticks before i found one that worked years ago before i knew about the diff in memory. however it does pay to check your memory if you buy a new stick and you then have stability problems.
i had a newbie friend who had just bought new ram and was as ususal cursing aol and blaming it on them, so i swapped out the bad ram with some good ram and problem solved.
From most affordable to most expensive, all 512MB, all Canadian funds:
266 Viking, 121
266 Corsair, 142
333 Corsair, 142
266 US Modular, 142
266 Premium, 149
266 K-Byte, 149
333 K Byte, 155
333 US Modular, 155
333 PNY, 162
266 PNY, 175
266 Corsair CAS 2.5, 230
I highlighted the 333 PNY 'cause it's actually less than the 266, and the Corsair stuff is the same price for either 266 or 333. I wonder if most of it ain't just the same RAM tested and rated differently?
In any case, I would imagine that using one of the 333 modules at a lower speed might improve its reliability.
Anyone with experience with any of these brands?
i just wasen't willing to throw out two perfectly good stcks that have always worked.
who knows it could even be the mobo? i have a soyo dragon plus board i got in 2002 maybe ill upgrade to a sata nforce 3 board
the 512 stick i got (2 weeks ago) was like 89.00 and it would not work with my older memory i have crucial memory in my machine and it works fine (not that im plugging crucial)
im told the high-end Corsair is good (if expensive)
i wonder if there is any type of software that tests memory sticks?
I strongly suggest that you check out RAMJET. They stand behind what the sell and they sell a quality product.
PowerBook Ram
If you have doubts, call them and ask them what's the difference between their RAM and others. Personally, I don't mind spending a little more money on a quality product - that's why I buy Macs
Chas
(anti-RAMJET comments removed, as they no DO ship to Canada)
It is thirty frickin dollars US though.
Here's the latest update.
I don't know if this is a Canadian phenomenon, but here's the latest problem I've run into when trying to buy RAM -- "The Configurator!"
Got me at the the local Staples/Office Place, a local computer shop, and now it's nailed me online aswell! Fvck!
I went to www.Tigerdirect.ca
They're mail order, they've got a warehouse and brick and mortar store about 30 minutes from where I live, and everyone I know has had good experiences with them. Free shipping over 100 Canadian too!
Why not, right?
So, I call in, and as soon as I say "notebook" I'm smacked in the face with the good old, "What brand? what model, etc etc..."
See, on the website they list DDR So-dimms without any such nonsense.
I checked my PB spec beforehand, 2.5v 266Mhz 200-pin DDR SO-DIMM non-parity. OK, so I'm ready to order, the only thing that catches me out is the stipulation that the module must be no taller than 1.25"
Bloody TigerDirect has no way to tell me about the actual dimensions of their chips.
I refuse to believe that the 175 dollar PNY 266 is the only module that fits my powerbook.
Are all SO-DIMMS 1.25" or less in height? Can I just go ahead and order the RAM or is this one of those cases where RAM suppliers play around with the spec and some of the modules listed may actually be too tall?
Click here and buy it from Crucial. $100 US and you don't have to worry about it not working.
Just order one less pint the next few times you go out, and you'll make up the difference.
It's worth a few dollars for the peace of mind.
http://www.corsairmicro.com/corsair/.../cm64sd256.pdf
meets the 1.25" spec
buy corsair
waht size chip were ya wanting? 256 or 512?