RAM - memory

Actual speed vs. listed speed

[ControlZ]

Wondering if anyone has any opinions on RAM. I have always stuck with Crucial brand memory but recently could not help but notice the much cheaper price of the some of the "lesser" brands such as Kingston, PNY Corsair, etc.

Does anyone have any real-world experience with some of the brands mentioned? I am also looking for a site that has performed benchmark tests on the various brands of memory, but cannot locate a site that is not obviously pitching one brand or another.

[jdMorgan]

If the computer is set up to use a particular speed of memory, then that's the speed the memory will run at. There are various detail factors that enter into this, such as minimum precharge times and other delays. But for a given 'speed' of memory device or module, these parameters will almost always be very similar, and the computer determines the performance, not the memory module brand.

So the reason you don't see legitimate memory module benchmark tests is that what would need to be benchmarked --given a fixed speed rating for the memory modules-- is the computer, it's CPU, front-end controller, memory controller, etc. and not the memory module itself.

The difference in price is more a reflection of the manufacturers' costs, marketing strategy, and to some extent, the 'quality' reputation of the parts used to assemble the module. If you want to actually analyze the quality factor, then determine the memory chip manufacturer by reading the markings on the devices on the module, and look them up. By referring to the manufacturers' data sheets on the parts, you can get information on MTBF, and compare detailed timing specs to be sure that a particular module manufacturer is not 'stretching' the devices' capabilities.

Also, some manufactures do more testing than others; Obviously a memory module that undergoes complete and thorough 'burn in' testing for 24/48/72 hours is going to cost more than one that is tested for 15 seconds and then shipped.

Jim

[lammert]

Also, some manufactures do more testing than others; Obviously a memory module that undergoes complete and thorough 'burn in' testing for 24/48/72 hours is going to cost more than one that is tested for 15 seconds and then shipped.

This is my experience with cheap memory brands. They either fail the first week because the limited factory tests didn't detect failures, or they will work for the entire life of your computer.

What CAN change the speed of RAM access in your computer is selecting all memory modules of the same size and brand, especially with modern motherboards. If you have four slots for memory modules and two are populated with brand A, and the other two with brand B, the motherboard may switch to a lower memory access mode. You should read your motherboard manual and look for terms like "single channel mode", "dual channel asymmetric mode" and "dual channel interleaved mode", to see which combination of memory modules will give the best performance on your specific motherboard.

[jdMorgan]

Since we've now pretty much fully transitioned to synchronous DRAM, the brands should no longer matter. However, for interleaved or dual channel modes to be used, the modules must indeed be the same size.

Jim

[ControlZ]

OK, as I sort of expected, no definitive answer. I am not interested in test data, etc. I was just curious to know if "cheap" memory stands up to more expensive memory. I read somewhere that cheap memory for example may be rated as 1 GB, but actually performs at a much slower speed.

[jdMorgan]

1GB is the size, not the speed.

In case you missed it:

If you want to actually analyze the quality factor, then determine the memory chip manufacturer by reading the markings on the devices on the module, and look them up. By referring to the manufacturers' data sheets on the parts, you can get information on MTBF, and compare detailed timing specs to be sure that a particular module manufacturer is not 'stretching' the devices' capabilities.

If a chip says that it's a certain size and runs at a certain speed, then it is and does. If the memory chip speed rating disagrees with that stated by the module manufacturer, then use the slower rating.

Jim