Need advice on RAM
Need some advice on RAM. I have a GA-ma78lmt-s2 motherboard and an AMD Anthlon quad core processor and a GTS 450, but I'm upgrading to a GTX 660 very soon. Anyway my primary focus is gaming with emphasis on modding games. I'm looking to upgrade to 8gb. I just want to know the difference between models/specs and stuff. What do you recommend? Feedback is greatly appreciated. Thanks
You didn't list your exact processor, so I can't be 100% sure it will work, but I would get CT2KIT51264BA1339. It is 1333 which should work on the board and I am pretty sure all the Athlon quad cores will recognize 1333 RAM.
The speed of the RAM (in Mhz) makes little to no difference higher than 1333 and your board only goes that high anyway, AFAIK, so This set of RAM would be good in the forseeable future.
The maker of the part number I mentioned (Micron) usually has prices on the high end, but its worth it because their stuff works right the first time pretty much every single time.
They sell the chips that fail their internal testing to other companies and that is where generic brand RAM comes from. The chips included on low priced RAM usually have failed QA from a major manufacturer, so I wouldn't mess with those.
Kingston is another high quality maker, with G.Skill falling a bit behind in reliability and Corsair farther still. I would avoid other makers than those 3 and Crucial (Micron) without even considering them.
As far as the difference between models goes, generally speaking higher Mhz is better and all the other numbers the lower the better. These will usually be written in the form of 9-9-9-24 or so or the RAM will just have a CL number listed (the first one of those dashed ones) as with CL9.
You can only compare the latter performance numbers within the same category of Mhz. So 1333 CL8 is better than 1333 CL9 in a theoretical sense. However, if you can't get your computer to recognize the CL8 RAM or you have to spend a lot of time messing around with configuration to get it to work right then you may lose whatever tiny advantage you gained by using the CL8.
A tiny increase in RAM speed isn't worth a major increase in RL headaches.
As mentioned before, this can only be compared within the same Mhz speed. CL9 RAM at 1333 is not necessarily better than CL 10 RAM at 1600, for instance, or CL 8 RAM at 1066.
The 1600/1333/1066/800/667/etc numbers refer to how quickly a "cycle" passes in RL. The larger the number the less RL time it takes to complete a cycle. The CL number refers to how many cycles it takes to do one thing, to essentially send a 1 or 0 from the processor to the RAM or to take a 0 or 1 from it.
So basically if a cycle happens more quickly but it also takes more cycles to do the same thing, then you don't really gain anything. Similarly, if it takes fewer cycles to do something and those cycles take longer to complete you also don't come out ahead.
If you were comparing 1333 RAM with CL9 and 1600 RAM with CL9, the 1600 would theoretically be better because it takes the same number of cycles to do the thing it needs (9) and those cycles happen more quickly.
The best way to compare widely different Mhz/CL numbers at the same time is to divide the Mhz by the CL for each RAM and then compare that total.
1066 Mhz / CL8 = 133.25
1333 Mhz / CL9 = 148.11
1600 Mhz / CL10 = 160.00
1333 Mhz / CL8 = 166.625
1600 Mhz / CL9 = 177.77
1600 Mhz / CL11 = 145.45
These numbers at least tell you a little bit about what you are looking at, but the problem is that they only really matter when RAM is the computer's bottleneck which is almost never the case.
If you are gaming, for instance, going from 1333 RAM to 2133 RAM (regardless of CL) will pretty much always get you less than 1 FPS increase in the game, because RAM is pretty much never the bottleneck in games.
About the only thing for which RAM is the bottleneck is archiving. If you sit there and use Winzip all the time, then faster RAM will benefit you somewhat. You might see as much as a 7% time drop for a Winzip job if you went from the before mentioned 1333 RAM to 2133 RAM (with average CLs for the speeds).
Most people spend less than 1% of their time on the computer RAM bottlenecked and the increase will only ease the pain by 7% in that 1% of the time, so that should probably account for about 0.07% of your decision-making.
Anyway, I hope you now have some kind of idea about how worthless it is to really pay a whole lot of attention to the RAM specs outside of the company that makes it.
tl;dr; The maker of the RAM is the only thing worthy of consideration.
Max supported ram is 1333, but I am sure you could get 1600 to work via OCing the mobo/cpu.
Raddinn's advice is sound. The only thing I would add is that faster memory has diminishing returns in performance. So while there is a decent bump in speed moving from DDR3 800 to 1066 and to 1333, there is very little real-world difference moving up to 1600 or 1866.
But your mobo will max out at 8GB of 1333 in a 2x4GB configuration, which typically will cost ~$40. In real world tests there is very little practical difference between cl7, 8, 9, or 10, so I would simply get what is affordable. For brands just stick with Corsair or gSkill.
except that Kingston and crucial typically have much higher dissatisfaction rates from customers on their modules than either gskill or corsair...
It's a lot easier to have a lower RMA when nobody purchases your product to begin with. As a percentage of RMAs to sales Corsair and gskill are cheap safe bets.
Its not easier to have a lower RMA -rate- when you sell fewer units. It may be easier to have a lower RMA -total- when that happens, but not a lower RMA -rate-.
Also, I don't know what satisfaction sites you are reading from, but satisfaction with Crucial RAM that I suggest people get and they do get it is at 100%. Considering I do huge numbers of new build threads and I exclusively suggest Crucial RAM, that's quite a lot of successes without a single failure.
Add to that, I pretty much never see anyone in here with a RAM problem that traces back to Crucial RAM. It only ever happened one time and only then because they had RAM not compatible with their board.
On the other hand, like 9 out of every 10 people that come in here with problems that trace back to RAM are using either G.Skill or Corsair RAM.
The numbers don't really appear to be lying to me.
for customer satisfaction I am looking at sales sites that allow for customer reviews such as newegg, amazon, and ncix, all three of which have several more reviews for gskill and crucial, while generally maintaining a more positive review ranking. Considering people are much more likely to give a review when something is bad than good, I would consider this with a certain amount of weight. It is not exactly a consumer reports review, but still...
Add to that my own personal use of both Corsair and Crucial (never used gskill in my own build), and I have only once had a bad stick of corsair, while I have had 3-4 bad sticks of crucial. I am only one person, I understand that, but it does seem to fit with the reviews I have seen elsewhere.
The big problem with Corsair, and perhaps this is what you have run into, is that when they have a manufacturing failure, they fail big-time. It happens, and there are some Corsair product lines that have obviously been plagued by such issues if you look at reviews. But in general they are few and far between, and if you do read product reviews before purchase you can generally weed them out before purchase.
CTKIT51264BA1339 = 5 eggs out of 57 reviews
CT2KIT51264BA1339 = 5 eggs out of 62 reviews
CT3KIT51264BA1339 = 5 eggs out of 5 reviews
Exact same stick just change the number in the package. Avgs 5 stars regardless of how many are in the kit.
I just typed CT2KIT51264BA1339 into Newegg and the numbers came right up. I didn't look for anything specific I just typed in the one that I always suggest and which I have memorized the part number for.
I just typed Crucial Ram into Newegg and it came up with 63 different model lines that are all rated at 5 eggs. There are 5 more on the list that are 5 models that are out of stock and probably never going to be restocked as well. 15 more are rated at 4/5 eggs. Exactly 1 is rated at 1 star and that only has 1 review for it. Two have 3/5 and they both have 5 and 1 ratings.
Total that up and that is 90% at 5 stars and 96% that is 4 stars or better.
Now some of the Corsair models have much more actual ratings, but that has more to do with marketing than quality.
Anyway, we could argue about this stuff all day and not get anywhere.
I would rather take meaning from actual returns figures than reviews. If Corsair avgs a 5x higher rate of RMAing than Crucial/Kingston in the raw data and all the people that come in with RAM problems have Corsair RAM then I get all the meaning I need from that.
It could just be that all the people that come in with RAM problems bought the bad lines of Corsair and indeed that is a strong possibility, but with with Corsair's worst (north of 10%) being like 10x the RMA rate of Crucial's worst (south of 25) I don't feel bad doing a blanket recommend for Crucial and telling people not to bother with hit or miss brands.
I am a solid supporter of people out there doing their research, but I am also a solid supporter of just telling them how to shortcut the process and buy the best brand so they don't have to research too.
Which brands of hardware have the absolute lowest RMA rates out of every brand out there, therefore, do mean a lot to me and I make certain that I know what those are to the best of my ability.
That really isn't Corsair for any of their product lines. They are a top performer, no doubt, but not the top performer. They have quite a ways to go before they get to that point.