What DDR 2700

[Aganack1]

I'm building a new system this week and was wondering what type of DDR 2700 you would suggest, and what the difference is?

 

[MikeMike]

just go with crucial which can be found cheap sometimes not sure what the prices are now

differences are minimal with maybe some design differences, and sometimes differt latency (which can normally be changed)

plus crucial can o/c well

 

[Jolyon33]

Nah, I've heard too many people have to return theirs nowadays. Go with Corsair!! They rock!!

 

[everman]

Get samsung chips on samsung pcb from newegg, pretty good stuff so far for me.

 

[Darien]

never had problems with crucial pc2100. upgraded my system a few days ago...now using corsair pc3200. can't go wrong with either company IMO.

 

[formulav8]

CTL Samsung

 

[Tirade]

According to this article it seems Samsung is the best for OC'rs
DDR 2700 shootout!
Any other reviews anyone knows about?
-Tirade-

 

[majewski9]

any name brand will do in my opinion. I like crucial!

 

[CaptnKirk]

NewEggs Samsung is NOT the 'True Samsung", as it's build on a third party PC Platform with Samsung chipsets.
I Haven't located a valid source for the 'True Samsung', they want you to buy local and pay full retail.
Here's the closest I can find to a source SAMSUNG DIRECT

Crucial is still the BEST available RAM, how do you beat Lifetime Warranty ?
I've NEVER heard of anyone having problems with Crucial, unless they blatently screw up when overclocking.

Oh, yeah - as an experiment I replaced Crucial DDR in a system with Samsung DDR - and my performance dropped.
Ovbiously I went back to Crucial, as it actually outperfromed the Samsung on my system.
I no longer buy Samsung - only Crucial, and I have the track record to back it up.

 

[Sappster]

You might have to pay a little more, but I would say go for Mushkin!! They all have heatspreaders on them if that's what you want. You could buy them without heatspreaders, but Mushkin is the place to go. They have the BEST customer service. They sell the very best. You will never get an automated machine, but always get a real person to talk to. They all hand test the memory and are specialized in what they do. I recently purchased the DDR 3200 with the heatspreader for dirt cheap. It's only about $150 for 1 -512 stick. I would say go for Mushkin, but if you don't feel like spending a lot, then go for Crucial.

 

[tbates757]

Crucial is about as stable as it gets, if you want overclockability go Samsung, Corsair, or Mushkin (yes, there are others but I dont feel like getting flamed).

 

[CaptnKirk]

Hmmm, interesting the comment about Mushkin being so good, what with the heatspreaders and all.

Take a peek in the lower right hand corner of MY MEMORY STORE,
which is an "Authorized Samsung" product dealer/source.
Isn't that a familiar name?

What do you want to bet that your 'Terrific Mushkin' is really another repackage of Samsung
- only this time with the heatspreaders hiding the Samsung chipsets?

 

[Ilmater]

Samsung (ONLY true Samsung chips on a true Samsung PCB)
Corsair XMS
Mushkin
Crucial
Kingston
TwinMos
Geil (fairly new and not completely proven, but I've had great luck with them)
OCZ (used to have some problems, but seem to have completely redeemed themselves)

In order, those are my picks from best to worst RAM. The Samsung is sometimes hard to be completely sure of, so I usually get Corsair (NOT the ValueRAM). While OCZ is last, they could quickly move to first as long as I continue to hear nothing but good things about them. Their prices are good, and as long as the quality is consistent (and so far, as long as they've been under new ownership, it has) they'll become the RAM of choice. From Samsung to Kingston, though, you really cannot go wrong. All are really good quality.

The only reason to buy PC3200 RAM (though not a bad reason) is so that you are pretty much assured of running at PC2700 with the most aggressive timings. However, most of these (most definitely the Samsung, Corsair XMS, and Mushkin) will be able to run at PC2700 with the most aggressive timings, especially if you're willing to raise the DRAM voltage a little. The other reason to either buy high-quality PC2700 or PC3200 is that running it at PC2700 with the most aggressive timings won't stress the RAM as much. If the stick is stressed and runs hot, it could lead to possible instabilities and a shorter life of the stick.

I would recommend buying two sticks if you don't mind the extra cost. If you get any board with dual-channel DDR in the future, you'll need two identical sticks to take advantage of it. Also, you'll never need more than 512MB of RAM unless you do a lot of video encoding or CAD work.

Edit: Let me rephrase my first sentence. Those aren't my picks from best to worst, they're my pick from best to worst of brands I would actually buy from. I don't think there is any other brand I'd consider buying. Don't buy generic as you could get burned. I don't mean to say that Geil or OCZ are the worst sticks of RAM out there, they're just not at the top of my list as the RAM that I'd buy. However, for my system, I just purchased 2 sticks of Geil 256MB PC3500 Ultra Platinum from Directron (here for $93) and I'm very satisfied. Just make sure you check the RAM when you get it, because they shipped me one stick of Platinum and one stick of Ultra Platinum; not much difference, but it's not what I paid for either.

 

[Ilmater]

Originally posted by: CaptnKirk
Oh, yeah - as an experiment I replaced Crucial DDR in a system with Samsung DDR - and my performance dropped. Obviously I went back to Crucial, as it actually outperformed the Samsung on my system. I no longer buy Samsung - only Crucial, and I have the track record to back it up.

I really hate to say this, but you're out of your mind. You can disagree if you want, but any performance difference you saw was in your head, period. Two sticks of RAM, set to the same speeds and timings, will perform exactly the same, within some degree of error. That's a fact. The only way they wouldn't would be if one of the sticks had serious problems. You could buy $10 generic RAM, and as long as it ran at the same speeds and timings of a $200 stick of RAM, they would perform the same. If your claim is that Samsung didn't overclock as well, then you could very well be correct (though that is not the norm, I can assure you).

 

[Zugzwang152]

newegg had a sale on kingston...40 bucks cheaper than the comparable crucial. might wanna check it out.

 

[CaptnKirk]

Excuse me, but when you run a test and the test reports a LOWER SCORE, and when you return to the other configuration and your score returns to the higher level - that may NOT be an anomolie. And to continue - if you put another set of the 'BRAND' into the system, and again the score drops, and when you return to the primary configuration you have something other than an unverified anomolie - you have what is a verified re-test. Now for your information that is NOT
" Out of your mind", you have a component that is NOT performing to the capability ot the system you are comparing it to.
Now if the higer scores are with Crucial, and the LOWER scores are with the Samsung, MY SYSTEM IS in fact
not performing as well with the Samsung as it did with the Crucial.

Thats 4 tests, 2 with higher scores, and 2 with lower scores.
If you think you're so smart - go figue it out.
The Samsung DID NOT perform as well in MY system.
It may perform well in some other system configuration, but definitly NOT in this one.

 

[blackhawk]

Unless you're high end overclocking/tweaking and have lots of money, go with micron.

Haven't heard anything about lots of people returning it and they offer free shipping within the USA plus lifetime warranty. I run 10 sticks in different computers and never had one fail plus they'll replace used sticks if you buy off the boards.

Great value and quality.

 

[Brian48]

Originally posted by: CaptnKirk
NewEggs Samsung is NOT the 'True Samsung", as it's build on a third party PC Platform with Samsung chipsets.
I Haven't located a valid source for the 'True Samsung', they want you to buy local and pay full retail.
Here's the closest I can find to a source SAMSUNG DIRECT

Is this something recent? I just bought a stick of 512mb Samsung PC2700 from newegg not long ago. It has Samsung RAM modules on a Samsung-made PCB (Brand name imprinted right where it should be). Supposedly, this is the 6 layer variety, but I can't confirm that right now 'cause it's already in the system.

 

[CaptnKirk]

This appears to be a bit 'Hit & Miss' on the Samsung. Some get 'True" as you did
others are getting the Samsung Chipset on a 'generic' PC board, not sure of the manufacturer.
A clue should be in the warranty, as Samsung 'True' would be Lifetime, generic would be 1 year.

 

[Ilmater]

Originally posted by: CaptnKirk
Thats 4 tests, 2 with higher scores, and 2 with lower scores.
If you think you're so smart - go figue it out.

OK, since you've now been kind enough (were you giving Samsung the benefit of the doubt?) to benchmark two different Samsung sticks (as opposed to the one you originally claimed to have tried), you find that your performance, once again, varies. There are plenty of things that could explain this. First off, what were those Samsung sticks rated at (CAS Latency, etc.)? What were the Crucial sticks rated at? Did you make sure to set the same timings with the Samsung that you did with the Crucial? How different were the benchmark scores? Were the Samsung sticks "true" Samsung chips on "true" Samsung PCBs? Were they current PC2700 sticks or were they made back when DDR333 didn't have an actual JEDEC standard? Is there some kind of problem with Samsung RAM and your board? Even the best sticks of RAM from the best companies sometimes have problems with certain boards.

It's this simple: as long as there isn't something flat out wrong with the RAM, then they will run at nearly the same level (again, within a very small margin of error). For example, if both sticks are running at DDR333, then they are sending two chunks of data every clock cycle (double data rate = DDR, I'm sure you all know that). Also, if they're set at the same timings, then they're finding data that is stored within them at the same rates. Therefore, it's simply impossible that - once again, barring any errors within the RAM itself - that they are running differently.

I don't say all of this just to pick on Captain Kirk. I think it's important for people that come here and request suggestions for RAM (and if they'd use the search function, we wouldn't have to go through this all the time). Hey, I've seen people before claim to know what they were talking about but actually having no idea, and frankly, Captain Kirk, I just don't know you or know the answers to all of those questions above.

Quick story: I once stumbled upon a review of three sticks of RAM: Corsair XMS, Samsung, and OCZ. This guy was claiming that one stick beat the other because it posted frames per second that differed by thousandths of a frame!!! And this was on a hardware review site. When I complained (and honestly, just made fun of that ridiculous review), he came in here and said that I should be nicer because he just reviewed hardware in his spare time, and it was a hobby of his. I told him the same thing I'm saying here: when you say that kind of stuff, people that don't know any better will think that you actually know what you're talking about. For the record, the highest difference in performance between the three sticks of RAM was within 1%.

 

[CaptnKirk]

Well let's see where to start.
Aerospace Industry having worked on some rather sophisticated computer systems.

I think I have a pretty good handle on what constitutes a drop in perfromance vs. holding parameters, and my tests dropped with these
2 particular sticks of RAM - 1 a 512 'True', the other a 256 'True' - in both cases they, and the Crucial are JEDEC certified PC2700, DDR333.
Perhaps what puzzles me the most is why the Samsung PC2700 banked 1=516 2=256 (768 Total) also scored LOWER than some Crucial
PC2100, DDR266 which was 3 banks of 256MB. I have absolutely no explanation for that.
I run my RAM VERY agressive, with CAS 2, 4 Bank Interleaf, and I boost the voltage.
Maybe my computer just didn't like these particular Korean parts. Will I try Samsung again? Maybe, but not in this system.
I'll build another system based on PC2700, DDR333 and see what that does - I can always move my Crucial to re-validate results.

PC3200, DDR400 - I'll stand by the JEDEC on this: NOT CERTIFIED, use at your own risk, DDR333 is where the certification ends.