Which DIMMs have a better chance of running well with tighter timings?

[Chicken76]

This one: http://www.crucial.com/usa/en/memory/blt4g3d1608et3lx0
Or this one: http://www.corsair.com/en/vengeance-8gb-dual-channel-ddr3-memory-kit-cmz8gx3m2x1600c7r

The Crucial is running 8-8-8-24 @1.35v so is it possible to overvolt it to 1.5v and run it at 7-7-7-21 ?
The Corsair is running at 7-8-8-24. Would it have a better chance of running at 7-7-7 timings without significant overvolting?

 

[bryanl]

The ones made with the fastest chips, something that can be determined only by removing the heatsinks (do not) and reading the markings printed on the chips, provided they're marked with standard part numbers (often they are not).

 

[john3850]

I am running 2x8 crucial vlp sticks.
These crucial sticks came with two different xmp profiles one is 1.35v and the second at 1.50v.
I believe it defaults to the 1.50v to make it more compatable with earlier mb.
Most of the reviews used 1.60 - 165v for the max oc on the crucial you listed.
The reviews were after the max speed rather then tighter timings.

 

[Magic Carpet]

This one: http://www.crucial.com/usa/en/memory/blt4g3d1608et3lx0
Or this one: http://www.corsair.com/en/vengeance-8gb-dual-channel-ddr3-memory-kit-cmz8gx3m2x1600c7r

The Crucial is running 8-8-8-24 @1.35v so is it possible to overvolt it to 1.5v and run it at 7-7-7-21 ?
The Corsair is running at 7-8-8-24. Would it have a better chance of running at 7-7-7 timings without significant overvolting?

You want Crucial. Either Crucial Tactical 1600 1.35v LP, or 1600 Sport VLP 1.35v. Your board will default them running at 1.5v, but boy there is lots of OC performance waiting to be unleashed. Depending on your board how high you can go, but 1666 Mhz at 7-7-7-20 1T, should be no problem @ 1.5v.

I am also running these low profile tactical in 4 x 8GB configuration. Top quality ram.

Found one of my old Thuban tests:

1333Mhz CL9 = 124 seconds to encode

1333Mhz CL6 = 107 seconds to encode

 

[BonzaiDuck]

You want Crucial. Either Crucial Tactical 1600 1.35v LP, or 1600 Sport VLP 1.35v. Your board will default them running at 1.5v, but boy there is lots of OC performance waiting to be unleashed. Depending on your board how high you can go, but 1666 Mhz at 7-7-7-20 1T, should be no problem @ 1.5v.

I am also running these low profile tactical in 4 x 8GB configuration. Top quality ram.

Found one of my old Thuban tests:

That makes all the sense in the world, really. I'd used Crucial and Corsair for DDR2 builds, then G.SKILL for DDR2 and DDR3. There is just a simple trade-off between voltage, speed and timings.

These new Crucial modules are more recent in the choices of DDR3. It seems just barely a year or two ago that those Samsung 1.35V modules were the rage. So this is new, improved DDR3 with timings like 8-8-8-24 @ 1600. You could probably run them at 9-9-9 and 1866 with the same voltage, but there's plenty of voltage headroom there. There should be choices within a voltage-spec and warranty range.

The problem with all this is in the testing for RAM configurations of 16GB and more. You can run the HCI memtest program in windows, which allows you to continue using the computer during tests, but I prefer the self-booting CD. Figure your system could be offline for the good part of a day or so with the latter approach. And if adjustments still need to be made, you have to start over.

Why I say that: the prices of higher-spec'd RAM aren't that different from the 1600 Mhz offerings, so it's just as easy to buy what you want versus overclocking the 1600's. If you already have DDR3-1600, you can choose to overclock -- and test-fail, test-fail and test-pass. I've done it both ways.

At least if you buy the RAM at the speed and timings you want, you'd only need to test it once at its rated specs.

EDIT: One more thing. Using XMP profiles may do more to reduce effort and get an optimum performance, since it should affect the "advanced timings." There are indeed a couple of advanced timings that affect performance substantially, but then you add another complication to testing, and this would be multiplied by further tweaking those settings manually -- "test-fail, test-fail . . . pass."

 

[SaltyNuts]

You want Crucial. Either Crucial Tactical 1600 1.35v LP, or 1600 Sport VLP 1.35v. Your board will default them running at 1.5v, but boy there is lots of OC performance waiting to be unleashed. Depending on your board how high you can go, but 1666 Mhz at 7-7-7-20 1T, should be no problem @ 1.5v.

I am also running these low profile tactical in 4 x 8GB configuration. Top quality ram.

Found one of my old Thuban tests:

Magic Carpet - I'm building my own PC for the first time in years, so I'm a bit of a newb these days, including when it comes to memory. Is a 1600 rated chip from Crucial as overclockable as a 2400 chip from some other company? I just ordered a Gigabyte GA-Z97X that supports speeds up to 3000 supposedly, and a Intel Pentium G3258 Anniversary Edition, and plan to overclock the crap out of it. I take it the 1600 chips would not do me good? Which memory in particular would you recommend? I'll probably get 8 gigs.

Thanks!

 

[Chicken76]

Thanks guys. Ordered 2 Crucial modules. Hope they're good as stock is very low.

 

[BonzaiDuck]

Magic Carpet - I'm building my own PC for the first time in years, so I'm a bit of a newb these days, including when it comes to memory. Is a 1600 rated chip from Crucial as overclockable as a 2400 chip from some other company? I just ordered a Gigabyte GA-Z97X that supports speeds up to 3000 supposedly, and a Intel Pentium G3258 Anniversary Edition, and plan to overclock the crap out of it. I take it the 1600 chips would not do me good? Which memory in particular would you recommend? I'll probably get 8 gigs.

Thanks!

Depends on what you mean there. What is the reference "starting point?" Do you mean "over-clocking the 2400-rated RAM to something faster?" Or do you mean a comparison of a rated 1600 and rated 2400 starting from the same (lowest) speed (1600) and going higher from there?

If you look at the offerings from either one RAM-maker or many given the current technology for DDR3, you'll see a pattern in three dimensions involving the voltage, the speed and the timings. The better or more promising RAM departs from the collective profile: you should be able to get the rated (highest) spec speed at lower voltage and tighter timings.

The higher the rated speed, the more likely a higher spec voltage, the more likely the timings will be looser. You can buy 1600's that are immensely over-clockable, and spend time and trouble tweaking and testing. Or you can look for the speed you want, accept the voltage and timings as a given, and avoid the trouble. IN FACT -- you could even buy RAM rated at a higher voltage, speed and looser timings, and down-clock it to either use (a) a lower voltage and lower speed, (b) lower or same voltage, lower speed and tighter timings, or ( . . ) other possibilities.

The recommended Crucials mentioned in this thread offer some great possibilities, but running them at higher-than-spec speed and timings or any combination of voltage, speed and timings other than spec means more testing and troubleshooting.

Heck. I might even decide to just spring for a pair myself, and see what they can do over my 1.5V/DDR3-1866/9-9-9-24 G.SKILL GZH's.

ADDENDUM: Lemme say somethin' else. Your RAM voltage spec from the manufacturer should have a range "within warranty." If you stay within that range, you should be "OK." Anything else -- you can shorten the life of the RAM, cause yourself headaches troubleshooting the RAM failures -- and more misery. If the Crucials will run at 1.35V, you might be able to boost the speed and loosen the timings. If they are safe to run within a range 1.5 >= X > 1.35V, there are even more possibilities. But in ALL cases, the first thing you want to do when you buy a set of RAM sticks is to test them at all-stock settings -- thoroughly. If you run them out of spec or beyond spec, you'll need to do double or more testing -- measured in time, sweat and trouble.

 

[SaltyNuts]

Thanks a ton BonzaiDuck. So I think what you are saying is that a 1600 chip with low voltage and tights settings might well overclock better than a 1800 chip with higher voltage and looser settings. If that is so is the thing to do just go with Crucial 1600 chips mentioned in this thread, cause they are likely good or better than even 2400 rated chips (which just use higher voltage and looser settings) anyways?

Thanks!

 

[SaltyNuts]

By the way, I ordered a Intel Pentium G3258 Anniversary edition, which I want to overclock it as much as possible (4.5ghz or higher), and I don't want the memory holding me back. Maybe its not that big a deal because if I max out the memory speed the chip's clock is unlocked so I can just raise the multiplier?

Thanks!

 

[BonzaiDuck]

Thanks a ton BonzaiDuck. So I think what you are saying is that a 1600 chip with low voltage and tights settings might well overclock better than a 1800 chip with higher voltage and looser settings. If that is so is the thing to do just go with Crucial 1600 chips mentioned in this thread, cause they are likely good or better than even 2400 rated chips (which just use higher voltage and looser settings) anyways?

Thanks!

That isn't quite what I said. Looser timings and higher speed generally trumps lower speed and tighter timings -- despite the fact that tighter timings yield more performance at any GIVEN speed.

I'm also saying that the Crucials look to be ripe for overclocking. Again, you can probably loosen the timings, set the speed at 1866, and leave the voltage alone. You might have to add a few millivolts to the IMC's voltage setting (on my Z68 board, it's called VCCIO).

You will WANT to put these overclock tweaks to the fire with extensive testing. Extensive testing with 16GB or 2x8GB of RAM takes time. I think it took 24 hours for me to get 400% coverage with HCI-Memtest-64, or four iterations of all the test sequence. All I wanted to do was to validate a change of the command rate from 2T to 1T. ANY memory failure due to the wrong settings, insufficient voltage, etc. has potential to corrupt your hard disk and multiply your misery.

That's why I recommend looking for the memory speed you want in a RAM kit spec, and then searching for the lowest voltage and tightest timings. It will save you time and trouble. If you want to spend the time and trouble OC'ing the set of Crucial Tactical 1600's, I wouldn't discourage it, but you need to EMBRACE that time and trouble as something you want to endure.

 

[SaltyNuts]

Thanks BonzaiDuck. Let's say I wanted the best quality memory chips available that will definitely handle the higher speeds without extensive testing. Do you have any suggestions? I'm willing to pay extra for a better chip that will definately handle the higher speeds, but don't want to pay extra if what I buy is in reality just the Crucial 1600 chips that are rated higher because they can typically handle the higher speeds naturally - I'd just buy the crucial 1600 chips if that is the case.

Thanks!

 

[BonzaiDuck]

Thanks BonzaiDuck. Let's say I wanted the best quality memory chips available that will definitely handle the higher speeds without extensive testing. Do you have any suggestions? I'm willing to pay extra for a better chip that will definately handle the higher speeds, but don't want to pay extra if what I buy is in reality just the Crucial 1600 chips that are rated higher because they can typically handle the higher speeds naturally - I'd just buy the crucial 1600 chips if that is the case.

Thanks!

Like stains in the toilet bowl, there's always a residual bias based on an unpleasant experience. I'd had troubles with Crucial Ballistix and Tracer DDR2 kits. They weren't IMMEDIATE troubles, but I thought that Crucial -- in an effort to address the overclocking market, had "over-spec'd" their modules. Too many of them blinked out on me within 6 months to a year. And I researched the warranty voltage RANGE on those -- which I stayed within.

It was the "black parts" that interested folks across RAM-makers. I think they had been Micron chips. I think Crucial and Micron are part of the same company. But no problem -- the "good" black-parts are often used by many RAM-makers.

Few years back, a certain low-profile, no-frills (or heatsink bling) model of Samsung modules generated rave comment here on the forums. I'm not so sure you can get those anymore: they were rated at 1.35V and DDR3-1600 -- whatever the timings were.

I'd used Crucial, OCZ, Corsair. I started using G.SKILL around 2011, and haven't looked back since. They aren't always the LOWEST in voltage specs. I've seldom bought a kit that required an RMA. However -- when I did -- I discovered their RMA process and tech-support very responsive. If you have a question about tweaking the RAM -- they'll answer it, and they'll do it via e-mail.

I think for timings, G.SKILL can be just as stellar as the rest. A lot of their good RAM can be found in kits/modules rated at 1.5V as opposed to 1.6 or 1.65 -- for DDR3.

Just imagine a 3-dimensional graph of voltage, CAS latency and speed. If you lower the volts below spec, you'll likely need to loosen the timings, or reduce the speed. Sometimes, your own testing may yield surprises: I still would like to find out if my G.SKILL GZH modules will run at rated speed and timings at voltage 1.4V < X < 1.5V. I'd have to test it.

If you raise the speed, you either need to loosen the timings, raise the voltage, or both.

The easiest thing to do is get the RAM with a spec matching your target operating speed. Get it with the tightest latencies and the lowest voltage among the available alternatives. Then, you only need to test once: to make sure you didn't get a defective stick so you can RMA. Beyond that, the more you tweak, the more you have to test.

I still think those Crucial 1600's look good. But I'd also wonder if you can get them to match a set of DDR3-2133's by applying looser timings. Maybe? Maybe you have to raise the voltage? You'd have to test them. Even if you wanted to OC the RAM, you'd do better to find a DDR3-1866 kit to shoot for a 2133 target. Why not just settle on a speed, follow the principles I outlined, and save yourself the trouble of too much testing?

Further -- you can (and I have done it myself) -- purchase RAM rated at a higher speed-spec and underclock it with tighter timings, if you think it will behave better with your given hardware. These days, that's not likely to be so much a positive option.