G.Skill Releases TridentX DDR3 Kits for Ivy Bridge

[jsc]

The only real way ram can gain "speed" is cap out to 1600 and get the lowest latency ram you can find. Many CL 7 kits can be purchased at that speed... these kits at CL10 and CL11 are a tough sell for anyone who knows what they are doing building a computer.

My first Core2 system used an E6600 and DDR2-1000 RAM (5-5-5-15-T2). I didn't want the RAM to be the problem with my overclock.

Pushed the FSB from 266 MHz to 367 MHz (3.3 GHz). Then I worked the timing down to a stable 3-3-3-7-1T. Increased memory i/o 7%.
Overall increase in system performance? Almost zero.
Amount of time I spent?

 

[RealBeast]

[citation][nom]Halcyon[/nom]Well, they do tend to look nice and that seems to be their usual purpose.[/citation]
The look is to keep you from thinking about those terrible timings and the 1.65V needed.

 

[kristoffe]

trolls

 

[dreadlokz]

[citation][nom]frombehind[/nom]funny story about that ~3 GhZ ram.... Ivy bridge can only interface with its ram up to 1600 MhZ. Anything above this can be considered wasteful. While faster ram CAN impact some synthetic benchmarks... FPS in games, nor any processing speeds in applications will not increase 1 bit past 1600. The only real way ram can gain "speed" is cap out to 1600 and get the lowest latency ram you can find. Many CL 7 kits can be purchased at that speed... these kits at CL10 and CL11 are a tough sell for anyone who knows what they are doing building a computer.[/citation]

High speed memory is not for average users! Like u said... 1600 with low CL are the best for most!
But with these kits you can do some good stuff like this:

http://www.youtube.com/watch?v=vAdRsQJBEBE

not for everyone... but some ppl could rly use all that speeeeeeed! =D

 

[JOSHSKORN]

and I'm still on the upper end of DDR2 This make me a sad panda.

Time to start praying my computer breaks so I can upgrade today.

 

[Guest]

This is how faster memory can really come in handy:
http://thessdreview.com/our-reviews/romex-fancycache-review-ssd-performance-at-13gbs-and-765000-iops-in-60-seconds-flat/

 

[hellfire24]

heat spreaders are useless.go with low profile ram if want to fit a good air cooler without issues.

 

[Uberragen21]

[citation][nom]amk-aka-Phantom[/nom]1.65V? Where's the promised voltage drop from Sandy Bridge's 1.5V RAM?[/citation]
It's because the RAM is at such a high factory clock rate to begin with, it needs the higher voltage to remain stable. It is after all still DDR3, not DDR5. Higher clock speeds without new technology will result in a higher power requirement.

 

[dalauder]

[citation][nom]frombehind[/nom]funny story about that ~3 GhZ ram.... Ivy bridge can only interface with its ram up to 1600 MhZ. Anything above this can be considered wasteful. While faster ram CAN impact some synthetic benchmarks... FPS in games, nor any processing speeds in applications will not increase 1 bit past 1600. The only real way ram can gain "speed" is cap out to 1600 and get the lowest latency ram you can find. Many CL 7 kits can be purchased at that speed... these kits at CL10 and CL11 are a tough sell for anyone who knows what they are doing building a computer.[/citation]3000MHz @CL11 will blow the pants off 1600CL7 considering 1866CL9 beats it: http://www.xbitlabs.com/articles/memory/display/sandy-bridge-ddr3_7.html#sect0

Now there's no reasonable use for this RAM, but I doubt it's for reasonable people. If you've got a heavily OC'd rig w/ a PCI-e SSD or raid 0 SATA III drives, this could provide a tangible benefit for batch processing or other things that take an hour total. Still, it's more fun that practical. It's just that I feel like you're criticizing a Ferrari when an Intregra gets the job done so well--they're aiming at different markets entirely.

 

[halcyon]

[citation][nom]dalauder[/nom]3000MHz @CL11 will blow the pants off 1600CL7 considering 1866CL9 beats it: http://www.xbitlabs.com/articles/m [...] html#sect0Now there's no reasonable use for this RAM, but I doubt it's for reasonable people. If you've got a heavily OC'd rig w/ a PCI-e SSD or raid 0 SATA III drives, this could provide a tangible benefit for batch processing or other things that take an hour total. Still, it's more fun that practical. It's just that I feel like you're criticizing a Ferrari when an Intregra gets the job done so well--they're aiming at different markets entirely.[/citation]
Nice, thanks. While I'm using a PCI-E SSD I just don't have any reason to OC' my i5 2500K beyond the 4.2Ghz its sitting at. Yes, I play Metro at high resolution but the visuals are already past where they need to be for slick immersive gaming. My RAM is set @ that slow 1866 CL9 (just Corsair XMS3) and performance seems good enough.

 

[xingmaodr]

Good article

 

[aviral]

Awesome.

 

[sykozis]

There's at least 4 people that can't comprehend the English language.

The IMC on Sandy Bridge and Ivy Bridge were both designed for a maximum memory voltage of 1.5V, according to Intel's spec. So, again, why do we keep seeing these memory kits made for Sandy Bridge and now Ivy Bridge using voltages of 1.65v when it's beyond the voltage spec for the intended processors? Intel even states that these voltages are dangerous for the memory controller.

 

[f-14]

According to the manufacturer, the TridentX memory ships with a removable top fin heat spreader to provide more flexibility when enthusiasts want to use the memory with other memory cooling systems.

didn't they mean big aftermarket cpu coolers. 2011 socket cpu's don't come with their own coolers any more.
and we all know how memory heights and cpu cooler clearances work when slots get blocked off or stacked over by cpu coolers vs. the amount of us that throw in memory coolers over the slots when the cpu cooler is blocking 1-2 slots.
i think g skill needs to hire a new marketing team.

 

[halcyon]

[citation][nom]sykozis[/nom]There's at least 4 people that can't comprehend the English language.The IMC on Sandy Bridge and Ivy Bridge were both designed for a maximum memory voltage of 1.5V, according to Intel's spec. So, again, why do we keep seeing these memory kits made for Sandy Bridge and now Ivy Bridge using voltages of 1.65v when it's beyond the voltage spec for the intended processors? Intel even states that these voltages are dangerous for the memory controller.[/citation]
So, since I've had my RAM at its stated voltage of 1.65v I should expect my 2500k to fail? Well, that's one way to have a reason for an upgrade.

 

[blazorthon]

[citation][nom]frombehind[/nom]funny story about that ~3 GhZ ram.... Ivy bridge can only interface with its ram up to 1600 MhZ. Anything above this can be considered wasteful. While faster ram CAN impact some synthetic benchmarks... FPS in games, nor any processing speeds in applications will not increase 1 bit past 1600. The only real way ram can gain "speed" is cap out to 1600 and get the lowest latency ram you can find. Many CL 7 kits can be purchased at that speed... these kits at CL10 and CL11 are a tough sell for anyone who knows what they are doing building a computer.[/citation]

Yeah, that's why some games show some benefits going from 1600MHz to 1866MHz (not big and maybe not noticeable, but they are easily measurable). You people complaining about the timings forget that the timings are the latency counted in clock cycles, so higher timings with a high frequency can equal or even be lower latency than lower timings on much lower frequencies. For example, a timing of 11 at 2600MHz is a lot lower latency than a timing of 9 at 1600MHz, in addition to the higher bandwidth.

Some things really benefit from faster RAM. For example, rendering, archiving (lots of WinRAR and such), and folding can benefit from faster RAM greatly. Llano can benefit from faster RAM greatly. I have to assume that with how fast HD 4000 is (it's creeping up on Llano A6 performance), that systems using the HD 4000 IGPs could also benefit from faster RAM for other uses than the above mentioned, or even more so for those uses.

AMD CPUs can benefit from faster RAM more readily than Intel CPUs due to the AMD CPUs having poor memory controllers (Llano and Bulldozer memory controllers have about 25% less bandwidth than Intel's Sandy Bridge controller does with the same number of channels at the same frequency, IE with the same memory kit).

[citation][nom]Halcyon[/nom]So, since I've had my RAM at its stated voltage of 1.65v I should expect my 2500k to fail? Well, that's one way to have a reason for an upgrade.[/citation]

Sandy Bridge should be okay with 1.65v RAM for a while, but it will not last as long as it would with 1.5v RAM. Having only one module per channel when you use the 1.65v RAM helps to alleviate this issue because although the increased voltage is straining the controller, using only two modules (one per channel) instead of four modules (two per channel) helps a little. Basically, using 1.65v shouldn't be too bad so long as you only use two RAM modules. Problems shouldn't really come along unless you have four 1.65v modules on a Sandy Bridge system. I don't know how well Ivy will handle high voltage RAM.

[citation][nom]f-14[/nom]didn't they mean big aftermarket cpu coolers. 2011 socket cpu's don't come with their own coolers any more.and we all know how memory heights and cpu cooler clearances work when slots get blocked off or stacked over by cpu coolers vs. the amount of us that throw in memory coolers over the slots when the cpu cooler is blocking 1-2 slots.i think g skill needs to hire a new marketing team.[/citation]

That is why it is recommended that you use a water cooling system with the LGA 2011 systems instead of an air cooler. If you want an air cooler, then use a motherboard that has eight RAM slots so that a large air cooler doesn't block too many slots. If you want memory like this, regardless of your system, then you're probably more than willing to just use a water cooler anyway.

 

[halcyon]

@ Blaze... I do have 4 x 1.65v modules on my 2500K system. If it fails no biggy (as long as it only takes the CPU, MB, and RAM with it). If it fails I'll be more careful next time.

 

[blazorthon]

[citation][nom]Halcyon[/nom]@ Blaze... I do have 4 x 1.65v modules on my 2500K system. If it fails no biggy (as long as it only takes the CPU, MB, and RAM with it). If it fails I'll be more careful next time.[/citation]

I highly doubt that if the RAM causes a failure, it will take anything more than the RAM and the CPU out. The problem is just that the CPU suffers damage over time (increased wear) from voltages that it wasn't designed to use.

It's not easy to accurately predict how long your system will last, but chances are that it won't last as long as it would if you had fewer RAM modules and/or the modules were being run with a lower voltage. Sandy Bridge's actual RAM voltage *cut-off* is closer to 1.57v or 1.58v, so if you can manage to lower the voltage to about that much without sacrificing stability, then you could extend the lifetime of that system, if you wanted to.

 

[halcyon]

[citation][nom]blazorthon[/nom]I highly doubt that if the RAM causes a failure, it will take anything more than the RAM and the CPU out. The problem is just that the CPU suffers damage over time (increased wear) from voltages that it wasn't designed to use.It's not easy to accurately predict how long your system will last, but chances are that it won't last as long as it would if you had fewer RAM modules and/or the modules were being run with a lower voltage. Sandy Bridge's actual RAM voltage *cut-off* is closer to 1.57v or 1.58v, so if you can manage to lower the voltage to about that much without sacrificing stability, then you could extend the lifetime of that system, if you wanted to.[/citation]I'll definitely give it a shot.

 

[dalauder]

[citation][nom]f-14[/nom]didn't they mean big aftermarket cpu coolers. 2011 socket cpu's don't come with their own coolers any more.and we all know how memory heights and cpu cooler clearances work when slots get blocked off or stacked over by cpu coolers vs. the amount of us that throw in memory coolers over the slots when the cpu cooler is blocking 1-2 slots.i think g skill needs to hire a new marketing team.[/citation]I have trouble fitting my cooler over the Ripjaws X. Of course, I probably shouldn't be using a mATX board, so that's pretty much my fault.

 

[blazorthon]

[citation][nom]Halcyon[/nom]I'll definitely give it a shot.[/citation]

Apparently it's not running RAM at 1.65v that is the problem, now that I've read into it some more:

[citation][nom]Jaquith[/nom]Thanks Thomas another Great Article! Don't like what I see, but I digress. Something's gotta be pooched with the ASUS P8Z77-V Pro BIOS (UEFI) -- hopefully. In the past the ASUS Pro line has been the meat & potatoes for my recommendations, and this is not the only review with similar performance numbers.Voltages, I am going to have a hard time recommending a vCore >1.2Xv, VCCSA and CPU VTT of 1.20v on the IB. I still need to see otherwise. From what I've seen the IB is more 'girlish' with voltages than the SB or SB-E, and there's little point having the fans spinning 'through' the case and creating high dBA with a high vCore. RAM (voltage), it goes back to my feelings that 1.50v DIMM was a bunch of Urban Myths especially since the SB-E and seemingly the IB can handle 1.65v DIMM RAM. Yeah, I noticed the XMP tried to set 1.25v VCCSA, or at least the set is encoded that way. Further, I don't wan to debate the OC until I get my hands on an IB, it should be any day now.Further, either the Engineers were dead wrong on the SB (1.50) or IB (1.65) they're wrong in both instances. I 'get' ultra fast kits (today) >DDR3-2133 e.g. DDR-2400 or faster are 1.65v kits, but only a few months ago IF 'I' recommended SB + 1.65v I'd have 20+ negative comments in the Forum. Seems counter intuitive step in DRAM voltage.Also, I am assuming you're testing the IB ES and I wonder how much of an impact that has in that the CPUID are geared towards the Retail. I remember all of the E5 (ES) problems and drops in performance compared to the Retail sisters. OC observation only, you seemed 'wimpish' with the SB-E compared to the IB - interesting?![/citation]

[citation][nom]Crashman[/nom]Actually, our engineering contacts recommended 1.40 to 1.45V for Sandy, and we only stepped down to 1.35V when a bio* board burned a CPU around 1.40V (apparently due to increased voltage during BIOS initialization). And the 1.65V memory thing...it was all about the difference between the memory controller and memory being less than 0.50V so VCCSA at 1.15V supposedly solved the 1.65V reliability issue.Read my first response, Tom's Hardware DID NOT SELECT ANY of these boards for participation. It's on the manufacturer, ask them The third slot on the MSI board isn't supposed to work with Sandy Bridge processors, due to the CPU-integrated controller supporting a maximum of two devices. Otherwise, go for it![/citation]

My suggestion would probably more or less solve the problem, but this might help too.