G.SKILL Releases Extremely High End DDR4 Memory Kit at 3400MHz

[btarunr]

G.SKILL International Co. Ltd., the world's leading manufacturer of extreme performance memory and solid-state storage, is proud to release the Ripjaws 4 DDR4 3400MHz 16GB (4GBx4) memory kit at CL 16-16-16-36, and a new Ripjaws 4 DDR4 3200MHz 16GB (4GBx4) memory kit with a faster CL15 timing. These extremely high end memory kits also come with awesome Turbulence III memory fans.

The new Ripjaws 4 DDR4 3400MHz 16GB (4GBx4) memory kit is made from hand-picked IC and G.SKILL's highly-selective binning process. It has also been tested for compatibility on the Asus Rampage V Extreme and Gigabyte X99-SOC Champion motherboards, shown in screenshots below.



Established in 1989 by PC hardware enthusiasts, G.SKILL specializes in high performance memory and SSD products, designed for PC gamers and enthusiasts around the world. Combining technical innovation and rock solid quality through our in-house testing lab and talented R&D team, G.SKILL continues to create record-breaking memory for each generation of hardware and hold the no. 1 brand title in overclocking memory.



View at TechPowerUp Main Site

 

[DeNeDe]

benefit only for adobe after effects, 4K/5K/8K video, editing.. hollywood producers and directors. CGI effects.. and all that stuff

 

[happita]

While I applaud RAM manufacturers efforts in advancing the DDR4 tech, I'm a little confused.

Back in the old days of DDR1, 400MHz was pretty much the standard speed for that generation. DDR2, it was 800MHz, DDR3, 1600 (or maybe 1333, not sure). Am I silly to expect that when DDR4 reaches maturity that the standard speed for the platform will eventually be 3200MHz with average timings? Or is that wishful thinking?

 

[Jorge]

While I applaud RAM manufacturers efforts in advancing the DDR4 tech, I'm a little confused.

Back in the old days of DDR1, 400MHz was pretty much the standard speed for that generation. DDR2, it was 800MHz, DDR3, 1600 (or maybe 1333, not sure). Am I silly to expect that when DDR4 reaches maturity that the standard speed for the platform will eventually be 3200MHz with average timings? Or is that wishful thinking?

Quite frankly it doesn't matter.

Most people have never taken the time to educate themselves on DRAM or system performance changes based on DRAM changes. Up until DDR3 came along the CPU/system could generate/process data faster than the DRAM so DDR and DDR2 were system bottlenecks. As such frequency and timings could make a noticeable change in system performance. With the advent of DDR3 the situation started to change as DDR3 running at 1600 MHz. or faster eliminated the DRAM bottleneck that previously existed. As a result even with the fastest CPU powered desktops and DDR3 RAM running at 1600 MHz. there is no tangible system performance gains with increased frequency or faster timings as these aspects become minute changes for a system that is no longer bottlenecked. APU powered systems can gain a small performance boost from RAM up to 2133 MHz. Above that you're just throwing money away.

DDR4 which is actually designed primarily for servers, it a solution to a non-existent problem at this point in time. LV DDR3 can provide essentially the same performance at a considerably lower cost. In addition with DDR4 the topology is different so you can't add to existing DRAM you must replace it all if you desire increased capacity. Naturally those who don't bother to do their homework can easily be duped into believing that they will see a system performance advantage by upgrading to DDR4, when it reality this is not true for CPU power desktop systems running 1600 MHz. or higher RAM frequency and 2133 MHz. for APU powered systems.

 

[hellrazor]

Quite frankly it doesn't matter.

Most people have never taken the time to educate themselves on DRAM or system performance changes based on DRAM changes. Up until DDR3 came along the CPU/system could generate/process data faster than the DRAM so DDR and DDR2 were system bottlenecks. As such frequency and timings could make a noticeable change in system performance. With the advent of DDR3 the situation started to change as DDR3 running at 1600 MHz. or faster eliminated the DRAM bottleneck that previously existed. As a result even with the fastest CPU powered desktops and DDR3 RAM running at 1600 MHz. there is no tangible system performance gains with increased frequency or faster timings as these aspects become minute changes for a system that is no longer bottlenecked. APU powered systems can gain a small performance boost from RAM up to 2133 MHz. Above that you're just throwing money away.

DDR4 which is actually designed primarily for servers, it a solution to a non-existent problem at this point in time. LV DDR3 can provide essentially the same performance at a considerably lower cost. In addition with DDR4 the topology is different so you can't add to existing DRAM you must replace it all if you desire increased capacity. Naturally those who don't bother to do their homework can easily be duped into believing that they will see a system performance advantage by upgrading to DDR4, when it reality this is not true for CPU power desktop systems running 1600 MHz. or higher RAM frequency and 2133 MHz. for APU powered systems.

If RAM gets fast enough there will probably be less cache levels on processors, which would help me (as a programmer) a bit.

 

[Harry]

even with the fastest CPU powered desktops and DDR3 RAM running at 1600 MHz. there is no tangible system performance gains with increased frequency or faster timings as these aspects become minute changes for a system that is no longer bottlenecked. APU powered systems can gain a small performance boost from RAM up to 2133 MHz. Above that you're just throwing money away.lt

The introduction of ddr4 and 2133Mhz+ dram frequencies will allow Intel/Amd to bump up the front-side bus capabilities of their new cpu's. I think if that happens we'll see a significant improve in computer power and the higher cpu bus speed willl be the new bottleneck.

 

[Aquinus]

If RAM gets fast enough there will probably be less cache levels on processors, which would help me (as a programmer) a bit.

Umm, no? You'll never replace cache because cache because it will always have lower latency than physical memory because of the memory hierarchy and how memory works. The fact that cache is essentially next to the CPU core makes latency super low and bandwidth relatively high. So to say that this would result in less cache levels is a little dumb because for it to replace something like L3 cache, latencies on memory would have to be faster than 10ns and right now they barely can respond faster than 50ns and with DDR4's higher latencies, we're seeing that kind of trend remain rather solid. The only cache level I might agree with you about is the L4 on Intel CPUs with Iris Pro, but to exceed that kind of bandwidth (not latency, doesn't get close) you need quad-channel memory. All in all, I think this is wishful thinking and doesn't represent reality.

The introduction of ddr4 and 2133Mhz+ dram frequencies will allow Intel/Amd to bump up the front-side bus capabilities of their new cpu's. I think if that happens we'll see a significant improve in computer power and the higher cpu bus speed willl be the new bottleneck.

Intel ditched the FSB when skt1366 was released 2008. That was 6 years ago... AMD also has been moving the PCI-E root complex to the CPU which means that they very well might not need HyperTransport in the future. All in all, your post is laughable to say the least (no offense).

Quite frankly it doesn't matter.

Most people have never taken the time to educate themselves on DRAM or system performance changes based on DRAM changes. Up until DDR3 came along the CPU/system could generate/process data faster than the DRAM so DDR and DDR2 were system bottlenecks. As such frequency and timings could make a noticeable change in system performance. With the advent of DDR3 the situation started to change as DDR3 running at 1600 MHz. or faster eliminated the DRAM bottleneck that previously existed. As a result even with the fastest CPU powered desktops and DDR3 RAM running at 1600 MHz. there is no tangible system performance gains with increased frequency or faster timings as these aspects become minute changes for a system that is no longer bottlenecked. APU powered systems can gain a small performance boost from RAM up to 2133 MHz. Above that you're just throwing money away.

DDR4 which is actually designed primarily for servers, it a solution to a non-existent problem at this point in time. LV DDR3 can provide essentially the same performance at a considerably lower cost. In addition with DDR4 the topology is different so you can't add to existing DRAM you must replace it all if you desire increased capacity. Naturally those who don't bother to do their homework can easily be duped into believing that they will see a system performance advantage by upgrading to DDR4, when it reality this is not true for CPU power desktop systems running 1600 MHz. or higher RAM frequency and 2133 MHz. for APU powered systems.

It's not about speed, it's about the fact that JEDEC calls for stacked DRAM chips in the DDR4 specification. As it matures, we should start seeing much higher memory densities once prices start come down and as DDR4 makes its way into the consumer market. It's just easier to hype up overclocking to an enthusiast forum, even more so when no mainstream platform uses it.

 

[Harry]

Intel ditched the FSB when skt1366 was released 2008. That was 6 years ago... AMD also has been moving the PCI-E root complex to the CPU which means that they very well might not need HyperTransport in the future. All in all, your post is laughable to say the least (no offense).

No offence taken, was sort of testing to see if anyone would correct me as I'm unsure about the legitimacy of my claim. I suppose I'd better do some research before contributing to this forum in the future.

 

[The N]

that is way to high for most users. i don't think many user will be interested in that high Bus speed RAM to wish for. more than half gamers or users, dont bother go above 2133mhz speed, 1600mhz or 1866mhz is so much standard these days. only Overclockers or reviewers want to have higher speed bus in benching area.

BUT GKSILL has its best overclocks past years and still they are the best overclockers in RAM. i remember RJ 1600mhz speed ram CL7, can easily go to 2133mhz with maximum loosen timings to CL9 the performance was better than RJ X 2400mhz cl11. it was just GREAT to compare.

 

[Aquinus]

that is way to high for most users. i don't think many user will be interested in that high Bus speed RAM to wish for. more than half gamers or users, dont bother go above 2133mhz speed, 1600mhz or 1866mhz is so much standard these days. only Overclockers or reviewers want to have higher speed bus in benching area.

BUT GKSILL has its best overclocks past years and still they are the best overclockers in RAM. i remember RJ 1600mhz speed ram CL7, can easily go to 2133mhz with maximum loosen timings to CL9 the performance was better than RJ X 2400mhz cl11. it was just GREAT to compare.

Wow, did you just not read my post I sent to @Harry ?

Intel ditched the FSB when skt1366 was released 2008. That was 6 years ago... AMD also has been moving the PCI-E root complex to the CPU which means that they very well might not need HyperTransport in the future. All in all, your post is laughable to say the least (no offense).

The FSB is gone, there are only two real external buses (aside from memory itself) for mainstream Intel CPUs now: DMI and PCI-E, QPI doesn't tend to get used unless your running multi-cpu setups on skt2011(-3). So really any bus speed other than the DRAM itself doesn't matter for memory speed. The MCH is a thing of the past.

With respect to memory speeds, I have 2133Mhz memory and it's significantly faster than 1600Mhz memory. The problem is that most CPUs and applications don't need it or can't use it. 2400 CL10 versus 2133 CL9 is measurably faster, bandwidth wise and latency drops almost 4-5ns. Once again, most computers and applications can't actually utilize it, so it really means nothing.

 

[The N]

With respect to memory speeds, I have 2133Mhz memory and it's significantly faster than 1600Mhz memory. The problem is that most CPUs and applications don't need it or can't use it. 2400 CL10 versus 2133 CL9 is measurably faster, bandwidth wise and latency drops almost 4-5ns. Once again, most computers and applications can't actually utilize it, so it really means nothing.

rightly said.......

tridentX seems faster even with higher CAS. no doubt 2133mhz faster than 1600mhz in any comparison. but these comparison are only limited to Benchmarking. otherwise gaming doesn't suppose to make any real difference at all.

andyou got good RAM 2133 @CL9 is not easily available. most are CL10 or CL11.

 

[Aquinus]

andyou got good RAM 2133 @CL9 is not easily available. most are CL10 or CL11.

I really notice no difference with them. I would opt for 1600 CL8, low voltage, and higher density DIMMs if I had to make the same decision over again to be honest.

 

[The N]

I really notice no difference with them. I would opt for 1600 CL8, low voltage, and higher density DIMMs if I had to make the same decision over again to be honest.

YES, Actually i Too. as recently i made decision and bought 1866 @CAS9 over 2133 CAS11

Tighter the Timings, better performance would be. we can only get 1600/1866mhz with lower/Good timings. i usually do overclocking and benching so i tried many different rams with different CAS.