[H]ard|OCP had a chance to try out Corsair's upcoming Vengeance LPX 3600MHz DDR4 on a Kaby Lake based system. The XMP settings for this DDR4 were 3600MHz with timings @ 18-19-19-39-2T and the system booted with no problems at these defaults, an improvement from some scenarios with Skylake based systems. Running Prime95 for over a day posed no problem for the system, however Memtest86 did until the RAM voltage was bumped up to 1.41v from the default 1.36v at which point it could pass the tests with no problems. This shows some promise for overclocking addicts planning on upgrading to the refreshed Intel chip.
"We were lucky enough to get our hands on a new set of Corsair Vengeance LPX 3600MHz RAM this week and we immediately put it work with the new Intel Kaby Lake Core i7-7700K processor that is to be launched next month."
Here are some more Memory articles from around the web:
Memory
- Team Group T-Force Night Hawk 3000 MHz DDR4 @ techPowerUp
- Kingston HyperX Predator DDR4-3000 64 GiB @ Hardware Secrets
- Crucial Ballistix Tactical 3000 MHz DDR4 (4x 8 GB) @ techPowerUp
I wonder why on my skylake
I wonder why on my skylake build, to get my DDR4 3200 stable, I had to increase the system agent and VCCIO voltage by a little? (enabling XMP by its self was unstable)
It sounds like a combination
It sounds like a combination of the memory, the board and the CPU. I have a Core i7-6700K (Skylake) on an Asus Maximus VIII Impact board with Corsair Vengeance 4000 MHz DDR4. It works like a champ with the default XMP setting. No additional voltage required.
For me, it is a cure i7
For me, it is a cure i7 6700k, and a gigabyte gaming 6 motherboard and 16 GB G Skill RAM F4-3200C16D-16GVK
+razor512 Your comment is
+razor512 Your comment is interesting because I also had a similar issue after enabling XMP on my Asus Sabertooth X99. My CPU is an i7-5820K (Haswell-E). I might build myself a second rig with Z270 / i7-7700k / 64 GB DDR4 3200 (16GB x 4).
I’d still love to see the
I’d still love to see the impact of RAM speed on Frame Rating/Latency. No one has done this yet, but there are hints that RAM might be a big deal here based on HardOCP’s early skylake review that showed big FPS gains at low resolution (i.e. CPU dependent benchmark).
Hopefully PCPer’s 7700K review comes with two speeds of RAM for at least a couple of the game tests so we can see if there is any impact from DDR4 speeds..
Digital foundry did some
Digital foundry did some testing on i3 6100, i5 6500 and even i5 2500K. Increasing the memory clock by 30% to 45% increased the framerate by 10% to 15% in some cases, while others there is barely any performance increase.
https://www.youtube.com/watch?v=frNjT5R5XI4
https://www.youtube.com/watch?v=qksXthUcbiQ
https://www.youtube.com/watch?v=Er_Fuz54U0Y
Maybe XMP is the
Maybe XMP is the theoretically stable setting given all other factors being perfect. In a real world situation with manufacturing tolerances and different circuit path layouts, you need the oomph of more voltage to operate with stability at those speeds under the loads you’re applying to the system.
I wonder if any review site has compared the efficiency of different motherboards and memory. It would be a massive endeavor to be even close to covering all the manufacturers and model lines out there now.
It’s not about other things
It’s not about other things being perfect, it’s about choosing a setting which is stable. So it’s the optimal balance of performance without pushing towards instability.
Looks like they used Windows
Looks like they used Windows 7 with their Kaby Lake CPU. Even though Win7 is not going to be officially supported, it looks like it’ll work just fine.
3600MHZ RAM?
Meanwhile, I
3600MHZ RAM?
Meanwhile, I still struggle with 800mhz ram.
Have a look about Samsung
Have a look about Samsung Galaxy Note 7 Galaxy Note 7 here.