New Posts  All Forums:

Posts by Sam OCX

Tested the 2x8GB 2133C14 version on Skylake recently. The best timings that I can get HCI-stable at 1500 and 1600: ... and some SuperPI 32M "binning" runs.
Micron D9RGx is no good. These chips don't scale above 1.45V and most of them have issues even with 1500MHz (DDR4-3000).
DS Hynix are not faster on Skylake, and much harder to clock. The temperature readout might actually be true. If you use a probe at the backside of the CPU, you'll see ~60c delta between that and the pot base temp. Who knows what temps are actually inside the CPU under heavy load.
Got to have a very good IMC for that. I've been through 13 5960X chips and none of them were capable of pulling HCI above 1666. The one I currently have won't even do 1666 32M in quad channel.First try of the new platform:
Tested another 32GB set of 2666C15 Fury. Similar to my previous one, this one is very good on MHz/volt, is even better than all 2666C13 that I had (no wonder, 1333c15 1.20 is a harder spec than 1333c13 1.35). However, it doesn't seem to like 1.6V+ very much.
Right now, only if you are very close to HQ of specific motherboard makers. Hopefully, the AFR will start making it into retail in a month or two.
What is the version number of the kit? (can be found on the stickers)
While everyone's on Skylake, got to test a Patriot 2400C15 kit that turned out to be Micron-based. What is worse, one stick on the kit is simply a tragic clocker.
Got hold of a 2x4GB Klevv DDR3-1600 kit recently. Turned out to be based on Hynix 4Gbit MFR. As you would expect on a low bin, the MHz/volt scaling could be better. However, due to decent tRCD/tRP limits and voltage tolerance, could fully stabilise my sample at 1400 11-13-13 and 1500 12-14-14. Some scatter SuperPi 32M runs just for the hell of it:
Specs of 3466C16 and 3600C17 models actually look much better than of 3800+ rated stuff.
New Posts  All Forums: