Pros: Pros - Worked in a non-QVL situation: Was able to make these work in an application that was not listed by G.Skill or the MoBo mnfr. - Good silicon: Slippery substrate where the current flow seems very fluid and the signal seems to flow very easily. You don’t have to jack up the voltage to get things moving and, in fact, they don’t respond well to pushing the voltage past 1.35V. Samsung B-Die. I had to increase procDOT a bit just to hold the signal in place. - Pliable: Once I found a groove that worked I could pull it in different directions to reduce voltage or SOC. Found two efficient profiles that I liked and that worked. - GTZN turns out to be some of the faster rated RAM sticks for 3600Mhz. Cons

Cons: - G.Skill might want to consider testing and listing these for more mobo applications. I had to lift LLC and boost to SOC to make it work so maybe that doesn’t count, but 4sticks at 64GB is likely to need some tuning anyway.

Overall Review: Building a 3950X rig with ASRock’s Creator motherboard, I wanted a significant amount of RAM for content generation as well as high-end statistics and machine learning. Not being sure about 2x32GB at this time I’m left with getting 4 sticks running across 64GB. I was also shooting for pure CL16 timing across all 4 primary timing steps if possible, but G.Skill is only willing to certify cl 16-19-19-19-39 sticks for this motherboard in 64GB sets. Synching that much RAM across 4 sticks is a big lift. ASRock wouldn’t certify them either, but I noticed they had listed a smaller capacity GTZN set in the 3800Mhz version. Since my original machine was still running and I had some time, I thought I’d give it a try. I don’t typically OC my CPUs and have never tuned RAM, but AMD recommends a 3600Mhz RAM speed for gen 2 Ryzen f you can make it work. Being in my 60’s, however, I can remember a time, long before Facebook, when most things needed tweaking to get them to working. If you tackle this you’ll start out working to contain RAM voltage as much as you can while lifting the environmental voltage known as SOC. With a little work I found I could then drop SOC back to 1.1v. That level it’s supposed to be low enough to avoid interfering with peripherals or the new high-speed PCI-Express 4.0 bus and I found that to be the case. As slippery as this silicon is, I had to lift procDOT to 48ohms just to keep the signal from leaking out too quickly. I also needed to add a heavy LLC to keep the internal DRAM voltage from fluctuating. Increasing LLC is less energy efficient, but with this set I was able to keep the voltage low and SOC somewhat low so overall I got a pretty energy efficient package under everyday use. Steve Burke on GN has noted that Ryzen runs best when you can keep voltages contained and keep things cool. I was able to develop two stable versions of the same design without pushing the voltage too hard: cl 16-17-16-16-34 at 1.338V and 16-16-16-16-34 at 1.35V. I started higher but was eventually able to back SOC down to 1.1v and still run a stable DDR4-3600Mhz with an 1800Mhz FCLK (ie. matched matched one-to-one). On the secondary/tertiary settings I took a lot of them from DRAM calculator as well as a few of the settings from the first design my Mobo came up with under its onw self-training. Did some reading and research as well to tune some settings. That kind of fooling around to fit a non-QVL low-CL set of RAM may be more than you’re up for but 4 sticks can be a challenge anyway and for me it was worth it: Both designs can successfully complete 10 full passes of MEMTEST86 pro with no errors (about 24hrs) and run 2nd level Prime95 overnight with no errors. Pretty solid. Staying with Cl16 across the board seems to be worth it as well. Worldwide, the 3950x tests at 39,259 under Passmark’s CPU benchmark, but I was able to get 41,609 on the 1.338v design with no CPU overclocking (see attached pics.). The 1.35v design did slightly lower on the benchmark because that extra voltage can’t quite produce as good a set of L-cache speeds. For this machine they seemed to do best with the 1.338v 16-17-16-16-34 design and it stays fairly cool at that level for something that’s putting out a reliable 3600Mhz. It’s hard to stress this chip but, so far, I’ve seen brief single core speeds of 4.5Ghz and would probably get more if I enabled AMD’s auto OC function that automatically adds 200Mhz. Achieving 4.5+Mhz with no overclocking is probably attributable to the slippery, low resistance, low voltage design I can achieve with this GTZN set. Using that design I’m able to gives Ryzen the thermal head-room it needs to run. If you look close at the attached images you'll notice that one of the two MEMTEST outputs has better L-cache speeds and that's attributable to the lower voltage of the 1.338v design. In comparing the Passmark cpu scores for the 3950x -- published average vs. my results with their test you can see some improvement in using this RAM. It may be that not everyone used 3600Mhz, but I like to think that some of that advantage comes from making a +/- pure CL 16 design work. Even if this set were listed for your MoBo I’m guessing it would take more than loading XMP to make 4 sticks at 64GB work at full 3600Mhz. It’s my understanding that is often the case. For me, the fluid pliable design allowed me build a stable, high performing production-oriented machine that will last me a long time without pushing into speculative ranges of overclocking. I know that was a lot to read, but, if you've gotten this far, hopefully, it did you some good. Good luck!

Pros: If you've passed your Cisco CCNA certification or can program in Cisco's IOS code and work with Cisco enterprise routers at work, you'll love this thing. It has all the command line code you're used to for setting up passwords, sub-domains, config settings etc, just like the big routers, for thousands less. As long as you have experience, and a Cisco service contract at work to tap into when you need to read some documentation, you'll probably be fine.

Cons: The initial password burn-in is trickey, and irretrievable if you make a mistake. (Don't make a plastic brick. I did first time around.) Cisco makes it's money from service contracts so documentation is sparse (12 pages plus a pamphlet) and poor quality. To download updated copies of IOS code, or SDM user interface, or to obtain basic documentation of IOS programming commands requires buying an annual service contract. The user forums are for experienced, certified, Cisco professionals and weren't reponsive to newbies. The "user-friendly" SDM GUI software is pretty but unclear about what it's doing, misguiding about how to avoid buning in mistakes, and it sometimes freezes up. Other Cisco reviews note IOS programming is the way to go vs. SDM (svc. contract required to obtain list of available commands) Some users with balky machines thought maybe the SDM froze at times because it didn't fit well in the routers memory.

Overall Review: This is a niche product for people who've mastered Cisco IOS at work and want to play with it at home. IOS is a powerful set of line command codes if you have access to a service contract to download the doc. Cisco's Service contract people couldn't find any record that the SB 101 existed, but based on my description of the machine, estimated a service contract for this size router would be $115/yr. through an approved Cisco distributor. They pointed out a svc contract is better than paying $400/hr. - minimum chge of 4hrs - for a frantic service call to Cisco's support center later. A good point indeed. Good luck.

Pros: Very resilient/fault-tolerant. For a while I had entered some of the latency settings backward and they still ran like champ. They're supposed to OC to about 215-220Mhz, according to OCZ but mine went a lot higher than that. Nice too that the warranty covers them up to 2.8v. I never needed the full. 2.8v and it was nice to know I could push them faster without violating their design limits. I found these to be a good stable foundation for Oc’ing my x2 Athlon CPU. The hefty heatsinks convey a feel of quality, if you like that sort of thing. Very happy I bought these, even if I paid more than the current price.

Cons: As heavy as those heat sinks are they qualify as a weapon.

Overall Review: I’m running these in a DFI Lanparty UT nF4 SLI-DR Expert, which is a great mobo, but it’s very picky about RAM. It’s hungry for something that will take the turns without slipping. My value RAM never could keep up but this stuff runs fast and smooth. Ordered on Tuesday, Newegg delivered it to me in the midwest on Thursday (no joke).

Comments: Some of the reviewers found this drive to be noisy, others have not. The one I recieved was quiet as ripples in water most of the time, but sounded like a blender inside the case every 5th or 6th read. This flaw must be limited to just a few copies. Newegg staff was professional, responsive and accomodating about arranging an RMA. Unfortunately, I kept tinkering with the drive and waited too long to send it in. The RMA expired (my bad). Also found drive to report errors in OS. Had some system crashes which ceased when I unplugged this drive. It's a new system build so I'm still testing other possibilities at this point. Might, for instance be that I'm approaching the limits of the power supply. Don't think so, but again it's a new system so things aren't always what they seem. May also have been the download from NEC to upgrade the drive, or the new TV card. I'm still testing things, but the noise thing some report with this drive can definitely be real depending on the copy you recieve. Will still order from Newegg (just ordered a great motherboard for an upgrade on another machine). Will be a little more cautious about NEC OEM drives in the future though until I verify the source of this error. (I hate to give up the compatiblity advantages of this drive as listed in some reviews. )