G.SKILL F4-3200C16D-8GVK Overclocked and Reviewed by HW-DB: DDR4-3600C14 at 1.72V

Last week we covered two HW-DB DDR4 overclocking reviews of kits that used not-so-overclockable ICs. This week, HW-DB reviews the G.SKILL F4-3200C16D kit, which uses Samsung ICs and is a lot better for Skylake overclocking practices. We are also delighted to see the return of the scaling graph, which you can find on page two of the review.

Let's have TaPaKaH from HW-DB do the talking: "When overclocking Samsung-based memory modules we usually observe the following behavior: 1) If the CAS latency is kept without change, then the dependency between stable frequency and minimal stable voltage is close to linear. 2) The minimal stable values for the rest of the primary timings (tRCD, tRP and tRAS) depend mainly on the memory frequency and operating temperature, with voltage playing close to no role whatsoever. However, on Skylake you also have to keep in mind that tRCD must be equal to tRP at all times and that the minimal tRAS value one can set in the BIOS is 28.

The conclusive lines read quite positively: "Having spent a few days testing the latest version of G.Skill's Ripjaws, we have established that it is still the daddy of all the enthusiast memory products. While the uniqueness of the looks might have eroded over time, the Ripjaws V still possess the overclockability and the flexibility that made its predecessors so successful among the enthusiasts." Check out the full review at HW-DB.com!


10

Belgium Massman says:


United States steponz says:

Come come we don't see a type performance chart.... mfr vs samsung vs afr... etc.... With all sorts of memory intensive benchmarks.. something for the overclocker..

United States Random says:

samsung is better than MFR for air I think AFR edges samsung out on air. On cold AFR and MFR are king.

United States steponz says:

Lets see the proof.. I already know what it seems to be.. but I want to see numbers.. screens.. etc.. Thats an article that needs to be written and I know it will be well received by overclockers... Just an idea for someone to get ahold of.

United States steponz says:

Also how will newer guys know which ones they should buy... Samsung might be better for them over afr.. etc.. I have several kits... maybe my samsungs are better than my mfr on LN2... and so on... Really would be a great article.. If I had time I would write it.. but that is not the case these days.

Romania Alex@ro says:

I will make that kind of article sir,just need some samsung and will have all sort of comparison,like 2 dimm fr vs 4 dimm,2 dimm samsung vs 4 dimm,mfr vs samsungs vs afr .Unfortunately only on air...

United States steponz says:

Sounds good to me

TaPaKaH says:

I'm not sure if an IC vs IC article will be a simple thing to do. The two major problems are IC quality (AFR currently exist only as cherry samples and there are no superbinned Samsung around) and platform preference (some boards like 4x4, some 2x4, some 2x8, etc). Also, (yet) there are no "common timing/subtiming" presets like we had on PSC or DDR3 Samsung, everybody "does their own thing" and values that work on one particular kit/board might not even POST on another kit/board.

Coming back to article, my plan was to use the same test platform for years and eventually build up a result base, which I can refer to in new reviews when doing performance comparison. However, finding the right memory-dependent benchmarks is not easy: XTU can gain +15% from one subtiming and nothing from CL18 to CL12, AIDA might vary from version to version, 32M and R15 vary too much from run to run. At this point, I am out of ideas. If you know of a sensible RAM benchmark, be my guest :)

websmile says:

BTW, very solid test, quality as good as usually

United States steponz says:

Those are very good points... but for a noob.... how does one tell which one would be better for them... I have stuff to test to tell... but when a rookie is looking.... not so much.. they go by advice...

Its just something to get the basics really. Community wise..

There does seem to be big difference from board to board also.


Sam OCX said: I'm not sure if an IC vs IC article will be a simple thing to do. The two major problems are IC quality (AFR currently exist only as cherry samples and there are no superbinned Samsung around) and platform preference (some boards like 4x4, some 2x4, some 2x8, etc). Also, (yet) there are no "common timing/subtiming" presets like we had on PSC or DDR3 Samsung, everybody "does their own thing" and values that work on one particular kit/board might not even POST on another kit/board.

Coming back to article, my plan was to use the same test platform for years and eventually build up a result base, which I can refer to in new reviews when doing performance comparison. However, finding the right memory-dependent benchmarks is not easy: XTU can gain +15% from one subtiming and nothing from CL18 to CL12, AIDA might vary from version to version, 32M and R15 vary too much from run to run. At this point, I am out of ideas. If you know of a sensible RAM benchmark, be my guest :)

Please log in or register to comment.

Leave a Reply: (BBCODE allowed: [B], [QUOTE], [I], [URL], [IMG],...)