[PRO OC] HyperX FURY DDR4 Memory: Comparing Retail and Engineering Samples

The OC community started to claim that for a few memory generations (ICs), it’s harder or even impossible to find a decent kit which would be able to match or outperform special binned ES samples that have been reserved for top overclockers or reviewers. After the launch of DDR4 with the Intel X99 platform the situation didn’t change at all, with more memory binners starting to feel unhappy and even somewhat de-motivated by the issue.

There were even opinions out there that a ‘weak’ ES kit would still be much better than the best available retail kit. The new Samsung B-Die made things even worse, with very big differences between retail kits – modules which can barely go above stock specifications, up to crazy clockers breaking Low Clock efficiency challenges.

After what I have seen so far and tested by myself I was really curious to compare and assess the differences. There could also be another factor involved. Could it be possible that some vendors were playing a more “dirty” game than others? So let the battle between AFR begin!

Full article at OC-ESPORTS.io


15

Belgium Massman says:

Love this article by Xtreme Addict! Nice to see one of the Elite guys put the theory to the test and check the difference between Retail and Engineering sample hardware :celebration: :celebration:

Noxinite says:

http://oc-esports.io/#!/404 Link not working?

Belgium Massman says:

http://oc-esports.io/#!/prooc/article/1/hyperx-fury-ddr4-memory-oc-review Works fine here. Try CTRL+R to be sure the cache isn't causing any issues :)

websmile says:

Interesting article, good work XA :)

Sweden Rauf says:

Great article as always XA!

United States GtiJason says:

Crazy how much the quality varies in the Retail HyperX Fury 2666C15 kits, with the best stick running 3866c13 at 1.575v and the worse at 1.9v. I understand these kits are not really binned as they are lower rated kits with 1.2v spec but 325mv seems like a lot. What surprises me even more though is that the 3866c13 at 1.9v stick booted 4000c13 when one of the 1.8v at 3866 sticks could not. Wonder how much it took to coheres them into that. It would be really nice to see a spreadsheet showing all voltages to get a better handle on how they behave but a sample size of 8 is far from conclusive. Oh and one more thing, not saying it didn't happen but the pic showing "And because it’s a review/preview of HyperX Fury 2666C15… I matched two best sticks in dual channel and clocked them very easily at 3866 12-18-18-28 1T, just like decent ES sticks." Is a single channel run of 32m ..... see




Still a well done and very interesting to me article, we need more guys like you around here, oh and btw is this the Pro OC article? either way good work, I approve :)

Poland Xtreme Addict says:

GtiJason said: Crazy how much the quality varies in the Retail HyperX Fury 2666C15 kits, with the best stick running 3866c13 at 1.575v and the worse at 1.9v. I understand these kits are not really binned as they are lower rated kits with 1.2v spec but 325mv seems like a lot. What surprises me even more though is that the 3866c13 at 1.9v stick booted 4000c13 when one of the 1.8v at 3866 sticks could not. Wonder how much it took to coheres them into that. It would be really nice to see a spreadsheet showing all voltages to get a better handle on how they behave but a sample size of 8 is far from conclusive. Oh and one more thing, not saying it didn't happen but the pic showing "And because it’s a review/preview of HyperX Fury 2666C15… I matched two best sticks in dual channel and clocked them very easily at 3866 12-18-18-28 1T, just like decent ES sticks." Is a single channel run of 32m ..... see


Someone posted a bad screenshot. Will be fixed soon :)

The proper screenshot:




Regarding booting and stability - those are different issues. I have seen that in the past, scaling is not linear for hardware, not only memories. For instance, I had Asus GTX 780 TI DirectCU, on LN2 between 1.4-1.65v I could do nothing. Not even boost like 50 MHz (it was maybe 20-30), after passing 1.7v card was flying 1800+ GPU MHz.

Regarding quality variety, well we can look at other memories, MFR, B Die, E Die and so on. For instance currently I have 10 sticks B die 3600 rated models and none of the sticks can do 3866 12-12-12-28 1T SPI32M copy waza even on LN2, even 3600 is not constant. How does it compares to those golden retail kits some ocers got? 4000+ 12-12-12-28 1T tight @ air? I think average @ AFR HyperX Fury seems much better overall. And regarding bigger sample size, I would love that, it's not the problem for me to spend more time to test like 40 sticks ;) The problem it's not that easy to get memories (okay, retails I can buy or get more from HyperX), but HyperX doesn't have more ES samples in fact, I asked ;)

I wanted to do something new, "unusual", not typical review like "how great the product is". My plan was to prepare something which might actually be useful for XOC community :)

United States GtiJason says:

"My plan was to prepare something which might actually be useful for XOC community" It appears that you've succeeded, and I know all about the variety of Samsungs ddr4 ic's, though I do believe that will tighten up a bit with a new chipset and when we start to recognize what voltages or ? seem to be freaking out many imc's. I have a feeling that some of the kits that have been written off as junk will end up surprising us when motherbord manufacturers figure out the "balance" between trace layout, proper volts and a new chipset to catch up to the swarm of ic's being released

Poland phobosq says:

First I had issues opening the article, and surprisingly refreshing 404 page helped. So much for wtfs, great reading from XA, thanks a lot Michal. Interesting hypothesis, good testing that involved a lot of work and encouraging conclusion, that actually retails stand a chance against ES. I would suggest going further with analysis - the scope of article is limited only to AFR-based mems, which are only a part of the ecosystem. Is it possible to have similar study about Samsung-based mems, especially B-die, which are already famous for its quality inconsistency? BTW one more thing - are OEM Hynix AFR available in retail? Maybe they could be also thrown in into the comparison, so that we'll see possible effect of magic juice added by HyperX.

Romania Alex@ro says:

Good facts and read for sure!

United Kingdom TX-OC says:

Link not working for me
It redirects me to http://oc-esports.io/#!/404

edit: It works now

United States sabishiihito says:

I guess I'm just unlucky with retail AFR, I've tried 2 kits of Predator 2666C13 and 1 of Fury 2666C15 and none of the sticks come close to these results.

TaPaKaH says:

Quality varies also on ES samples. I have a pair of Kingston 3466C16 ES AFR that can barely do 1800 13-18-18 at 1.70V.

United States sabishiihito says:

One thing I noticed, the AFR sticks I tried all had a different PCB from these. I have some MFR 2666C15 Fury that has the same PCB as the AFR samples used in the test and a couple of those kits are strong (I hear they were 3333C17 that got put in retail).

Australia Bullant says:

Nice XA

Please log in or register to comment.

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