Galax HOF DDR4

[NumLock21]

New limited edition ddr4 memory from Galax. Chrome plated and cherry picked Samsung B-die chips. Only 50 of them availabe through the Galax store. Comes in 16GB (2x8gb) kits. 20% discount on these rams, for those who have purchased their galax 1080ti oc lab edition video card. The rams can achieve,  4133MHz, with CL 19-21-21-41 timings, at 1.35V. 

$368

$294 aprox., for those that got the gpu

 

 

 

 

https://www.techpowerup.com/236227/galax-hof-extreme-limited-edition-chrome-plated-ddr4-modules-now-available

 

http://galaxstore.net/GALAX-HOF-EXTREME-OC-Lab-Limited-Edition-Memory-16GB_p_140.html

[MageTank]

Guess I'll still be waiting a while for XMP 16GB DIMM's that can match or exceed my 3600 C14 setup. Really wish they'd figure out multi-rank DDR4 4000 already, lol. 

[Gungpae]

That's a pretty penny :^)

[Sauron]

I always find it funny how these make no discernible difference in 99% of games and yet they are marketed to gamers 

[Bananasplit_00]

Looks amazing 

[Techicolors]

this might be interesting to test on a ryzen rig.

 

also i recall Galax making a different DDR4 ram, think it sold out relatively quickly cause it was on sale. 

[WereCat]

26 minutes ago, MageTank said:

Guess I'll still be waiting a while for XMP 16GB DIMM's that can match or exceed my 3600 C14 setup. Really wish they'd figure out multi-rank DDR4 4000 already, lol. 

Your 3600 C14 is XMP or your OC?

 

Edit:

nvm... seen your profile

[MageTank]

18 minutes ago, Sauron said:

I always find it funny how these make no discernible difference in 99% of games and yet they are marketed to gamers 

Memory overclocking does impact many games, just these XMP kits are slower than almost every manually tuned kit, and extremely overpriced. The key to seeing the difference, is to pay attention to minimum framerates/frametime. Average framerates are not impacted much at all. 

 

Just now, WereCat said:

Your 3600 C14 is XMP or your OC?

My overclock. Hard to drive 3733 on 16GB DIMM's (have tried on 3 different boards, including the Apex, impossible to stabilize without compromising on latency) and I can't get any tighter. I am currently testing 3466 C13-13-13-28-2 (7.5ns absolute compared to 7.77ns absolute of the 3600 C14-14-14-28-2). If I can get that to stick, I will be happy. Would get me below that 35ns latency goal I set for myself.

 

The fastest 16GB DIMM XMP is 3866 C18-18-18-38, which is 9.3ns absolute, 21% slower than my current OC. You would need 3866 at C15-15-15-30-2 to match my current absolute latency. 

[Sauron]

3 minutes ago, MageTank said:

Memory overclocking does impact many games, just these XMP kits are slower than almost every manually tuned kit, and extremely overpriced. The key to seeing the difference, is to pay attention to minimum framerates/frametime. Average framerates are not impacted much at all. 

Wouldn't buying a better cpu or graphics card impact your gaming experience far more than any memory tuning?

[PlayStation 2]

I know it's clocked fucking high but holy hell, the price on that.

[Carclis]

7 minutes ago, Sauron said:

Wouldn't buying a better cpu or graphics card impact your gaming experience far more than any memory tuning?

I think you're confusing memory overclocking with buying a higher clocked memory kit. But yes money is definitely better spent on those components.

Very nice looking memory though.

[MageTank]

3 minutes ago, Sauron said:

Wouldn't buying a better cpu or graphics card impact your gaming experience far more than any memory tuning?

It depends. GPU will help framerates when GPU is the bottleneck, but if you are CPU bound, or in the case of memory, I/O bound, you have to relieve that specific bottleneck. 

 

Often times, people only look at average/max FPS when they compare hardware, but only a few people actually test minimum framerates (Digital Foundry, GamersNexus, etc). In their tests, you can actually see the impact that faster memory has. The difference between dipping from 100 fps down to 50, and instead dipping from 100 down to 60. That 20% difference being only 10fps may not seem like much on paper, but for the gaming experience, it feels quite different. It's not just memory either.

 

Cache clocks are often ignored as well, often said to be "useless for real world", but I've seen what cache can do for the entire uncore subsystem on Intel platforms, including it's impact on the IMC. Changing my cache clocks from 4ghz up to 4.8ghz netted a 5% difference in minimum framerates in Witcher 3. Granted, that's a massive 20% difference in cache clocks for only 5% performance in minimum frames (and zero benefits to averages), so take that for what it's worth. I certainly would not recommend sacrificing on core/memory clock for a higher cache clock, just tighten it all up after all else is stable. 

 

I am currently testing various overclocks based entirely on latency. If you'd like, I can bench and publish those results once I dial in a few rock solid profiles to show latency's impact on gaming. If it's anywhere near as impactful as it is on Ryzen, I'll certainly be happy with the results. A guy did test this a few years ago, but he mostly touched primary timings, not quite as thorough as my complete primary/secondary/tertiary overclocking.

 

Here is his study: http://www.overclock.net/t/1487162/an-independent-study-does-the-speed-of-ram-directly-affect-fps-during-high-cpu-overhead-scenarios

 

At the end of the day, if you own an overclocking platform to use your overclockable CPU, you should manually overclock your ram. It's free FPS, and anyone can do it if they take the time to read. I tried to make my guide as thorough as possible, and even to this day, I walk several people through memory overclocking per week, with not a single person walking away without an overclock of some kind. For Skylake/Kaby, I even guarantee 3000mhz, which is a far cry from 2133 JEDEC. Would I recommend buying any 4000+ XMP? Certainly not. I know exactly how they achieved these speeds, and the sacrifices made on tertiary timings and RTL/IO-L is not worth it. They are trading every ounce of latency and efficiency for a higher frequency that does nothing for them, even in synthetics. A slower, tighter dual rank kit will top any of those 4000+ kits in any test, outside of the people that do 4000+ at C12 with obscenely unstable volts. 

[XenosTech]

3 minutes ago, MageTank said:

-snip- 

Now that you've said that can the fact my processor being locked be th reason it can't push past 2555 mHz ?

[MageTank]

7 minutes ago, XenosTech said:

Now that you've said that can the fact my processor being locked be th reason it can't push past 2555 mHz ?

On memory? Certainly not. I ran 3200 C15 Hynix IC's on a Pentium G4400 and my current ram kit ran 3600 C14 on my i5 6600T. I've never met a Skylake or Kaby IMC that could not do 2800. 3200 is where IMC quality starts to matter in the tests I have done. 

 

I can say however, since you are using Crucial (Micron IC's) that overclocking will be significantly harder for you. I would dial in 2800 C16-18-18-36-2 at 1.35vDIMM, and start from there. You might be able to hit 3000 C16. You can also push 1.4vDIMM, but Micron doesn't scale all that great with vDIMM, so it's probably pointless to go any higher.

 

If you'd like to try, I can walk you through it in private. I know for a fact we can hit 3000, just takes time. 

[XenosTech]

2 minutes ago, MageTank said:

On memory? Certainly not. I ran 3200 C15 Hynix IC's on a Pentium G4400 and my current ram kit ran 3600 C14 on my i5 6600T. I've never met a Skylake or Kaby IMC that could not do 2800. 3200 is where IMC quality starts to matter in the tests I have done. 

 

I can say however, since you are using Crucial (Micron IC's) that overclocking will be significantly harder for you. I would dial in 2800 C16-18-18-36-2 at 1.35vDIMM, and start from there. You might be able to hit 3000 C16. You can also push 1.4vDIMM, but Micron doesn't scale all that great with vDIMM, so it's probably pointless to go any higher.

 

If you'd like to try, I can walk you through it in private. I know for a fact we can hit 3000, just takes time. 

I'll get around to that again... I tried going through all the XMP profiles without touching voltage but none stuck besides the 2555mHz then I bumped it up to 1.35 as you suggested and nothing would post other than the 2555mHz

[MageTank]

Just now, XenosTech said:

I'll get around to that again... I tried going through all the XMP profiles without touching voltage but none stuck besides the 2555mHz then I bumped it up to 1.35 as you suggested and nothing would post other than the 2555mHz

Did you touch primary timings at all? You can't get fast AND tight with Micron, you have to choose one or the other. Part of the reason why no overclocker uses Micron. For stability, they are absolutely amazing, and can handle higher heat than any other IC that I've seen, but this also comes at the cost of extremely limited scaling with voltage and clocks. 

 

Still, I've had my hands on 3 different Micron kits, and managed to make 3000 happen even on the cheapest ASRock board (Z170 Pro4) so I am sure we can go higher than 2555, just takes some advanced tweaking, and a ton of time to validate stability. 

[XenosTech]

12 minutes ago, MageTank said:

Did you touch primary timings at all? You can't get fast AND tight with Micron, you have to choose one or the other. Part of the reason why no overclocker uses Micron. For stability, they are absolutely amazing, and can handle higher heat than any other IC that I've seen, but this also comes at the cost of extremely limited scaling with voltage and clocks. 

 

Still, I've had my hands on 3 different Micron kits, and managed to make 3000 happen even on the cheapest ASRock board (Z170 Pro4) so I am sure we can go higher than 2555, just takes some advanced tweaking, and a ton of time to validate stability. 

Nah didn't touch timings at all just wanted to see what would POST based on your guide but like I said I'll have to sit down a day and actually tinker with it til I get it to work

[Jito463]

1 hour ago, Sauron said:

I always find it funny how these make no discernible difference in 99% of games and yet they are marketed to gamers 

Because gamers will buy them, so they do make a discernible difference to the bottom line of the company.  

[dexT]

1 hour ago, Sauron said:

I always find it funny how these make no discernible difference in 99% of games and yet they are marketed to gamers 

These are for performance enthusiasts like the HOF cards or Kingpins. 

 

For overclocking and benchmarking.

 

 

[BuckGup]

2 hours ago, Sauron said:

I always find it funny how these make no discernible difference in 99% of games and yet they are marketed to gamers 

Not true. Look at Ryzen and i3s. From 2133mhz to 3200mhz you can get anywhere from 10-20fps

[EnergyEclipse]

If I had DDR4, whatever RAM I had now would be going on Ebay pretty quick, as that has to be the best heat spreader design I ever seen.

 

but one question,

 

WILL IT RGB

[Bouzoo]

Nobody should be shocked by the price, it's a limited edition of 50 pieces. 

[koji]

If you think expensive memory will improve your gaming, then you might wanna consider adding the budget to the GPU instead.

[VerticalDiscussions]

O-O!

 

Thats look almost as good as the Corsair SE RAM sticks!

[Sauron]

9 hours ago, BuckGup said:

Not true. Look at Ryzen and i3s. From 2133mhz to 3200mhz you can get anywhere from 10-20fps

Why would you buy ultra expensive premium ram and then pair it with an i3? Why not just spend the extra money to grab an i5 or a ryzen 5?