G.SKILL Hits 4500MHz With All-New Trident Z DDR4-4333MHz 16GB Memory Kit (betanews.com) 72
BrianFagioli quotes a report from BetaNews: G.SKILL is a respected RAM maker, and the company is constantly pushing the envelope. Today, it announced a new DDR4-4333MHz 16GB Memory Kit (2x8GB) -- the first ever. While that alone is very cool, the company is bragging about what it accomplished with it -- an overclock that hit 4500MHz using an Intel Core i5-7600K processor paired with an ASUS ROG Maximus IX Apex motherboard. Pricing and availability for this kit is unknown at this time. With that said, it will probably be quite expensive. What we do know, however, its that the insane overclock to 4500MHz is for real. This was achieved using timings of CL19-19-19-39 in dual channel, which resulted in read/write of 55/65GB/s and copy speed of 52GB/s.
Holy crap, betanews is still around? (Score:4, Informative)
I had no idea.
Also that's some criminally fast memory, shame they are only 8GB sticks and shame it's about 70% more expensive than it was 7 months ago.
I'm literally not upgrading anything due to this, I can take 20% but this has become ridiculous. Count me out of the upgrade game.
Re: (Score:3)
Once you hit a level where your hardware does what you need the upgrade race becomes dumb. E.g. in two years lets say you'll have 120Hz binocular 4K goggles, what then?
Re: (Score:2)
Re: (Score:2)
AMD? (Score:2)
Why does this have an AMD tag on it. It's an i5.
Re: (Score:2)
Re: (Score:3)
This kit is only going to get that speed on one very specific ASUS Maximus Apex motherboard, which is Intel specific.
Re: (Score:3)
Please. I can take the same memory to another motherboard and fuck with the bus manually to overclock it. Don't you know how to bit-bang-bus, old timer?
Re: (Score:2)
not really, ryzen doesnt scale well past a point, that point being around 2900-3200 ish
Re: (Score:2)
Re: (Score:2)
every overclocker on the internet that makes a simple bar graph
Re: (Score:2)
Re: (Score:1)
Hopes?
As for AMD benefiting from faster RAM, sure, but Intel does too, sometimes more.
Re: (Score:3)
Fuck Intel and AMD, I'm going to add this to my ATmega328P!
Re: (Score:2)
Lol thank you for the laugh! Now to just figure out how to pop it in the breadboard
Re: (Score:2)
https://pics.onsizzle.com/elvi... [onsizzle.com]
Re: (Score:2)
Lol I thought you were linking me to an actual article showing Dimms linked to an atmega
Re: (Score:1)
I expect Texas Instruments to throw this into their calculators really soon now.
The level of progress they show is simply amazing!
Re: (Score:2)
Fuck Intel and AMD, I'm going to add this to my ATmega328P!
Not really enough enough pins to do a 30-pin SIMM... but it is fast enough to refresh DRAM and still do useful work. They have atmegas with enough pins to pull it off.
Re:AMD? (Score:4, Insightful)
So it's basically awkwardly written article with a summary that is trying its best.
I just wonder why they used an i5 setup for the overclocking and not an i7 (or Ryzen). I suppose there must be some super special i5 only motherboard out there that makes it possible? This should have be explained in the article.
AMD's CPUs are heavily bottlenecked by RAM (Score:2)
Re: (Score:2)
Re: (Score:2)
Citation please.
Every benchmarker on the planet. Go youtube em?
But it is not the ram but WIndows 10 scheduler bug. Under WIndows 7 it doesn't have this problem. Basically what is hurting Ryzen and handing the crown to Intel for games is due to the CXX NUMA architecture which it calls Infinity Fabric.
Basically a Ryzen is really a server oriented chip with 2 4 cores. Not 8. Cache is shared only on 4 cores each. Windows 10 spins all the freaking threads like a merry go round to the cores for power management so cell phones an
Re: (Score:2)
Re: (Score:3)
I've been using G.Skill memory modules in my builds for the last 10+ years. Never had a problem with them.
I bought two dimms and then I bought two more dimms and then my first two dimms BOTH went bad. However, G-Skill replaced them rapidly and graciously with the precise same P/N and I had no downtime, just time with half the RAM. I strongly suggest the same route to others. Buy your second pair of DIMMs later so they may have come from a different batch :)
Re: (Score:2)
Buy your second pair of DIMMs later so they may have come from a different batch :)
One time I bought two pairs of G.Skill DDR2 memory five years apart. Same part number and specs. Heat sink on the newer pair had a slightly lighter shade of blue than the original pair.
Re: (Score:2)
RAM has caught up with CPU speeds? (Score:3)
Back in the old days, RAM was clocked at the same speed as the CPU, so RAM could be accessed with a minimum of wait cycles. Then speeds diverged, making various levels of cache necessary.
Does the advertised 4.2 GHz speed mean we're back to RAM that's synchronized with the CPU? Or is the issue more complex than that? The 19-19-19-39 timing mentioned suggests that it is, but TFA is light on detail.
Re: (Score:1)
Everything in computers is only fast when "streaming". It's the same for CPUs, RAM, graphics, network, mass storage: Everything is pipelined. Fast processing requires avoiding random access, because there's latency everywhere.
Re: (Score:3)
It's also 4.2 GHz "data speed" as multiple bits are sent on a single clock cycle, hence I think it's running at 1.05 GHz.
So let's say a 1ns cycle time and the timings suggest latency might be a bit under 20-40 ns although who knows how the numbers have to be added up or not.
It adds up I think, consider the distance traveled, memory controller and CPU memory hierarchy the CPU-to-RAM latency may be something like 50 to 70 ns very roughly.
L1 cache still is well over 10x faster.
Re: (Score:2)
2.1 Ghz- it is DDR, not QDR RAM
Re: (Score:3)
Then speeds diverged, making various levels of cache necessary.
It is not the difference in clock rate that necessitated the use of cache, it is the latency. The physical constraints of having memory located on a DIMM external to the CPU result in unavoidable latency. Once you implement a cache to get around the latency, the CPU speed and memory speed are no longer linked. Introducing memory that runs at the same speed does nothing to change this - you are still using a cache to avoid the latency.
Meanwhile... (Score:2)
Re: (Score:2)
If you werent so full of yourself, you would know that DDR3 has been declining in price for the last 6 months.
I check Amazon daily. Prices for the G.Skill Ares 16GB (2 x 8GB) DDR3 1866 memory modules [amzn.to] is still $110. That price haven't changed for several years.
but I guess its hard to pay attention with all the stupid in your head.
Go find someone else to play with, troll.
Re: (Score:2)
I have about 2.3kg of 30-pins memory modules if you want to buy them.
Re: (Score:2)
Same with HDD prices. I'm thinking about buying used 2TB drives for my own storage array.
If I was to rebuild my file server, I would get the Seagate 2TB BarraCuda Compute hard drive [amzn.to] for $67 each. These are not NAS drives but should work all the same. That's slightly more in price than what I paid for Western Digital 1TB Red NAS hard drive [amzn.to], and an extra 1TB of storage space is hard to ignore. Alas, I won't be replacing my hard drives until they hit the five year mark./p.
Re: (Score:1)
Re: (Score:2)
There were huge gains in Core i3 6100 tests, the particularity being it applies to many games there. i3 are not overclockable but have a high clock in the first place. Save for one recent model which is overclockable but costs the same as an i5.
The only annoying thing, with Intel, is that you need a motherboard with Z170 or Z270 chipset to clock the RAM higher. So, they make it harder to max out single thread performance on a cheap budget. i3 7100, DDR4 2400 (up from the older 2133 limit) and B250 motherboa
A small correction (Score:5, Informative)
G.Skill does not manufacture the memory dies, it purchases the memory dies and assembles them into a DIMM memory module ready for sale to customers - Wikipedia.
Which means that technically they are assembling memory modules, instead of producing them from start to finish.
There are just four companies in the world which actually produce memory chips and they are: Micron (Crucial), Samsung, Hynix, and Toshiba.
That stuff is newbie trap though (Score:2)
I recently built a new computer, something I don't do very often (every 5 years~ and I got lazy last time and got a prebuilt, so it's been a while).
First, RAM speed barely makes a difference for most people since not everyone is editing videos all day (and in games it barely does anything).
Then, these kits only reach these speeds with the timings properly setup, on the right motherboards/cpu combo (even if all your hardware is advertised as being compatible with the speeds). Often only if you only use 2 chi
Re: (Score:2)
I think you have it wrong, video encoding only cares about megaflops pretty much while games react more to memory latency, or more precisely the whole { L1 + L2 + L3 + memory }.
Of course, the über RAM is silly with as you say the extreme frequencies hardly properly working but eventually when RAM will cost say (dummy prices) $103 for DDR 2133, $103 for DDR 2400 and $107 for DDR 3200 you should take the 3200 pretty obviously.
Re: (Score:2)
Fair enough :) Thus the n00b trap (I'm a noob), where it's pretty hard to figure out what it will actually do.
When I was looking at benchmark for DDR4 RAM, most showed no real improvement in most games, but some drastic improvement in certain high end photoshop or 3d rendering tasks ::shrugs::
Either way, you really have to know what you're doing if you're buying something above 3200~
Re: (Score:2)
Games differ a lot between them, they can also be where you need a +5% performance the most (even though you can really barely see it)
Yes "useful" photoshop-like tasks vary as well. Benchmark quality and selection varies.
Also generally, for games you might want 4 or 6 cores at 4GHz (or more Hz if possible and cheap), if you're a professional doing media or server/dev things you might like 16 cores at 2.7GHz better (say), have more memory channels and you don't give much of a shit about RAM speed.
Re: (Score:1)
Re: (Score:2)
Yup, I was mostly talking about 4000+ RAM.
meh (Score:3)
Wake me up when RAM speed has any noticeable effect on real world performance.
Besides,all they're doing to get this, is overclocking standard slower speed memory on a tester then binning the individual chips accordingly. You aren't getting RAM that was actually made to go this fast, whcih brings a LOT of reliability questions up.
Personally I'd rather have good reliability than a slightly higher score on benchmarks that you'll never notice in real life.
Re: (Score:2)
Say, you run games on a Ryzen CPU, on a 1080p 144Hz or higher monitor so small framerate difference don't go to waste.
You can notice it in very specific situations like that.
With early motherboard revisions, early BIOS, first gen CPU you might be almost certain the stupid overclocked speeds won't run though.
This story about 4500 MHz RAM is really pointless right now, the only way it's interesting is it signals that in the long run, you might/should have reliable and fairly affordable memory at 3600 or even
Re: (Score:2)
RAM speed affects Ryzen quite a bit. For some reason parts of the CPU internal bus are linked to RAM clock speed, and benchmarks show quite big gains in some applications.
Re: (Score:2)
Thanks. I just read that as yet another reason to avoid AMD.
Re: (Score:2)
THat is due to the WIndows 10 schedular bug which clears out the L3 cache forcing it to constantly reload. Intels share the same L3 cache for all cores. If that bug is fixed ram speed on AMD won't be so critical