G.Skill TridentZ5 6400 CL32 DDR5 scaling review
Click here to post a comment for G.Skill TridentZ5 6400 CL32 DDR5 scaling review on our message forum
ngoni615
What's the deal with testing ddr4 3200 cl36? wow. That timing is extremely loose. 3200cl14 would be much closer to those high-frequency ddr5 kits.
Undying
You are always gpu bound at high resolutions so memory and cpu matters little. Still you should give a crap :P
TLD LARS
Thank you for this test.
This is the second test with the same outcome, 400-600€ memory kits are only worth it for overclock highscores.
There are going to be comment like: "you need to disable E-cores and overclock ringbus" "why not 7000? this is where the magic starts to happen" the timing comment already happened.
chinobino
Just a small note, page 7 says "Memory DDR3 / DDR4 Read MB/sec".
Great article, very interesting!
asturur
Ok seems like ddr4 on intel 12 is still future proof.
Since gen 13 or gen 14 will force a mother swap again you can postpone the decision
Kool64
overclocking used to be the budget way of gaining performance. Now it's the "rich guys" way of gaining 5fps.
fry178
because of everyone ocing everything, and they just decided to do it themselves, and make some extra cash for binning stuff more.
XenthorX
Well that was interesting, might be the first batch of DDR5 that trully feels above DDR4 across the board.
nizzen
For Wow players 😛
icedman
wow once pricing comes down to sane levels this kit actually makes sense performance wise i would have liked to see some higher speed ddr4 thrown into the mix to actually compare 3200-3866 with cl 14-18 on the same cpu, I don't really get why 3200/cl36 is even being shown here its an unrealistic speed that i hope no one uses.
user1
I would really like to see how ddr4 at 4800mt/s + on alderlake performs, particularly for software that actually gains from the extra bandwidth.
Agent-A01
Alder lake can't run speeds that high. The absolute max you will see is around 4266MTs
nizzen
Gear 1 yes, gear 2 5600mhz+ ddr4
Mufflore
Take a look on page 6, last image:
https://www.guru3d.com/index.php?ct=articles&action=file&id=76858
3200CL36 is DDR5.
The point being to show how a slow DDR5 memory kit will perform.
Mufflore
Those results dont look feasible.
Take the last 2, there is an increase of 3% in memory b/w resulting in a 13% fps increase.
I dont think so 🙂
nizzen
Most likely it's just you that reading the numbers wrong 😉
Source;
[youtube=sgonfT3fCAs]
Mufflore
I took the figures from that graph, the graph also shows the large fps increase for the small b/w increase.
The issue didnt stem from my post.
Agent-A01
DDR5 3200CL36 kits don't exist.
The slowest DDR5 you can buy is 4800MTs
True but Gear 2 is irrelevant for DDR4.
There are two performance metrics in memory bound scenarios. Bandwidth limitations and latency. You can increase bandwidth infinitely but it won't matter if WoW is limited by total latency(which it is)
That's why there is a huge gain with "fully" tuned timings because the tweaked timings reduce total latency by a significant amount
IceVip
Would've already swapped my x299 for this platform if it wsn't for the absolutely poor ddr5 performance right now. By sheer mistake did i learn
that if you occupy all 4 slots the ram won't even go near 4800mhz.. Why the hell did they rush to release the platform if it scales like a bish.
Agent-A01
Unfortunately that's how all new platforms have been since a new DDR revision has been introduced. DDR3 and DDR4 both had teething issues at launch.
At least on DDR5 we get junk micron kits that are unstable out of the box at slow speeds/timings lol
