G.SKILL Trident Z5 NEO RGB DDR5-6000 32 GB CL26

[ir_cow]

Today’s review focuses on the G.SKILL Trident Z5 NEO CL26 memory kit. Tailored for AMD Ryzen processors and systems, this kit aims to deliver peak performance thanks to ultra-low timings paired with a 6000 MT/s clock config, hitting the Ryzen Memory Controller's sweet spot.

Show full review

 

[MxPhenom 216]

Am I missing something here? That 285k is looking suspiciously competitive. Within single digit fps compared to AMD competitors

Edit: oh on a few games 1080p. Basically true for 4k on majority of games

 

[vbq7qK68eyYAH4iR]

this is tough to write, what’s the point of this review? The product is EOL and can’t be bought by consumers. I know the numbers should be identical between this kit and the newer 1.45V but it just seems odd. Loving the low latency memory testing. Thanks for the review.

 

[Pepamami]

No ZenTimings screenshot in Test Setup :CCC

 

[ir_cow]

this is tough to write, what’s the point of this review? The product is EOL and can’t be bought by consumers. .

I just review whatever is sent to me. I assume G.SkIll PR didn't think it would be discontinued either. Demand was so high this reviewsample is a CKD repurposed.

I assume they probably can't produce enough at the 1.4v bin and it was easier just to discontinue this one and make a new version that has a higher voltage so binning becomes easier. Just my thoughts. No idea the reason reason for 1.4v being discontinued.

 

[_roman_]

Some other side give those modules away. I also saw that 1.40 vs 1.45V DC thing today.

• Already EOL. New revision is using 1.45 V for 6000 MT/s EXPO profile

Some stuff behaves not to my expectations. Thanks for the review.

Cl26 looks awesome. The difference is sometimes barely noticeable. Seems after the mainbaord you could also burn a lot of money for dram with nearly no benefit.

 

[Hugis]

Good review , maybe put this link in page 3 for people wanting to know about the secrets of CKD

 

[Makaveli]

Great memory but will be on the expensive side, this kit with sub timings tweaks should be a beast.

Right now about $259 CAD and it shows a preference for 800 series motherboards(Have noticed this for most sub CL30 kits) so you will have to keep an eye on the QVL also.

No ZenTimings screenshot in Test Setup :CCC

I agree would be nice if that was added.

 

[ir_cow]

No ZenTimings screenshot in Test Setup :CCC

Will add it eventually. I have them all from the 9800X3D which is not what I want to show.

 

[rv8000]

I think it would be important to start listing trefi and trfc values in memory comparisons. There are some large differences when comparing m-die kits who can have trfc values set at double what some of these faster a-die kits set via xmp/expo.

Low cl kits at relative frequency with better refresh windows are obviously going to be faster.

 

[ShrimpBrime]

Yes, this ram released January this year and still not available to the US.
According to some close friends, the kits are rated for 1.40v and 1.45v, the 1.40v kits are more desirable.

Here's a screen shot of the potential pulled from HWBot

 

[Calmmo]

Looks like a super bin to get that at 1.40v (also why they already discontinued it) Proly can push to sub 45ns on RL.
Even my old 2022 a-die could do sub 50ns/8000mts. The xmp numbers really say it all tho, just buy the cheapest xmp/expo kit if you cant tune it your self at all, they all virtually perform the same.

 

[Makaveli]

Looks like a super bin to get that at 1.40v (also why they already discontinued it) Proly can push to sub 45ns on RL.
Even my old 2022 a-die could do sub 50ns/8000mts. The xmp numbers really say it all tho, just buy the cheapest xmp/expo kit if you cant tune it your self at all, they all virtually perform the same.

Facts.

My CL30 Kit with EXPO on and sub timing tweaked already provides better numbers and lower latency than this CL26 kit with just EXPO on.

 

[Guwapo77]

Yes, this ram released January this year and still not available to the US.
According to some close friends, the kits are rated for 1.40v and 1.45v, the 1.40v kits are more desirable.

Here's a screen shot of the potential pulled from HWBot

View attachment 395094

Guess I got really luck and was able to snag this kit from Newegg.com (USA). It didn't stay in stock long at all.

 

[ShrimpBrime]

Guess I got really luck and was able to snag this kit from Newegg.com (USA). It didn't stay in stock long at all.

View attachment 395103

I can find some, but would have to come from half way around the world. I have no need for it honestly with an Intel setup. These are great for AMD 1:1. Just set it and forget it.

 

[docnorth]

Not bad, but even on AMD the cheaper 6400 cl32 looks slightly better.

 

[Waweq]

I have LEXAR version of 6000mh CL26 EXPO sticks, amazing sticks and they got thermal pads so they don't heat up easily

 

[evernessince]

I can find some, but would have to come from half way around the world. I have no need for it honestly with an Intel setup. These are great for AMD 1:1. Just set it and forget it.

You don't buy this kit to set and forget. You absolutely want to tune subtimings. Most of your performance boost is going to come from that. Otherwise might as well get a CL30 kit and use buildzoid's timings. It's going to be faster than this at EXPO.

 

[The King]

@ir_cow What happens to lifetime warranty when a G.Skill product reaches EOL?

End of Life so end of warranty?

 

[ShrimpBrime]

You don't buy this kit to set and forget. You absolutely want to tune subtimings. Most of your performance boost is going to come from that. Otherwise might as well get a CL30 kit and use buildzoid's timings. It's going to be faster than this at EXPO.

For sure. But it's A-Die. Pretty much seems most kits are reasonably tweakable.

I'm fine at 6800 CL28. With a different kit.

 

[Sol_Badguy]

@ir_cow

Page 4: the MEG Z790 ACE isn't mentioned in the test system table at the top of the page. I know it's pictured at the bottom of the page but still.
Page 5: the CPU-Z screenshot shows the 285K with the Z890 Unify-X instead of 13900K and (one of) its corresponding mobos.
Also at the bottom of the page, some confusing typos whatever with the naming of the 3DMark charts, were those supposed to be TimeSpy, Steel Nomad, CPU Profile in that order?
Page 7: The 1% lows charts for Alan Wake 2, TLoU 1, Cyberpunk 2077, Elden Ring, Spiderman Remastered, Starfield have the resolution 1440p instead of 4K. A typo given the results look correct for 4K but you know what that is? Karma for excluding the most relevant gaming resolution.
On the same page the average fps charts for Baldur's Gate 3, Counterstrike 2, Remnant II are the 1080p ones from the previous page.
Page 13: The average fps charts for Baldur's Gate 3, Counterstrike 2, Remnant II are the 1080p ones from the previous page.
On pages 17-19 the frametime analysis graphs have 6200 MT/s on them, a typo again, no big deal.

Some additional things I would have liked from this review:
- the actual models of kits listed in the comparison tables at page 4, would've made it easier/faster to identify them;
- that the G.Skill 30-38-38-96 Z5 Neo kit AMD was sending out to reviewers testing the Ryzen CPUs was also present here just to show how pointless this new kit is for everyday use, paying 50% more with additional voltage for academic gains in SOME workloads.

P.S. Also about that (hopefully still upcoming) Arrow Lake memory scaling 20+ kit royal rumble round-up shootout mega-review, honestly looking at this review (along with other samples/teasers in the past) I'm seeing similar if not better scaling with 13900K than with 285K, so about those supposed benefits of CUDIMM, well most likely for future platforms where huge speed means more gains (than right now) at much more affordable prices, but right now the sweet spot for Arrow Lake looks like 7200-7600 MT/s, speeds which don't warrant CUDIMM.

 

[edlee]

I believe subtimings make a much larger difference, compared to primary timings on AM5, can you please post the subtimings, and do a separate review using easy ddr5 timings from buildzoid

DDR5 timings

 

[ir_cow]

@ir_cow What happens to lifetime warranty when a G.Skill product reaches EOL?
View attachment 395121

End of Life so end of warranty?

This goes for other brands too. If your DDR3 that just went bad, it is unlikely to get the same bin back. Lifetime warranty for most computer components are this way.

Obsolete or Discontinued Products​

Whenever possible, an obsolete or discontinued product will be replaced with the same product. If G.SKILL is unable to replace with the same product, G.SKILL will replace that product with a comparable product of similar specification or function.

 

[Makaveli]

I think it would be important to start listing trefi and trfc values in memory comparisons. There are some large differences when comparing m-die kits who can have trfc values set at double what some of these faster a-die kits set via xmp/expo.

Low cl kits at relative frequency with better refresh windows are obviously going to be faster.

Agreed this is the default for my Kit with EXPO on

Default tRFC 884 tREFI 11677

This has me at about 81ns in Aida for latency.

 

[ir_cow]

@ir_cow

Page 4: the MEG Z790 ACE isn't mentioned in the test system table at the top of the page. I know it's pictured at the bottom of the page but still.
Page 5: the CPU-Z screenshot shows the 285K with the Z890 Unify-X instead of 13900K and (one of) its corresponding mobos.
Also at the bottom of the page, some confusing typos whatever with the naming of the 3DMark charts, were those supposed to be TimeSpy, Steel Nomad, CPU Profile in that order?
Page 7: The 1% lows charts for Alan Wake 2, TLoU 1, Cyberpunk 2077, Elden Ring, Spiderman Remastered, Starfield have the resolution 1440p instead of 4K. A typo given the results look correct for 4K but you know what that is? Karma for excluding the most relevant gaming resolution.
On the same page the average fps charts for Baldur's Gate 3, Counterstrike 2, Remnant II are the 1080p ones from the previous page.
Page 13: The average fps charts for Baldur's Gate 3, Counterstrike 2, Remnant II are the 1080p ones from the previous page.
On pages 17-19 the frametime analysis graphs have 6200 MT/s on them, a typo again, no big deal.

Thanks for finding all the errors - they have been fixed expect for the Z790 MB. It is absent from the list because I am slowly switching over to it, but the data in the review is from the previous ASUS Z790 Apex MB. To avoid people thinking 8200 works on the Z790 ACE, I just removed it from the list for now.

Working on 3 reviews at once and being sick gives you massive brain farts. I better make sure I didn't make the same mistakes for the next few.

Some additional things I would have liked from this review:
- the actual models of kits listed in the comparison tables at page 4, would've made it easier/faster to identify them;

I remove it on purpose. I believe it distracts from the actual product being reviewed. Also complaints before when I use to include what brand is what, the tables become extra hard to read. Also once the memory kit is no longer being produced it is meaningless to have it as a comparison. This current setup is the best evergreen scenario I could come up with so far.

- that the G.Skill 30-38-38-96 Z5 Neo kit AMD was sending out to reviewers testing the Ryzen CPUs was also present here just to show how pointless this new kit is for everyday use, paying 50% more with additional voltage for academic gains in SOME workloads.

I don't have a CL30 NEO kit in the benchmarks. I do agree that you can easily overpay for memory. It really comes down to what is important to you and how your computer budget is looking like.

P.S. Also about that (hopefully still upcoming) Arrow Lake memory scaling 20+ kit royal rumble round-up shootout mega-review, honestly looking at this review (along with other samples/teasers in the past) I'm seeing similar if not better scaling with 13900K than with 285K, so about those supposed benefits of CUDIMM, well most likely for future platforms where huge speed means more gains (than right now) at much more affordable prices, but right now the sweet spot for Arrow Lake looks like 7200-7600 MT/s, speeds which don't warrant CUDIMM.

Ultra 200 (1851 socket) benefits the most from Ring Cache and D2D increase. You push those up and suddenly 8800 MT/s is leaving everyone else in the dust. Sadly in a stock setup, faster memory isn't the savior for this socket like the previous one.


Just a side note. tRFC is often greyed out for DDR5 because it is controlled / linked to another timing. The important ones are tRFC2 and tRFC(SB). These values have also been added to the memory comparison list. Just remember that motherboards have a will of their own to change these values. ASUS XMP-1, ASRock "Aggressive", Gigabyte "High Latency" and MSI "Efficiency Mode" will change the tREFI and tRFC values for better or worse. These reviews are done with the correct values used in order to be more universal.