ADATA XPG Lancer DDR5-5200 2x 16 GB

[ir_cow]

XPG takes DDR5 to a whole new level with its Lancer line and an XMP profile of 5200 MT/s. It is time to test the waters and see how it performs compared to contenders. We will also take a look under that heatspreader and do some overclocking.

Show full review

 

[dicktracy]

They said they will have DDR5-8400 "later this year." I highly doubt that now

 

[mechtech]

Hmmm so does DDR5 have actual ecc? How does it differ from classic ecc that’s on ddr4 and prior?

does it work or help? Is there a way to test it in reviews?

 

[Wirko]

Hmmm so does DDR5 have actual ecc? How does it differ from classic ecc that’s on ddr4 and prior?

does it work or help? Is there a way to test it in reviews?

No, it doesn't. It's been discussed a couple times at TPU, for example here:

That's exactly how I understand it too (but there's not exactly an abundance of technical data available about DDR5).

On-die ECC has become somewhat of a necessity as DRAM cells keep shrinking and becoming less reliable with each new process node. It should detect bit flips due to cosmic rays, other radiation*, rowhammer attacks, bad cells, other reasons, or no reason at all. ECC on the data bus complements that and should be able to detect bit errors due to bad electronics in RAM chips or IMC, bad contacts in slots, overclocking, other reasons, or no reason at all.

* because most materials, and chip packaging too, contains some unstable isotopes.

How to test the on-die ECC ... I'm wondering, too. With full ECC you can probably overclock high enough and force some bit errors during transfers over the data bus, and those errors will be detected by the IMC. With on-die ECC you can't expect the same effect if you do the same.

 

[Honda_tpu]

1.35v is kinda low after looking at gigabyte's tachycon qvl ...

 

[Sabishii Hito]

Early word on the street is that Hynix is the IC to get for highest clocks, then Samsung, and Micron pulling up the rear.

 

[mechtech]

No, it doesn't. It's been discussed a couple times at TPU,

Ok. So NO ecc on desktop DDR5

Thanks

 

[ir_cow]

Early word on the street is that Hynix is the IC to get for highest clocks, then Samsung, and Micron pulling up the rear.

I have yet to get my hands on a samsung kit. Hynix is nice. Have a Teamgroup 6400 in-hand. Not many MB can boot that though...

 

[mechtech]

Early word on the street is that Hynix is the IC to get for highest clocks, then Samsung, and Micron pulling up the rear.

Looks like 4800 is stock speed

DDR5 Desktop RAM | Computer Memory Upgrades

Enjoy the next generation of multi-core CPUs with a DDR5 desktop RAM upgrade from Crucial. Buy today and get FREE US Delivery.

www.crucial.com

 

[Sabishii Hito]

I have yet to get my hands on a samsung kit. Hynix is nice. Have a Teamgroup 6400 in-hand. Not many MB can boot that though...

I actually have a Samsung-based kit (G.Skill 5600C40) but the highest I've gotten it to is 6200. That was on the Hero, I replaced it with an Apex so I'm expecting better.

 

[W1zzard]

I actually have a Samsung-based kit (G.Skill 5600C40) but the highest I've gotten it to is 6200. That was on the Hero, I replaced it with an Apex so I'm expecting better.

Good luck with that, let us know how it goes

 

[bonehead123]

Somebody please wake me up when the latencies & prices go way, way, way down. Until then.......

back to nappy time, hehehehe

 

[ir_cow]

Somebody please wake me up when the latencies & prices go way, way, way down. Until then....

Latency isn't everything. I still have DDR-400 with CL2. By your logical that should be the fastest thing on the planet.

 

[acerock]

I feel like the article was slightly misleading when it said "The 32 GB (2x 16 GB) XPG Lancer DDR5-5200 kit is available for around US$350."
It's not available at all and won't be available for the next few weeks, at the earliest.

 

[ir_cow]

1.35v is kinda low after looking at gigabyte's tachycon qvl ...

Looks like the highest for Micron is 1.25v on that list. Samsung it seems you can blast it with 1.5v though. I need to get my hands on one of these MBs for memory overclocking!

Z690 AORUS TACHYON (rev. 1.0) Support | Motherboard - GIGABYTE Global

Lasting Quality from GIGABYTE.GIGABYTE Ultra Durable™ motherboards bring together a unique blend of features and technologies that offer users the absolute ...

www.gigabyte.com

 

[Hyrel]

Latency isn't everything. I still have DDR-400 with CL2. By your logical that should be the fastest thing on the planet.

He never said latency is everything, but when cheaper 3600Hz DDR4 with CL14 gets you more fps in games than this DDR5 set, then yeah the latency is far too high for this to be worth the price.

 

[asdkj1740]

Early word on the street is that Hynix is the IC to get for highest clocks, then Samsung, and Micron pulling up the rear.

that is fact.

Looks like the highest for Micron is 1.25v on that list. Samsung it seems you can blast it with 1.5v though. I need to get my hands on one of these MBs for memory overclocking!

Z690 AORUS TACHYON (rev. 1.0) Support | Motherboard - GIGABYTE Global

Lasting Quality from GIGABYTE.GIGABYTE Ultra Durable™ motherboards bring together a unique blend of features and technologies that offer users the absolute ...

www.gigabyte.com

some said micron datasheet states lower max voltage than hynix datasheet's max voltage.

 

[Mussels]

The 1:1 ratio generally stops between 3600 and 4000 MT/s for Alder Lake CPUs. My Core i9-12900K maxes out at 4200 MT/s, which isn't rare if going by the sheer number of forum posts about many struggling to reach 3800 MT/s. It is safe to say that anything greater than 3600 in a 1:1 ratio configuration is completely dependent on the CPU memory controller.

Wow. You could swap that to say Zen3 and it would be word for word correct as well. Seems like there's a limit around that mark for memory controllers with todays tech, for everyone.

Decoding some of those graphs is a bit hard on the brain, is it possible to edit in a smaller font or something and mention the clockspeed and core timings, with the SKU? (And if its DDR4 or 5)
It's a brain melter

 

[ir_cow]

Decoding some of those graphs is a bit hard on the brain, is it possible to edit in a smaller font or something and mention the clockspeed and core timings, with the SKU? (And if its DDR4 or 5)
It's a brain melter

Can't control the font size. Frequency and timings (along with if its DDR4/5) is listed on the setup page to avoid cluttering the graphs.

My reasoning is brands can have multiple kits with the same speeds and that is hard to read in a chart. When I go buy memory, I look for the part number and just google that. Find the lowest price. Even better when I read a review and I see different memory I like, its easier to find. Instead of having to search for that old review to get the part number.

 

[ir_cow]

@Mussels Charts have been updated. Thank you for the feedback. It is always welcome and only makes TPU content stronger.

 

[Mussels]

@Mussels Charts have been updated. Thank you for the feedback. It is always welcome and only makes TPU content stronger.

I have ADHD, so maybe it was just me - but when i tried to find and compare the same kit between graphs i just couldn't keep up
thanks for the tweaks - that's a LOT easier

Now that i can make sense of it:

Wow. a lot of intel users bagged out AMD for high memory latency results, but 3600 C16 there is a good 11ns slower than my 5800x and i have FAR worse timings.
Theres so much to unpack with AL.

 

[W1zzard]

but when i tried to find and compare the same kit between graphs i just couldn't keep up
thanks for the tweaks - that's a LOT easier

your feedback is much appreciated

 

[Mussels]

Okay one more question before my brain explodes: Does memory performance change in P core or E-core only mode, like Ryzen does with inter-CCX latency?

 

[asdkj1740]

Okay one more question before my brain explodes: Does memory performance change in P core or E-core only mode, like Ryzen does with inter-CCX latency?

default ring clocks are different , 3.6g vs 4.7g for 12900k out of the box vs p cores only.