Question Looking for experience with some DDR5 6000 RAM ?

[instawookie]

Just in the process of upgrading some memory with an amazon gift card I received, kind of torn a few brands that I hadn't really used before (g-skill and TeamGroup). Just looking for some feedback or suggestions. I feel like the corsair is the worst of that bunch speed wise however I've haven't had any bad luck with their products either. Current rig is a 13700k with ddr5 Aorus 5200 and a 4090-gigabyte gaming OC, with an Aorus Ultra Z-690 board. Things I don't know and have read, Hynix vs Samsung vs Micron, Samsung is the worst of bunch currently? Not sure how true that is, and I lack knowledge of timings, but it seems tighter is preferred and the lowest latency possibly is what you want from memory. I know I won't gain too much from the upgrade but I'm not going to really get any more speed anywhere else aside from overclocking. None the less which would you choose between the 3 or have any other suggestions?

TEAMGROUP T-Force Delta RGB 32 GB (2 x 16 GB) DDR5-6000 CL30 Memory (FF3D532G6000HC30DC01) - PCPartPicker

G.Skill Trident Z5 Neo RGB 32 GB (2 x 16 GB) DDR5-6000 CL30 Memory (F5-6000J3038F16GX2-TZ5NR) - PCPartPicker

Corsair Vengeance 32 GB (2 x 16 GB) DDR5-6000 CL36 Memory (CMK32GX5M2D6000C36) - PCPartPicker

Thanks in advance.

 

[Darkbreeze]

G.Skill is generally about the best memory brand out there. Corsair is pretty good too for the most part, but they quite often use a variety of different IC's even for same model kits and that introduces a lot of random BS into their compatibility sometimes.

Some of the Team group memory kits are really good, while others are really crappy, low end garbage. For me, G.Skill Trident series and Ripjaws series are almost always the first lines of choice. Then it just depends on the particular model of the memory kit for other brands.

For Intel platforms it's usually not AS critical as with most all Ryzen platforms, but IS still a consideration and in this way I've found G.Skill Trident Z, Trident Z5 and Ripjaws to be of the highest quality AND have very broad compatibility with the most platforms and generations.

You do NOT want that Trident Z5 Neo kit, as the "Neo" indicates that that memory kit was designed to be used with Ryzen platforms. Regular Trident Z5 (Non-Neo) is what you want to look for, or Ripjaws, then after those it just depends on the model but I generally avoid the standard Vengeance series kits as they are lower quality and not as broadly compatible in most cases. It's just not the same kind of high quality product as the Trident Z/Z5 or Ripjaws, or the Corsair Dominator platinum (Which is expensive, but also usually very good quality).

I'd avoid the Teamgroup memory altogether unless you can verify that it is a kit using one of the higher end ICs (Memory chips) like these others. Some of their kits are, some of them are not.

Samsung ICs are not the "worst of the bunch". Generally speaking, Samsung ICs have been the absolute best out there especially for kits using their "B-die" ICs. I think the discontinued the B-die memory chips though and not sure which of them currently is at the top of their product stack, but in general if you get memory from a highly regarded series and it has Samsung ICs, it's probably pretty decent. SK Hynic and Micron both have their own high end IC packages as well.

All of them, Samsung, SK Hynix and Micron have both lower end and very high end products so it just depends on which of them a company uses when building their modules.

As a fast and dirty rule, the lower the CL memory is, the more desirable at any given speed a memory kit is. A 6000MT/s kit with a CL30 primary timing is going to be a significantly faster, and usually also more expensive kit than say a 6000MT/s kit with a CL36 latency. In some cases, if a kit has fairly low latency, despite the fact that it's frequency (6000MT/s for example) is moderately slower than another kit with higher CL latency it will still end up being the faster kit due to it's lower true latency.

 

[instawookie]

G.Skill is generally about the best memory brand out there. Corsair is pretty good too for the most part, but they quite often use a variety of different IC's even for same model kits and that introduces a lot of random BS into their compatibility sometimes.

Some of the Team group memory kits are really good, while others are really crappy, low end garbage. For me, G.Skill Trident series and Ripjaws series are almost always the first lines of choice. Then it just depends on the particular model of the memory kit for other brands.

For Intel platforms it's usually not AS critical as with most all Ryzen platforms, but IS still a consideration and in this way I've found G.Skill Trident Z, Trident Z5 and Ripjaws to be of the highest quality AND have very broad compatibility with the most platforms and generations.

You do NOT want that Trident Z5 Neo kit, as the "Neo" indicates that that memory kit was designed to be used with Ryzen platforms. Regular Trident Z5 (Non-Neo) is what you want to look for, or Ripjaws, then after those it just depends on the model but I generally avoid the standard Vengeance series kits as they are lower quality and not as broadly compatible in most cases. It's just not the same kind of high quality product as the Trident Z/Z5 or Ripjaws, or the Corsair Dominator platinum (Which is expensive, but also usually very good quality).

I'd avoid the Teamgroup memory altogether unless you can verify that it is a kit using one of the higher end ICs (Memory chips) like these others. Some of their kits are, some of them are not.

Samsung ICs are not the "worst of the bunch". Generally speaking, Samsung ICs have been the absolute best out there especially for kits using their "B-die" ICs. I think the discontinued the B-die memory chips though and not sure which of them currently is at the top of their product stack, but in general if you get memory from a highly regarded series and it has Samsung ICs, it's probably pretty decent. SK Hynic and Micron both have their own high end IC packages as well.

All of them, Samsung, SK Hynix and Micron have both lower end and very high end products so it just depends on which of them a company uses when building their modules.

As a fast and dirty rule, the lower the CL memory is, the more desirable at any given speed a memory kit is. A 6000MT/s kit with a CL30 primary timing is going to be a significantly faster, and usually also more expensive kit than say a 6000MT/s kit with a CL36 latency. In some cases, if a kit has fairly low latency, despite the fact that it's frequency (6000MT/s for example) is moderately slower than another kit with higher CL latency it will still end up being the faster kit due to it's lower true latency.

Thank you for the feedback and giving me some clarity of things I didn't know.

 

[Darkbreeze]

Of course. Anytime.

 

[instawookie]

Of course. Anytime.

Ripppppppp So I was in the process of ordering memory today and the Gigabyte Aorus ddr5 6000 has become available but the catch is its a cl40 memory..... the G skill same price point basically is cl30.... Part of me wants the Aorus memory because thats 90% of my rig and I know that has to be the worst concept to validate buying it....

https://www.newegg.com/gigabyte-32g...&ranSiteID=8BacdVP0GFs-rhEgoinWBiqy9aKKIenEOQ

vs

https://www.amazon.com/gp/product/B09Y1BM3ZH/ref=ox_sc_act_title_1?smid=A3TOECTKC4OEBD&psc=1

 

[Darkbreeze]

I would absolutely buy G.Skill memory long before I'd buy any Gigabyte branded memory. Being "Gigabyte" branded memory does not, in any way, make it more compatible with the rest of your hardware whether that hardware is also Gigabyte branded or not.

 

[instawookie]

I would absolutely buy G.Skill memory long before I'd buy any Gigabyte branded memory. Being "Gigabyte" branded memory does not, in any way, make it more compatible with the rest of your hardware whether that hardware is also Gigabyte branded or not.

Without a doubt, doesn't make it any more compatible. Hate to say being brand loyal sometimes causes bad decisions . I've had great luck with Gigabyte in my past 2 builds, I've also seen my fair share of rants on FB pages and Forums about their parts. More than likely I'll end up buying both kits for my own testing and sale the extra kit afterwards, not seeing many comparisons for CL 30 vs CL 40 @ 4k resolution. I've seen a few videos at 1080p and 1440p using a 3080 and 12900k which does show some marginal differences 5-10fps. I did find this article fairly interesting as well so it has my wheels turning on the value of what you get. https://www.pcinq.com/ddr5-6200-c40-vs-6000-c30-a-quick-take-on-timings/

 

[Darkbreeze]

It's nothing against Gigabyte. I actually PREFER Gigabyte over ASUS, ASRock and MSI when it comes to motherboards, it is simply Gigabyte "memory". They are new to the memory game, and to the best of my knowledge they do not make their own memory but instead simply rebrand somebody else's products unlike G.Skill, who designs and assembles it's own memory modules.

 

[instawookie]

Seeing how this hasn't died yet, I guess I can provide a follow up. Aorus memory has done fairly well, was a little rough @ first playing COD with xmp @ 6000 and would get random crashes. After a couple Bios updates, I've experienced very little crashing since then with exception of yesterday playing WoW, I received a Kernal 41 error, what causes this I actually don't know, my guess is memory, but it's only happened once over the course of 2 weeks. I could never get this memory to operate at 6200 though, so around black Friday I'll probably try to swap a G-Skill 6400 cl-30 or cl-32. I'm curious of benchmarks differences between a cl30 vs cl40 nobody really does those reviews that I've found.

 

[Darkbreeze]

You don't need a review to determine the result of a different CL latency at equal frequencies. There is a very simply formula for determining this AND the fact is that True latency, or "First word latency" is the real factor in determining improvements in performance. It's not just CL and it's not just frequency. It's a combination of things but primarily it's a result of frequency AND CL latency.

And it is determined as follows, and this is taken directly from the person who explained it to me many years ago, Computronix.

1 / Frequency (not DDR) x Latency = True Latency (nanoseconds).

Stock 3200 @ 14 is faster than Stock 3600 @ 16:

1 / 1.600GHz x 14 = 8.75nS
1 / 1.800GHz x 16 = 8.89nS


So, if you have a 3200mhz kit with a CL 14 latency, you divide the frequency by 2 since it's double data rate and you need the actual frequency, not the double data rate, so you get 1600. 1600 equals 1.6. Using the formula above that results in an 8.75nS true latency which is faster than the 8.89 true latency you'd get with a 3600MT/s CL 16 kit despite the fact that the 3600MT/s kit has a higher frequency/speed. The lower the CL latency is at equal speeds, the faster the kit will be. And in many cases a lower CL latency will make even a slower kit, faster.

 

[instawookie]

You don't need a review to determine the result of a different CL latency at equal frequencies. There is a very simply formula for determining this AND the fact is that True latency, or "First word latency" is the real factor in determining improvements in performance. It's not just CL and it's not just frequency. It's a combination of things but primarily it's a result of frequency AND CL latency.

And it is determined as follows, and this is taken directly from the person who explained it to me many years ago, Computronix.

1 / Frequency (not DDR) x Latency = True Latency (nanoseconds).

Stock 3200 @ 14 is faster than Stock 3600 @ 16:

1 / 1.600GHz x 14 = 8.75nS
1 / 1.800GHz x 16 = 8.89nS


So, if you have a 3200mhz kit with a CL 14 latency, you divide the frequency by 2 since it's double data rate and you need the actual frequency, not the double data rate, so you get 1600. 1600 equals 1.6. Using the formula above that results in an 8.75nS true latency which is faster than the 8.89 true latency you'd get with a 3600MT/s CL 16 kit despite the fact that the 3600MT/s kit has a higher frequency/speed. The lower the CL latency is at equal speeds, the faster the kit will be. And in many cases a lower CL latency will make even a slower kit, faster.

I greatly appreciate the breakdown of all that and the education behind it, however it still doesn't show me real world results, in terms of FPS and Benchmark numbers. I already know that a lower CL latency will make a slow kit faster, but the question is how much? 3/4 fps? 5-10 fps? What are the benchmark differences between a 6000mhz @ cl30 vs 40? Thats what I'm looking for, trying to justify that value, same as 64gb vs 32gb.

 

[Darkbreeze]

See, that IS the thing. You are 200% right. It does NOT show you the "real world" results of ANYTHING, at ALL.

But let's cut the crap and tell it like it is. Nothing is going to do that UNLESS it happens to be a direct comparison of two identical platforms, using the same software for comparison, and ALL else being equal, and the fact is that unless is is ALL happening on the same bench with the same motherboard (Excluding motherboard comparisons, obviously.) there is absolutely ZERO chance of a fair comparison of anything. Ever. At all.

But, you can get a decent "idea" about a lot of things.

How much? Who in the hell do you think is going to give you a definitive answer on that when all of us just take it case by case? Pfffffft. Sorry man, I don't intend to insult anybody, ever, especially not by laughing at them, and we sure as hell try to be an inclusive bunch of geeks around here in contrast to a lot of the less tolerant veterans on most other forums, but we still can only draw from real experience. We don't have an inside track to the magic 8 ball, much as we'd really like to have that. LOL.

So, that formula I gave you that was loaned to me, I mean, take what you can from it and then reach out to learn, grow and expand your knowledge of how the finer points of tweaking specific platforms, happens. That is a much better effort, and from your replies I can tell that it is not an issue of intellect. Let me know about any further "results" and I will offer whatever help or insights that I can, as will others.

 

[Darkbreeze]

THIS, might also be very helpful. Also, a part of MY learning process, few years back. Be sure to click on "See more", because this is the only spot I think that still hosts the entire guide, and it is honestly about the most honest one you are going to find anywhere if you don't personally know somebody that specifically has a degree in memory related technologies as an engineer.

Comprehensive Memory Overclocking Guide

www.overclock.net

 

[instawookie]

See, that IS the thing. You are 200% right. It does NOT show you the "real world" results of ANYTHING, at ALL.

But let's cut the crap and tell it like it is. Nothing is going to do that UNLESS it happens to be a direct comparison of two identical platforms, using the same software for comparison, and ALL else being equal, and the fact is that unless is is ALL happening on the same bench with the same motherboard (Excluding motherboard comparisons, obviously.) there is absolutely ZERO chance of a fair comparison of anything. Ever. At all.

But, you can get a decent "idea" about a lot of things.

How much? Who in the hell do you think is going to give you a definitive answer on that when all of us just take it case by case? Pfffffft. Sorry man, I don't intend to insult anybody, ever, especially not by laughing at them, and we sure as hell try to be an inclusive bunch of geeks around here in contrast to a lot of the less tolerant veterans on most other forums, but we still can only draw from real experience. We don't have an inside track to the magic 8 ball, much as we'd really like to have that. LOL.

So, that formula I gave you that was loaned to me, I mean, take what you can from it and then reach out to learn, grow and expand your knowledge of how the finer points of tweaking specific platforms, happens. That is a much better effort, and from your replies I can tell that it is not an issue of intellect. Let me know about any further "results" and I will offer whatever help or insights that I can, as will others.

Not insulted in any way, I might not ever know if I don't ask right? it's funny you linked to the overclocking forum. I'm currently sifting through those as well in the process of undervolting, which has been hit or miss. Runs under load great and crashes at idle but I'll get there. As far as what I'm looking for something like this

View: https://www.youtube.com/watch?v=xyr0qzZBfRc&t=722s
seems to be as close as I can get 9/10 times. While the link I sent shows marginal gains, 2160p resolution changes kinda changes the dynamics and shifts the workload to the GPU and less CPU, I just don't know how much 2160p affects the memory side of things. My best bet at this point is to yolo 200$ and run some benchmarks lol.

 

[Darkbreeze]

Crashing while idle is usually MORE likely to be a result of a sub-par quality power supply than other things, but, I have seen it. But for the sake of not disregarding the obvious, what is the EXACT model of your power supply? And, how long has it been in service?

Going back to the original question, I would ALWAYS choose G.Skill over ANY other memory manufacturer. That is not to say that others don't also have quality products, especially Corsair with their Dominator platinum line of products, but I do not believe that ANYBODY out there has the same quality and performance as G.Skill, if compared at ANY price point. There simply is no other brand that has the quality at even close to the same price, and if price is no consideration, they still have the best memory kits on the market. Bar none.

 

[instawookie]

Crashing while idle is usually MORE likely to be a result of a sub-par quality power supply than other things, but, I have seen it. But for the sake of not disregarding the obvious, what is the EXACT model of your power supply? And, how long has it been in service?

Going back to the original question, I would ALWAYS choose G.Skill over ANY other memory manufacturer. That is not to say that others don't also have quality products, especially Corsair with their Dominator platinum line of products, but I do not believe that ANYBODY out there has the same quality and performance as G.Skill, if compared at ANY price point. There simply is no other brand that has the quality at even close to the same price, and if price is no consideration, they still have the best memory kits on the market. Bar none.

Power supply is a 1200w P2 EVGA Platinum supernova and has maybe 8 months of use, machine is only used for gaming. This issue only started occurring during WoW as of lately, temps aren't out of wack when gaming, 60C average gpu side 55-60C for cpu. Machine has ran fine for months playing dota2/forza horizon/tiny tinas etc.... I did have issues in the past with memory being set to 6000mhz for CoD and it crashed periodically, a few updates came out and it went away. Other crashes occurred during me playing with undervolting, I didn't try anything crazy, it did benchmarks just fine. up to a .0105 undervolt if I remember right, however it just wouldnt idle for long on desktop. WoW played fine the first week, after an nvidia update I did start gathering a couple Kernal 41 errors. Now this could be tied to a crap ton of things I'm still sifting through that lol.

Current repair attempts

Windows troubleshoot and repair drivers etc...
Nvidia drivers up to date... (May possible rollback)
Cleaned pc and double checked all connections after cleaning.
Memory Reseat.
WoW repair/scan tool.

Next attempts
(Running the 4090 @ 95% powerdraw and may even reduce to 90% just to see if it doesnt like 60C when playing, However every timespy benchmark gpu hits 61-63C and has been fine

disabled fast boot in bios) Being tested in the next 10 mins. together.
Reinstall of WoW...
Reinstall of Windows
I do have an available f25 BIOS but its relatively new and I prefer to wait a month or two on those.
Bite the bullet and buy the G.SKill and hope its not psu

I've also attached an image of the Kernal error thats populating.

Screenshot-2023-08-16-154535 hosted at ImgBB

Image Screenshot-2023-08-16-154535 hosted in ImgBB

ibb.co

 

[Darkbreeze]

So, I think I'd give the newer BIOS a shot.

Also, after installing Windows did you then go to the Gigabyte product page for your motherboard and download the latest chipset, network adapter and audio controller drivers and install them? Using the native Windows drivers is not really acceptable except when it is the only option because the hardware is too old and is not supported any longer by the manufacturer for the OS version in question.

Power supply and drivers are generally the main considerations when it comes to Kernel 41 errors.

 

[instawookie]

So, I think I'd give the newer BIOS a shot.

Also, after installing Windows did you then go to the Gigabyte product page for your motherboard and download the latest chipset, network adapter and audio controller drivers and install them? Using the native Windows drivers is not really acceptable except when it is the only option because the hardware is too old and is not supported any longer by the manufacturer for the OS version in question.

Power supply and drivers are generally the main considerations when it comes to Kernel 41 errors.

Yea downloaded all the gigabyte support software as of lately nothing really new has released since 2022 for it, the f25 bios has a couple issues per the z690 forums on the overclock.net page some seem to revert to f24 for audio crashing and xmp issues. I didn't receive any crashing yesterday, but I also didn't game alot either. Hoping the disabling of fastboot was the culprit. Also meant to note, the prior Kernal 41 errors in my log from 7/28-7/29 was caused by myself during my undervolting attempts.