• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

What is the cheapest high-end memory terms of latency?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.
K

Komar

Registered
‍
Joined
Jul 30, 2017
Hi.

Please help me find appropriate choice for MSI Z270 TOMAHAWK + i7-7700K.

All that matters is latency in practice. No benchmarks, no bandwidth numbers, just the difference the RAM can show in tests like GTX 1080Ti at 480p with no AA ;)
The actual difference might be even 10% between mid-range RAM vs. high-end RAM. That's too much.
But let say the PC will be used for older games on 240Hz monitor - 2% better framerate won't make any difference, so it's not about the memory kits that would be chosen by any of you "serious" overclockers here. :)

The memory will have to work at low latency settings for many years. If anyone knows any advices how to keep 7700K IMC healthy through all the years, I would greatly appreciate any info.




One more thing. If price turns out to be too high, is it even possible to buy 1 out of 2 sticks, so 8GB now, and another 8GB in the future?
 
Ok. Let me rephrase that: least expensive memory which will be at the high-end side in terms of latency.

Any info will do. I know nothing about memory now, but few years ago I knew enough about DDR1 and DDR2 to know that this matter is very complicated.

- comparison tests - hard to find, especially when you focus on gaming latency. Tests like this should be done at ridiculously low resolution on high-end cards.
- which synthetic benchmarks overlap with gaming latency (back in the days, I knew SuperPI results and Everest (memory latency tests) were the ones to focus at. Now it might be misleading.
- what Nebulous wrote - tests from a year ago might be misleading. Same kit might have different chips now
- long-term use and influence it can have on IMC of the processor. Memory that is much better in tests, and is most expensive, might turn out to be inappropriate for long-term use with setting focused on low taltency, due to the fact that it means higher voltages, which means high risk of performance degradation. Or it might be other way around.
- current situation. It might be good to know if, for example, just one vendor produces chips that are fast, and the rest is significantly behind. If today is like 2 weeks away until another vendors plans to introduce something to compete, it could help to make a decission. Maybe slower RAM now and faster 2 months later?

Any info would be welcomed!
 
Sorry... a 1080ti @ 480p... wth????

There are no worries RE: the IMC...

The ratings and latency are on the packaging. You cam also do some math... speed/CAS latency to get a ballpark idea.
 
EarthDog said:
Sorry... a 1080ti @ 480p... wth????

There are no worries RE: the IMC...

The ratings and latency are on the packaging. You cam also do some math... speed/CAS latency to get a ballpark idea.
Click to expand...

Well, the more ridiculously low the res, the more you can see actual RAM importance, so yes, the lower the better, unless it's a game with dynamic geometry engine, where smaller res means less elements being rendered.
Specs alone might be tricky. IMC behaves strangely sometimes. I stopped reading into it after I've bough i5-2500K, so I don't know how 7700K's IMC behaves.
 
This Video is what you want to see.
Kaby Lake DDR4 Memory Guide, Scaling Performance Tested.

 
Komar said:
Well, the more ridiculously low the res, the more you can see actual RAM importance, so yes, the lower the better, unless it's a game with dynamic geometry engine, where smaller res means less elements being rendered.
Specs alone might be tricky. IMC behaves strangely sometimes. I stopped reading into it after I've bough i5-2500K, so I don't know how 7700K's IMC behaves.
Click to expand...
but scaling isnt linear. You are exxagerating a difference which does nkt extrapolate up. ;)
 
Gotta keep in mind that most folk buy "ram" we buy the chips. I could care less who "makes" it, most of the OEMs just slap a sticker on a pre-built board and that's it. That being said it would be wise to purchase ram in pairs as suggested because it will be difficult to find a matching pair if you space it out too far because processes change, products change etc. My personal favs have been Hynix (Hyundai) or Micron; Samsung is normally garbage. Looks good on paper but has poor endurance in the long term which seems to be a repeating theme if you look at their NAND products.

So for me once the IPhone releases and after the holiday rush I'm going to start looking for a 32GB set from either vendor.
 
If it was as easy as just looking at few tests, I wouldn't create this topic.

Please refrain from sliding away from this topic main purpose. I don't want to talk about why (or what for) I need latency. It doesn't matter how ridiculous you think 480p tests would've been, if they were made. And I definitely don't need to hear that it doesn't matter cause no benchmark shows it. I've been through the same thing over and over again, since Athlong 64, through C2D and further. Really. This is not a thread about what you think is a good test or not. Just let me focus on what I want without asking me to explain, cause when I start doing that, I'll just unleash an avalance of offtopic (when I start talking about high refresh rate in games, low persistence, attempting to lock the MINIMUM framerate at 120fps without any drops, about configuring the game options in a way noone else does, etc.)
The good idea would be to focus on worst case scenario, no matter what people think about the test it was done in. So if any of you have any suggestions what such a test might be, please tell me. I don't think anyone made 720p high/no AA tests using 1080ti or Titan, but maybe you saw some synthetic benchmark that showed surprisingly big gains on the expensive, fastest memory kits? That would be a good start for me to search for the info. :)

@wingman99 Thanx for the video but unfortunately it brought me no help/info whatsoever.

@Sentential Any advice where could I hope to find information about it? I know it might be difficult. Same memory might be based on different chips, just a few months after the positive reviews.
 
Techpowerup did something close to what you were looking for (games testing at 720p..not sure if they played with memory speeds and timings) with the ryzen 3 review. Perhaps check that.

Otherwise....considering running these tests yourself with what you have and post back so we can see the results. :)

Good luck. :)
 
Komar said:
If it was as easy as just looking at few tests, I wouldn't create this topic.

Please refrain from sliding away from this topic main purpose. I don't want to talk about why (or what for) I need latency. It doesn't matter how ridiculous you think 480p tests would've been, if they were made. And I definitely don't need to hear that it doesn't matter cause no benchmark shows it. I've been through the same thing over and over again, since Athlong 64, through C2D and further. Really. This is not a thread about what you think is a good test or not. Just let me focus on what I want without asking me to explain, cause when I start doing that, I'll just unleash an avalance of offtopic (when I start talking about high refresh rate in games, low persistence, attempting to lock the MINIMUM framerate at 120fps without any drops, about configuring the game options in a way noone else does, etc.)
The good idea would be to focus on worst case scenario, no matter what people think about the test it was done in. So if any of you have any suggestions what such a test might be, please tell me. I don't think anyone made 720p high/no AA tests using 1080ti or Titan, but maybe you saw some synthetic benchmark that showed surprisingly big gains on the expensive, fastest memory kits? That would be a good start for me to search for the info. :)

@wingman99 Thanx for the video but unfortunately it brought me no help/info whatsoever.

@Sentential Any advice where could I hope to find information about it? I know it might be difficult. Same memory might be based on different chips, just a few months after the positive reviews.
Click to expand...

I've been in the OSX scene for awhile so I haven't had to look until just recently now that I'm on an AW13r3. From my experience Gskill is garbage, Kingston, Corsair and Crucial are always on point. Sure enough looked at a set of 32gb DDR2400 HyperX Impact and a review shows them using Hynix so I found my RAM. Always had a soft spot for Hynix after setting some HWBOT records with em, eventhough they're *******s and eventhough they bankrupted my favorite ram supplier (Winbond) and almost bankrupted my second favorite (Micron) If you're concerned about getting garbage your safest bet is Crucial Ballistix because Crucial is owned by Micron and they are their official wholesaler. Micron is always solid, often not the VERY best but I cant say there has been a single bad set of chips chips in the 10+ years I've overclocked.
 
Last edited:
There are 3 popular memory IC in single and dual ranks. Dual rank = double sided modules = worse overclocking = higher latency. Somehow no one is saying about it in any reviews and in some tests/games can't just compare frequency and main timings.
Micron is in kits up to 3000
Hynix is in kits up to 3466
Samsung is in kits up to 4500
then is single or dual rank version where dual rank is barely ever above 3200 for all IC

The lowest latency has Kaby Lake so 7700K+Z270 but the best if it's overclocked. Then you get Samsung 2x8GB kit at 3200 CL13/14, 3600 15/16 or any higher frequency and set it to 3200 13-13-13 / 3600 15-15-15 / 3866 17-17-17 1.35-1.45V if you want it to run stable for longer.
Also low latency = balance between high frequency and low timings.

However, because of the way how new platforms work, I doubt that memory latency will affect gaming.
 
Last edited:
Tried to obtain some info last night, a lot of time spent, but no info found :(

1. No, techpowerup has no 720p tests, I guess noone has, and it would be pointless anyway, if the tests wasn't done on Skylake/Kabylake i7 or desktop i5, since IMC surely plays a significant role.
2. I can't find most of memories people test or talk about, here in Poland.
3. When I find something, I cannot find a single test, not to mention the info about what chips are under the radiators and how many.

I found theese: https://www.gskill.com/en/product/f4-3200c14d-16gtzr
Are they the famous B-die? (I don't know much about it, aside from the fact that I see it often in discussions)

Now, the questions are:
- is it good? I found no intel OC tests for this kit.
- if they were good, are they still made on same chips?
- F4-3200C14D-16GTZ) are 10% cheaper and same 14-14-14 3200MHz 1.35V, so I have no idea what is the difference here

And most importantly: Can I find something cheaper than that?
Those cost 990zł, whereas the absolute best and most expensive I looked at, are 1100-1200zł (3600 15-15-15 and faster).
Cheapest 2600+MHz 2x8GB costs 620zł, for reference, and there are things that might be a better choice at 750-800zł.
But I cannot find any info.
Then there are those: https://www.gskill.com/en/product/f4-3200c14d-16gvr
again, supposedly same specs, and 880 instead of 990zł.

Then there are those: https://gskill.com/en/product/f4-3200c15d-16gvr
at 780zł. They are 15-15-15, if I could reach 14-14-14, it might be good enough (or not, still don't know) if that more expensive ones are still too bad to allow for something more, like 13-13-13 or 3600 for example

And then there are even cheaper ones, but without knowledge about thier OC potential or even what's "under the hood" it's hard to say.

And on top of that, if IMC worked in simple way, it would be easy, but I have no idea how it actually behaves. I saw many strange things and that tought me to be very careful with any assumptions. I don't know how intel programmed 7700K's IMC. Maybe it's really not worth it to go for the fastest? Or maybe it is. The fact that the real performance might be different by even 15% where it's only 4% on all the reviews out there (done at 1080+Ultra (plus MSAA sometimes (facepalm) makes it a really hard task.
 
Last edited:
Either I'll soon be very embarassed for being blind, or there are no comparisons to any memory speeds/timings/frequencies etc.
 
Thanx. That techreport test is pretty good. The differences between 3000 and 3800 are surprisingly accurate (I mean they are big), which suggests that it might be worth it to hold of for a few months to be able to afford really expensive 2x8GB kit (just buy cheap 2x8GB for now)
Of course the kit they used is unavailable in Poland, but I can assume 3600 15-15-15 will behave similar to the ones they tested.

About the google phrase - of course I tried the same, but as you can see - a ton of Ryzen tests pops up, since it caught big attention recently (for average Joe, memory matters for Ryzen gaming PCs, but does not for Intels).

BTW. I wish someone pushed PC gaming into stacked memory. Capacity doesn't matter so much nowadays, and we can only wonder how much boost would be possible if we moved from 2006 in terms of latency.
Yeah. DDR4 3600 15-15-15, and DDR2 1200 5-5-5 available in 2006/7. We are basically stuck, just bandwidth increases, which is not all that important, especially for high framerate (100+) gaming. I think PC would get bigger boost in gaming from something like Micron's HMC, than it will get from next 5 years of CPU advancements. And I'd buy a 30% faster PC even if factory equipped 12-16GB of RAM was unupgradable.
Sadly for big corporations, PC gaming doesn't matter. :(
 
You must log in or register to reply here.

Similar threads

A
Making the Jump to AM5...
Replies
4
Views
447
Zerileous
Zerileous
BROOKLYN ZOO
coldboot (warmboot) issues @ R7-5800x | B550 Aorus Elite V.2 rev.1.0 | AGESA 1.2.0.5
2
Replies
31
Views
921
EarthDog
EarthDog
mackerel
1Rx16 to 2Rx8, filling in 1Rx8 testing?
Replies
2
Views
2K
mackerel
mackerel
R
Still Considering New Laptop
Replies
8
Views
854
DeadSmiley
DeadSmiley
Tyerker
Your dollar can go a LOT farther these days (mostly)
Replies
3
Views
982
stompah
stompah
Back