Why not to always follow tech tubers.

"if you put the data in the perspective of his audience, the data is 100% accurate".

By that I think he's implying that HUB audience are 100% normy users which I agree with. I would also say that most of the people on this forum are normy users so, there's that?

I also disagree with him on his opinion that there is no reason to go DDR5. There are a few reasons and Buildzoid also says so.

i would argue that going DDR5 as a large dataset processor or video editor is probably the same price as DDR5 where you need a lot of RAM.
 
3200 is a good choice, in the sense that most DDR4 systems can support 3200 without much trouble (and they're stock for 11th/12th gen and Zen 3 with 2x1R), while being very widely available.

The Vipers @ 3733 are currently pretty cheap though, so would be interesting to see the comparison, but DDR5-4800 is like DDR4-3200, fairly comfortably stock. Those Crucial sticks have gone up now though, £33 to £43 and £66 to £83 (for 2).
 
HUB posted another video and literally explained how people misinterpreted their low cost DDR5 video

youtu.be

DDR5 For BUDGET CPUs, Does It Make Sense Now? Core i3-12100 + DDR5-4800

Support us on Patreon: https://www.patreon.com/hardwareunboxedJoin us on Floatplane: https://www.floatplane.com/channel/HardwareUnboxedBuy relevant products ...
youtu.be youtu.be

Also, Tetras is 100% correct with 3200 being easily supported.

In this new video, HUB used very fast DDR4 ram.
 
JollyJamma said:
....In this new video, HUB used very fast DDR4 ram....
Click to expand...

If you're referring to the DDR4 4000 he used then this highlights HUB's lack of knowledge when it comes to memory. The bandwidth of the 4000 memory he used might be a little ahead of the 3200 but the all important latency would have been much worse. Plus the 4000 would have been in gear 2.

This is shown in the Watch Dog Legions graph where you have DDR4 3200 out performing DDR4 4000 - which to anybody with more than a superficial knowledge of memory is a little daft.
52366572890_16ec132091_z.jpg


If he really knew his beans then he wouldn't have used a CPU that locks the SA voltage in the first place as by doing so there was no point in using DDR4 4000 ram. If he wanted to show the difference DDR4 4000 made then he should have simply taken those same B die 3200 CL14 and clocked then fairly simply to something like C16 4000. (using a better CPU that doesn't lock the SA of course)
 
Last edited:
MartinPrince said:
If you're referring to the DDR4 4000 he used then this highlights HUB's lack of knowledge when it comes to memory. The bandwidth of the 4000 memory he used might be a little ahead of the 3200 but the all important latency would have been much worse. Plus the 4000 would have been in gear 2.

This is shown in the Watch Dog Legions graph where you have DDR4 3200 out performing DDR4 4000 - which to anybody with more than a superficial knowledge of memory is a little daft.
52366572890_16ec132091_z.jpg


If he really knew his beans then he wouldn't have used a CPU that locks the SA voltage in the first place as by doing so there was no point in using DDR4 4000 ram. If he wanted to show the difference DDR4 4000 made then he should have simply taken those same B die 3200 CL14 and clocked then fairly simply to something like C16 4000. (using a better CPU that doesn't lock the SA of course)
Click to expand...
He is aware of gear mode 2 and running it in an FCLK ratio of 1:1 so I trust him to have done it properly.
 
JollyJamma said:
He is aware of gear mode 2 and running it in an FCLK ratio of 1:1 so I trust him to have done it properly.
Click to expand...

'Properly' would have meant not trying to use DDR4 4000 and then making a deduction about it. If his intent was to make DDR4 look as bad as possible then it was done 'properly'.

He's only served to hoodwink people like yourself into thinking "he even used very fast DDR4 memory", when in actually he did nothing of the sort.
 
MartinPrince said:
'Properly' would have meant not trying to use DDR4 4000 and then making a deduction about it. If his intent was to make DDR4 look as bad as possible then it was done 'properly'.

He's only served to hoodwink people like yourself into thinking "he even used very fast DDR4 memory", when in actually he did nothing of the sort.
Click to expand...

I knew something wasn't right with this whole thread and it's this. Intel can and does run DDR4 4000 at gear 1 and does see a performance improvement.

here's a techspot review by Steven Walton of DDR4-4000 at gear 1 and gear 2 (i've extracted one image because I don't expect you to go to the review site here) with an extract of the review stating "For anyone looking to maximize the value of an Alder Lake processor, DDR4 memory is what you'll be after, and there isn't any point as far as we can tell going beyond DDR4-4000. So if DDR4-3200 CL16 and DDR4-4000 CL18 are priced fairly similarly in your region, you might as well go with 4000."

cp1.png


Also, HUB has been historically very good at providing both Intel and AMD DDR4 reviews with manual v automatic timings for speeds ranging from 3000 to 4000 (see below) so I don't think you have understood the HUB review method.


Also, I've manually tested 4 by 8gig 3200CL14 Samsung B die versus 2 by 8gig 3600 CL14 versus 32gig dual rank 3200 CL16 Samsung (I think it's S die) and the dual rank kit is the fastest so "faster' RAM isn't always the best. Rank also has a significant performance impact, regardless of tertiary timing improvement for the other kits.

I really think you need to revaluate your understanding of DDR4 as I don't think I'm the hoodwinked one here.
 
MartinPrince said:
If you're referring to the DDR4 4000 he used then this highlights HUB's lack of knowledge when it comes to memory. The bandwidth of the 4000 memory he used might be a little ahead of the 3200 but the all important latency would have been much worse. Plus the 4000 would have been in gear 2.

This is shown in the Watch Dog Legions graph where you have DDR4 3200 out performing DDR4 4000 - which to anybody with more than a superficial knowledge of memory is a little daft.
52366572890_16ec132091_z.jpg


If he really knew his beans then he wouldn't have used a CPU that locks the SA voltage in the first place as by doing so there was no point in using DDR4 4000 ram. If he wanted to show the difference DDR4 4000 made then he should have simply taken those same B die 3200 CL14 and clocked then fairly simply to something like C16 4000. (using a better CPU that doesn't lock the SA of course)
Click to expand...

You've missed the annotation pointing out dual rank versus single rank. Dual rank significan't improves performance. Whoops. Looks like someone doesn't read their charts properly.

The performance different of 2 FPS between the single rank kit to DDR4-4000 can be a run to run variance and the game may prefer low latency RAM over faster RAM.
 
Last edited:
You're not really getting the point are you?

Let me make this simple - no person who really knows about memory and it's timings would do a test where they have DDR4 3200 and then throw in DDR4 4000 but do it in such an inept way that the latency on the DDR4 4000 is worse than the DDR4 3200!

Ask anybody who knows a modicum about memory; you would never pay the extra for DDR4 4000 and run it with such slow and loose timings that it turns out slower than DDR4 3200.
The only people that buy DDR4 4000 and run it with such loose timings are those that have been hoodwinked by the hype and the higher cost.

It just serves to demonstrate a level of ineptitude or ignorance only more compounded by those people that try to defend it.
 
Last edited:
So, to sum this up. We should check many different sources for our tech information. Which I think many people already do. Plus, only getting your information from a single source could indeed be incorrect. As mistakes can be made from time to time.
 
JollyJamma said:
You've missed the annotation pointing out dual rank versus single rank. Dual rank significan't improves performance. Whoops. Looks like someone doesn't read their charts properly.

The performance different of 2 FPS between the single rank kit to DDR4-4000 can be a run to run variance and the game may prefer low latency RAM over faster RAM.
Click to expand...
Dual ranks impact varies depending on platform, alder lake doesn't have as much of a difference between single and dual rank as say skylake based cpus, especially CML.
 
1st, both these tests use games that are very rarely if ever ram-speed limited. Why not use a RAM benchmark (i.e. AIDA64) or a game that in certain conditions is 100% ram-speed limited (like factorio with a super massive factory).

2nd, using a 12100F to test DDR4 4000? Wtf? No chance in hell of fully utilizing that ram's speed with the bottom-of-the-barrel IMC. Is the reviewer a moron?

There are two main metrics, memory bandwidth and memory latency, most games are not really affected by either, and the rest are usually affected by one and not the other. These two factors should ideally be investigated separately with a variety of games (and on a machine that can fully utilize the fastest ram) to really see what's going on.

I've tuned my RAM because a) I enjoyed learning how to do it and b) I have a massive factorio base and it makes a difference there if nowhere else. Currently running Samsung B-die, 32GB DDR4 3600MHz 15-15-15-32-275-2T-Gear 1
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom