How Memory Sticks are tested and programmed
Vložit
- čas přidán 28. 06. 2024
- Support me on Patreon:
/ der8auer
---------------------------------------------------------
Music / Credits:
Outro:
Dylan Sitts feat. HDBeenDope - For The Record (Dylan Sitts Remix)
---------------------------------------------------------
Paid content in this video:
- /
Samples used in this video:
- /
der8auer Advertisement Policy:
der8auer.com/advertisement-po...
---------------------------------------------------------
Timestamps:
0:00 Intro
0:29 RamCenter test stations
1:28 Test module & software
2:59 Smart I/O
4:06 Smart SPD
5:04 "Outro" - Věda a technologie
That was my last Computex Video. Had to relax for a few days after shooting too many videos in Taiwan. Next week we will continue with "normal" videos. As far as you can consider them normal :D Enjoy your sunday!
Thank you
Where do we get the software, ramCENTER rct-pro v2.00?
Buddy! Take some time for you :)
You're doing so much between Thermal Grizzly, CZcams, and your various partnerships! Man, you're such a driven person, mad respect!
I guess when you do what you truly love, it's never really work, eh!
This is so much better than all that "AI" bullsh!t in the big booths.
Wanted to write a similar comment. +1
Probably the most interesting Computex video I've seen - I actually learned something!
I remember seeing my dad going to a 7-11 in the 1950s with a handful of vacuum tubes from our TV. The store had a machine that tested the tubes for free. The machine would indicate if a tube was faulty. Below the tester was a cabinet stocked with new vacuum tubes. You removed the new tube(s) you needed and paid the store for them.
Sometimes they bring in those devices in our repair dept. That testing bench is totally different from with solid state devices. We are going way back then, lol.
Analog devices are resilient to failure and easy to diagnose. I wonder if each memory integrator customize their own testing equipment the way the ORCA company does.
That "conductive rubber" stuff is really interesting! It's almost certainly an "anisotropic" material where it's conductive the in Z direction but not X or Y (so you don't need to align it), and a primitive version of it is used in the "zebra strips" that connected old fixed-layout LCDs to things like hand calculators and other small electronics.
I wasn't aware that they made material like this with sufficiently fine conductive channels for connecting modern DDR4/DDR5 ICs, and that they can support enough current and isolation to maintain signal integrity for this kind of testing with modern high-speed circuits. Very cool!
Love how that ram tester looked like it out the 80s/90s and the software looks right out of the Window 98/XP era. Something comforting in that.
Yeah, I was staring at the design for so long.
Is there really no one left that is designing and coding for RAM anymore in the year +2000!?
Feels like we are losing so many engineers in the RAM department and none want to go and learn it all.
"If it ain't broke, don't fix it"
@@duccc Yes, Ducky that mantra is solid.
But we are talking about advancing the human master race here.
I believe you can test broken GPU's with a software that looks like that. Checking if the VRAMs are working or not.
@@N0N0111not much reason to. The software is only used by skilled/trained technicians and engineers. It has all of the information and controls needed front and centre. Plus, it works. Redesigning it would mean not only training everyone on the new software (which will probably temporarily increase the error rate due to incorrect usage), but also might introduce bugs that didn’t exist before. So why go through the effort when the existing tech works?
They do all that without AI? How are such things possible these days! 🤣🤣
I LOVE these videos, on tech and How it is done. Always wondered how do they test them, how they debug them. I wish you made more of these factory/engineer tours...
Just wondering, how could the "hotswappable" coupling module with rubber contact perform in daily operation. Just imagine the options.
You could like 50 RAM modules when new dies come out, swap it for the ones in your system, test which combinations performs the best and run it...
Maybe not as easily swappable as those in the video for increased stability and cooling, but ... damn, the TRUE modularity of even RAM modules on the sticks would be awesome.
Or having these test devices in local stores to test the faulty DIMMs, swapping the faulty modules and selling them as refurbished would save tons on shipping back and forth across the globe. And create local jobs. I am not really green fanatic, but in the RAM space this could be sensible, especially as you have so many modules per system in every PC... They could just order reel of modules from few major manufacturers and do the fixes without shipping it to other continent.
Problem is the cost of those IC test modules, I've been interested in buying them for years but the prices are insane just for 1.
And this is coming from a guy who has paid over $500 for DDR3, DDR4 and DDR5 kits
Wow this is by far the most interesting video from computex.
They didn't explain how they make the ram AI ready though
BLEH
This kind of booths are super cool and interesting.
This is honestly one of the most interesting videos from computex this year
G'day Roman,
WOW! this was so cool & along with GN Steve's catch up with Vince my 2 fave videos from Computex
Nice Video , it's always interesting to see out of the box procedures , the specific tools normally aren't visible or detailed . for the conductive rubber , we can find it a band in a lot of lcd displays to make the contact between the cristal/glass LCD and the pcb that controls it . the bands are a sandwich of rubber with carbon like pins/rods in the middle . in these case instead of a band it's a square , the production of this squares must be very specific , maybe layers side by side to achieve the pin array spacing
Thanks for sharing these type of technical videos.
that can make a good video, tools like that
That was pretty chill. Thanks DB!
My 7200 teamgroup can do 8200 but my 8200 vipers couldn't do 8200
And that’s why I’m using 5600 ram. I don’t need that extra few percent in exchange for instability. Mind you I bought it in late ‘22 when higher really seemed to have issues on many z690 boards. Wish it was 6000, but for a 12700k… it’s fine.
@@Akkbar21 For all the years I have overclocked my main gaming systems, I have learned this.
You find the desired speed (MT/s) you want to run your RAM.
High speed for high data transfer.
Medium speed for super tight timings with superb low latency.
The key, after you have found your settings is to find the golden voltage that your RAM will be stable at.
For DDR4 and DDR5 1.4v is the standard to run the RAM optimal for both scenarios. Always have your RAM temp below 55C in an hours burn in stability test.
After it has passed the +1 hour 99% load test, up your voltage with 0.01v, this is an extra save barrier for random voltage instability.
As last buy a motherboard that has Load Line voltage calibration settings for RAM, this will super extend RAM stability.
As final don't be afraid to read through some overclocking forums about RAM.
@@N0N0111 that’s great and all. I prefer to set xmp or whatever amd calls it and it’s rock solid and works.
@@Akkbar21 My experience's over the years for having a dialed in RAM that is load tested. will last you a super stable system with near to none blue screens.
That's cool stuff! Thanks for sharing.
A company I worked for had a memory tester that tested sim modules. Problem was that it was bloody expensive, but as the first DDR generation were just introduced it became useless as it could only test the older memory standard. it was bought when we had a lot of problems with the memories and it did it's work well. But as I said it became unusable when DDR became standard.
We also had a machine that washed motherboards. This was not often used, but it was bought when an entire batch of motherboards from one manufacturer turned out to have problems caused by badly washed motherboards from the factory. So we bought these cheap and washed them in that special machine. Well over 90% of them worked perfectly after that. The warranty period was one year for our customers so it was pretty important that they were reliable. The rate might have been close to 100% but I can't really say. It was however well over 90%.
A problem with test machines are the changing standards. That machine you bought for a lot of money suddenly is useless.
That was pretty cool. Thanks!
Really enjoyed this video and learning how ram is set up. Thanks
Looks like they took Memtest86+ and repurposed it
it's just the enterprise equivalent
I couldn't find it.
yea probably a GPL violation too. As it is released as GPLv2.
@@tuttocrafting If they're only using it internally, it's fine, no?
@@concinnus it seems that the company sell solutions for small ram modules manufacturer.
So I suppose that the software is sold with the zif kit.
This is so awesome. I've been wanting one of those DDR5 IC testers for a while. They're not cheap or practical but damn would it be cool to bin an insane kit for yourself. Maybe one day...
very interesting stuff, nice one
That's very interesting how RAM is actually certified and compiled for each speed. I wonder if companies like Corsair have a largely automated process or how theirs differs from a small company like this. I'm going to keep my eye open for their RAM modules here in Canada. It'd be neat to support a smaller company and get away from the grips of major PC companies, especially knowing just how thorough their testing and validating speeds are.
Neat to see! Thumbs up.
I find this very interesting! thanks for sharing.
also, is that an IDE cable being used to program te SO-DIMM modules board at the end of the video?
wow... my first 3 or 4 PC's used IDE cables..... man , now i feel old hahaha (im almost 39)
Great video!
I wonder what type of test is used for GPU VRAM
Cool video it was hidden from my feed but thanks for sharing
so basically all RAM is the same between companies, they just get it to that final step you showed and THAT makes it branded LOL - Nah I know there's more to it than just branding, but I guess I never thought about what goes into the different brands other than buying specific IC's from sk Hynix or Samsung or Micron ... I guess they could use different voltages, bin the ICs better, or even the quality of the PCB... so there are a lot of factors that go into each brand being different.... and you just opened up out eyes to how they figure that all out.. PRETTY FRIKKIN COOL!!!
Peak educational content
I`ll have To Remember this next time My Memory fails , wonder if i can make a sticky note for that , 🤔
What is that testing chassis? Can you buy this? I like it, it would be great for display in our shop. With or without custom cooling. And with or without RGB at the bottom.
This is very cool. Is how they actually bin high speed modules? Check each IC on the whole DIMM until they get a good combo of parts? I thought machines did all of this work.
Maybe for the bigger manufacturers. But smaller manufacturers often rely on more manual processes.
I wish der8auer can visit a GPU test site one day and make a video about it.
This chip/soldering tester could be on motherboards. Not with all the leds of course.
Very interesting. I was also intrigued by the fanless GPUs, but found that they are likely just Asus GT730 (or similar).
Is there a hardware solution for testing SSD ( SATA, M.2) ? Thank you
Are those Orca RAM any good? Do they have ANY overclock margin or they get boosted to the max like cheap RAM? Get them tested. Give us information! :D
speaking of "Testing"
Can you make a video about M.2 NVMe slot voltage tester?
There are several cases where a brand new M.2 SSD died so quickly because the slot itself gave too much Voltage
This could help regular users figure out what is wrong with the ram sticks. Probably expensive though.
Did you get some sweet sweet golden sample ram tho?? 😂
So you find the ONLY stand without IA?!
Can anyone give a hint how hard it is to DIY a JEDEC standard SPD? As a passion project I’d like to build my own unbuffered ECC DDR5 DIMMs with DRAM ICs harvested from various overclocking DIMMs to get modules that do higher frequencies and tighter timings.
It's actually easy, the expense is in the module testers themselves but it can be done without one using individual sticks and testing per IC. Has been done before by guys in the XOC scene
Thanks for the information!
RAM programmed via EPROM?
test software reminds me of memtest...but customized 😄
Why was there no AI memory training, I am so disappointed! /s
Framchasers....... You know you want this haha......
God I wish stuff like this existed in my country. America COULD do it. We are smart enough. Rich enough. We just don’t. Too many consumers, too few creators. Proper leadership can change that. It sure isn’t Trump.
Lol
It's MHz
I'm here to see the feline.
2:05 oh look a GPL license violation on video.
Absolutely based on the MemTest86+ that as far i know is released under GPL v2 license.
Who is the manufaturer booth? So I know that I will never have to buy from them...
Why should you care?
@@basshead. As a FOSS developer I hate when my work (not in this case) is used in violation of the license.
Indulging in this behavior is wrong. Both for the people that works on such project and for the industry.
Unfortunatly most ASIA brands don't care about liceses and this is something that will have to chage.
@@basshead.Because there are legal obligations for every company. Imagine if there's no copyright protection anywhere...
I find it hard to tell it being based on MemTest86+. I mean sure the interface looks somewhat reminiscent but for all we know code base could be completely different
@@yensteel We would get cheaper and better products.