Thread for LTT Labs Test Suggestions

[Indian pc builder]

say someone is considering switching over from one WiFi setup to another, they would test the range and speeds and provide a side by side comparison so someone who is switching can get an idea of wether it will give the same coverage and speeds or should he consider something else. The assumption would be that one can get an india for the relative range of one to the other. As the comparisons and testing from the labs would be in no interference conditions, and the interference at a place where such a setup would be deployed would affect both devices the same way, there would be an easy relative comparison 

[thisisjaymehta]

Test and compare Real SAR values with SAR values given by phone manufacturer. (Like on android when you do *#07# it shows SAR value).

[HKPolice]

Hey Linus can you please repeat this experiment with modern NVMe drives? 

[WeiSheiLindon]

As part of your phone test suite, you should test real framerates in video recording. Many phones will not record in the framerate you set them to once they start to thermal throttle. It would be great to know which phones actually did what they said they were doing.

[LTTMobileJake]

3 hours ago, WeiSheiLindon said:

As part of your phone test suite, you should test real framerates in video recording. Many phones will not record in the framerate you set them to once they start to thermal throttle. It would be great to know which phones actually did what they said they were doing.

I'm planning this once our camera testing room is built!

[Altrius II]

Posting this here instead of letting it die in pilot chat on discord:

I would find it interesting if Labs tested encoder / decoder performance on GPUs, either as a standalone test or as part of a review dataset. Whilst it would be difficult to explain the results in a way which a general user might understand, it would almost certainly be an interesting test for those of us who run plex servers and actually have a use case in which that would be beneficial information to know before buying a GPU.

 

Thanks Jake for giving it the time of day in the Discord

[Dogzilla07]

After watching the latest HWBusters/Cybenetics + GamersNexus, I wrote this piece on a SFF discord channel, and just wanted to post it on reddit as well, sry for the wall of text
 

With Mike Chin from SPCR + Steve from GamersNexus and Aris from HWbusters/Cybenetics peer reviewing each others frequency analysis, results from anyone else who doesn't join them are effectively not accurate enough
 

I mean Hwcooling is infinitely more accurate compared to Igor's Lab, and a very good source of subjective conclusions and experience, but a test without peer review is not objectivelly accurate
 

Ideally Hwcooling would fully cooperate with GamersNexus HWbusters and Mike Chen to bring their fan specific frequency expertise, and then all 4 would settle on a methodology and explanations for frequency testing

And the beauty of it, it doesn't require them to have expensive machines, or expensive chambers, if the cooperating is only for frequency methodology and Hz impact explanations (+ if HWcooling builds/buys their own hemi-anechoic chamber, then they'd be fully equal)
 

Additionally, there could be two Tiers, or two Peer Classes

Hemi-Anechoic Chamber Tier/Class at 6dB SET/~10-14dB GET noise-floor @ 1m

Semi-Anechoic Chamber Tier/Class at ~15-20dB SET

And then whoever is the Lead authority on the 2nd Tier/Class can form a "club" set standards for Expreview, HardwareCanucks, Major Hardware, LinusTechTips, Thermaleft, Igor's Lab, etc, ...

SET = ideal/intented/theoretical noise-floor for a chamber

GET = actual noise-floor for Hemi chamber due to outside noise/vibration differences

[Theweedwarf]

Testing GPU expansion for laptops with thunderbolt 4 would be great. 

[Ferrex]

Test Apple's $130 thunderbolt 4 cable VS cheaper alternatives on your expensive cable testing machine 

 

https://arstechnica.com/gadgets/2023/10/apples-130-thunderbolt-4-cable-could-be-worth-it-as-seen-in-x-ray-ct-scans/

[jufy111]

With energy prices increasing across the world, I think idle power metrics are becoming more and more relevant to many people. Just about every outlet reports maximum system power that occurs load during benchmarks, however there is very information available on idle power usage.

Idle power stats are particularly relevant for people who may be running a NAS or home server on consumer hardware, which may spend a lot of time idling. But this is also relevant for every day people who may run gaming PC, but day to day usage.

Below is a number of proposed tests to test power usage. The number of tests may below may be excessive, but I think include at least idle power draw would be highly beneficial.

• Idle Load,
  • System tested on a standardized OS image

  • May have to be left for some time to allow for the PC to drop into lower C-states

• Minimal Load,
  • Full screen video playback (1080p YouTube)

• Moderate Load
  • 2x 1440p monitor

  • Full screen 1440p playback on 1 screen

  • Some other tasks that will put moderate load on the CPU

• High Load
  • Gaming benchmarks

• Extreme Load
  • Make both CPU and GPU hurt (CineBench + FurMark)

Obviously, there is an infinite combination of CPU/MOBO/GPUs, however a standardized set of hardware could be used for generation.

For example, a standard test of hardware could be used to test power consumption for a variety of motherboards. There are a million different motherboards out there but one test for each chipset would be great.

• ·         Stock Motherboard settings (or maybe stock + DOCP/XMP enabled)

• ·         Standard number of fans (i.e., 3x120mm Noctua S12A)

• ·         RGB Disabled (sorry gamers)

• ·         Standard GPU (RTX3060?)

• ·         Standard 2x16 GB Memory sticks

• ·         Standard storage (1x M.2. NVME SSD)

 

For each series a standard tier of CPU for example

·         Intel 400 series (10400,12400,13400)

·         Ryzen 600X series (3600X, 5600X, 7600X).

Alternately, you could test with no standardized hardware going forward, and then publish all the system hardware to something that can be easily searched and filtered. (Similar to the below google doc warning: It’s in German).

https://docs.google.com/spreadsheets/d/1LHvT2fRp7I6Hf18LcSzsNnjp10VI-odvwZpQZKv_NCI/edit?pli=1#gid=0

 

 

[Mattwilson83]

PSU cables including rgb cables. There’s so many cheap options but hard to find we’ll know branded cables. 

[Paslo]

Hi! Just made account in here to mention that i would be really pleased to see some kind of CPU test in BEAM NG car physics game.

It can kick the s**t of any cpu coz u can just add more ai cars in there and to make chaos :D

 

In that game you can add Ai traffic and it' super heavy. With my old i7-8700k i can run about 12 cars decently and as my cpu is 6core/12threads i assume game can run 2 cars with 1 core, but what if you have some efficiency cores etc etc.... 

 

I wanna see difference between best i9 and best amd (both being 32threaded[or hopefully same amount of threads])

 

Thanks and hopefully someone sees this and gives it a go or if someone has 32threaded cpu give it a go and give information when ur fps drops to unplayable when u add cars.

 

sry english not my native language

 

happy weekends!

 

[manikyath]

it's an interesting idea, but i see two major flaws:

- repeatability might be problematic, for a good test every CPU should be subjected to the exact same scene, this would require the engine to be 100% deterministic, and the sccene to be set up with a specific set of starting points... essentially building a 'canned benchmark'... at which point you may as well just use an off the shelf canned physics benchmark.

- it might be difficult to turn that into a usable metric aka "bar graph on the screen"

[DANK_AS_gay]

43 minutes ago, manikyath said:

it's an interesting idea, but i see two major flaws:

- repeatability might be problematic, for a good test every CPU should be subjected to the exact same scene, this would require the engine to be 100% deterministic, and the sccene to be set up with a specific set of starting points... essentially building a 'canned benchmark'... at which point you may as well just use an off the shelf canned physics benchmark.

- it might be difficult to turn that into a usable metric aka "bar graph on the screen"

There's a built in physics benchmark in BeamNG drive

[DeadOnArise]

Don't really know where to put this, been a long time yt watcher but green grass here. Anyways. 

 

Ark Survival Ascended has been quite the challenge to many gamers recently with resources. Felt like I shoul just bring this up. 

 

Memes are its new crysis

[LogicalDrm]

21 hours ago, DeadOnArise said:

Don't really know where to put this, been a long time yt watcher but green grass here. Anyways. 

 

Ark Survival Ascended has been quite the challenge to many gamers recently with resources. Felt like I shoul just bring this up. 

 

Memes are its new crysis

Merged to Labs suggestions.

[Kheelan]

Is there a way to get testing done to see if monitors are as color-accurate as they clam to be? 

[Guille.R]

I’d love to see some machine that tests durability on folding phones, something that opens and closes it a lot of times and see how it evolves.

[KrazyMoose]

I'm heading to a concert soon and not wanting to damage my hearing, I looked up ear plugs for concerts and discovered that "high fidelity music ear plugs" are a thing.

What I think would be interesting is to test the manufacturer claims on actual noise reduction, foam vs silicon and affect on music quality. Some manufactures even have tables that outline noise reduction at each frequency.

[HammondComputers]

Amazon laptop power supplies and batteries vs oem

 

ive that often the cheap no name power supplies get very hot, dont always supply the correct power, are very easy to break apart (for swapping cords because im cheap) compared to oem versions or even Targus universal power supplies (which you can buy for 10$ or less on ebay, im guessing they get a bunch of returns because they only give you 5 tips in the box now. So super cheap open box deals.)

 

no name batteries, easier to take apart, missing safety features, the controller doesnt always match the cell capacity. not that you should, but maybe you bought a bunch of cells for pennies when Radio Shack went out of business and also discovered that the generic controllers often accept cell swaps where the oem ones do not.

 

The  benefit to the community and the entire planet.. (lol) is not testing the generic supplies and batteries for performance but the safety features. Is it dangerous to our equipment or is it a fire hazard, when they are missing temp sensors, or over voltage protection, or even under voltage when trying to charge a lithium cell that has fallen to low in voltage to safely charge?

 

also not really related except also being sold on Amazon. Cat 7 patch cables.  People dont seem to realize that Cat 7 is older then cat 6 and not really related to 5,6,6,8 etc.  Sure cat 7 was great for long video runs, but not certified by the internet gods.  Maybe a real life test vs cat 6 and 8 and a video explaining what cat 7 even is?

 

 

[Jdude2005]

I saw a video on YouTube where they utilized industrial CT scanning on different charging cables and were able to see the individual transistors on the pcbs and all the different layers of complexity built in. It was a really awesome technology that could be utilized in many videos for LMG. I'm not sure if links are allowed because I haven't read the rules yet, but the video I watched was https://youtu.be/AD5aAd8Oy84?si=zwGK7N4WVSaDTAGy It is part of a sponsored series with Lumafield. I know Linus is really good at spending money and getting sponsors and this technology looks like it would be exactly something the Labs team would love to play around with. 

 

 

On a side note, I love the work everyone at LMG and partner companies do to make informative and entertaining content for the world. Keep up the great work. I've been watching since before the very first whole room water cooling video 3 houses ago. Still a fan and recommend the channel to anybody with any tech questions.

 

[TerpMike28]

About a year ago, the main LTT channel did a review of the various dashcams available on the market at the time. Since that video was released, Sony has released it's v2 Starvis sensor that supposedly significantly improves image quality on dash cams, especially at night. With a relative lack of independent reviews/standardized testing of the new tech out there, it's hard to determine which camera to purchase. Having the labs produce a follow-up video with standard testing across all cameras being tested at a variety of price points would be helpful for consumers as the prices for these cameras can run in the hundreds of dollars.

[yesiamzach]

Screen polarisation. I got a new pair of polarised shades and now I have to tilt my head 45° right to see my phone at full brightness (pixel 7)

[Tan3l6]

How about thermal paste "Aliexpress edition" test.

There are at least four thermal pastes, native to Aliexpress that promise very high thermal conductivity:

1) CTG9 - 13.5W/mK

2) ZF-12 - 12W/mK

3) LK-17  - 17W/mK

4) Coxbyte CG-50 - 18.2W/mK

 

I do understand that the advertised thermal conductivity in this case is (marketing) bs, but still would be good to know if any of them is at least decent.

Maybe a bit narrow scope/niche for LTT labs though.

[bdc]

I'd like to see the LTT Labs get their hands on some Frore Airjets and for you to revisit the ultimate Steam Deck Cooling challenge (or the Rog Ally/Legion Go if their design works better).

 

Unfortunately the AirJet Mini dimensions are too wide to be able to fit two side by side into the Steam Deck fan cutout which may have allowed 4 of them to be used? It is probably the same for the other systems too. Anyway it would still be really cool to see some testing done and see the impact on temps, battery life, noise, performance and the amount of effort required to do the mods for the following scenarios.

 

1. Mount 2 AirJet Mini's onto the SoC directly (or the heatpipe/heatsink, custom or stock)
2. Create a new heatpipe that wraps around into the area that the fan currently sits and mount 2 AirJets (4 if they fit), one on each side of the heatpipe.
3. Mount a 2nd pair of AirJets onto the heatpipe right above the CPU and combine with the setup in scenario 2.

Frore themselves have a video

with some details about doing just scenario 1 but it is a marketing video targeting manufacturers, LTT verifying the numbers they quote or trying other methods would be awesome.

This could be a multi-part video, maybe 2 videos on the Steam Deck and if 3 or 4 AirJets on the Steam Deck offered no performance boost over 2, then do another video with 4 AirJets on the Asus Rog Ally and see how a higher TDP device benefits.