The Workshop - You're doing it wrong (in my opinion)

[theme1256]

Hi There!

In my opinion you (Luke) are doing it wrong in the workshop. Love the concept, but would like to see the following things changed, it might change the output of the episodes.

I'll go trough the 3 episodes: fan placement, cable management and thermal compound application.

 

One thing I have to comment on all of the episodes is, that I would have used a hotter CPU, like the 4770K or 4790K, they put out way more heat, compared to the 6700K.

 

The "Case Fans" episode I don't have any gripe with, you did it right (scientifically), you kept everything other than the amount of fans constant. Nice! Tho, still a hotter CPU, would be better.

The "Cable Managment" episode I do have some comments on.

First, you didn't check how the GPU boosted, and as we know it boosts when it has the cooling headroom, so it would be interesting to see how that changed as you stuffed stuff into the case.

I personally have seen benefits from cleaning my cables, tho that was in a Lenovo M-series case (it's s**t). My CPU dropped about 5 C.

And again the hotter CPU, please.

 

The "Thermal Compound" episode I also don't have any gripe with, other than the hotter CPU.

 

Hope you change the CPU to a hotter one and when checking GPU temps you log how much it boosts

 

Regards

A long time (lurking) viewer

[don_svetlio]

And a open-air GPU cooler

[shymi]

On the topic of the hotter CPUs - should've used AMD ones. Some of them give away a ton of heat and are easily overclockable for even more heat.

[Naeaes]

I cringed a bit when they did only that short test in the cable-management test. Poorly managed case lets dust build up on the wires and which can dislodge and be sucked to the GPU or CPU heatsink. A huge dustball all of a sudden hurts more than evenly spread layer of dust over time in my opinion. I just mean, leave it for a year and re-run the test. Also the mastercase is one of the best-case scenarios in terms of cooling. Do the same on a tiny case with poor optimization, maybe.

[Yoinkerman]

Those are pretty minor changes to claim he's doing it wrong imo.

[brandishwar]

2 minutes ago, Yoinkerman said:

Those are pretty minor changes to claim he's doing it wrong imo.

Except he is doing it wrong because he's not controlling for variables and making wide generalizations from limited testing.

 

They're testing on the latest-gen CPU instead of one that runs hotter. They're testing on only one CPU instead of having several systems. They're trying to say that cable management doesn't matter except to make a system look pretty, without checking whether the processor and/or graphics card might be throttling to maintain a lower temperature. Same with the CPU thermal compound application. It didn't seem like they check for fan speeds or throttling. Go back several generations and you'll have systems and components that didn't have throttling as a safety measure. At minimum they should've disabled any thermal throttling options in the BIOS if that's configurable (motherboard dependent, I believe), and set any fans to run at a constant percentage if the fan is PWM. For the GPU, they should've also seen about disabling any throttling if possible, or checked if there was a lesser boost clock compared -- again, safety measures to protect from overheating.

 

Their fan placement test was reasonable in that it showed that fan placement can make a different to temperatures, and that there's a threshold beyond which more fans won't make much of a difference. They also weren't using the greatest CPU cooler on the block for that test, so any differences were more pronounced. But they still should've controlled for any safety measures the BIOS would take to keep the CPU from cooking.

[theme1256]

1 minute ago, brandishwar said:

Except he is doing it wrong because he's not controlling for variables and making wide generalizations from limited testing.

 

They're testing on the latest-gen CPU instead of one that runs hotter. They're testing on only one CPU instead of having several systems. They're trying to say that cable management doesn't matter except to make a system look pretty, without checking whether the processor and/or graphics card might be throttling to maintain a lower temperature. Same with the CPU thermal compound application. It didn't seem like they check for fan speeds or throttling. Go back several generations and you'll have systems and components that didn't have throttling as a safety measure. At minimum they should've disabled any thermal throttling options in the BIOS if that's configurable (motherboard dependent, I believe), and set any fans to run at a constant percentage if the fan is PWM. For the GPU, they should've also seen about disabling any throttling if possible, or checked if there was a lesser boost clock compared -- again, safety measures to protect from overheating.

 

Their fan placement test was reasonable in that it showed that fan placement can make a different to temperatures, and that there's a threshold beyond which more fans won't make much of a difference. They also weren't using the greatest CPU cooler on the block for that test, so any differences were more pronounced. But they still should've controlled for any safety measures the BIOS would take to keep the CPU from cooking.

Exactly! Tho I forgot to mention the constant fan-speed part

[Yoinkerman]

I think you're overestimating how those changes will affect results.

 

Also, they're not a peer reviewed scientific journal, nor do they need to be to produce meaningful results.

 

Cpus don't throttle until they get too warm or don't have enough power delivery.  Nor do they throttle to stay at the temps displayed in the video.  

 

I'm not saying your criticism isn't valid, just that you're overestimating the importance.

[Prysin]

43 minutes ago, Yoinkerman said:

I think you're overestimating how those changes will affect results.

 

Also, they're not a peer reviewed scientific journal, nor do they need to be to produce meaningful results.

 

Cpus don't throttle until they get too warm or don't have enough power delivery.  Nor do they throttle to stay at the temps displayed in the video.  

 

I'm not saying your criticism isn't valid, just that you're overestimating the importance.

Neither is the Mythbusters. Which is a TV show, which ALSO could have gone sloppy (despite catching flak for it) and kept going. Instead, said TV show goes to great lengths to stay CONSISTENT and to ELIMINATE OBVIOUS VARIABLES.

 

none of which has been done in several of the latests LTT videos.

[Yoinkerman]

10 minutes ago, Prysin said:

Neither is the Mythbusters. Which is a TV show, which ALSO could have gone sloppy (despite catching flak for it) and kept going. Instead, said TV show goes to great lengths to stay CONSISTENT and to ELIMINATE OBVIOUS VARIABLES.

 

none of which has been done in several of the latests LTT videos.

Sure they can be improved.  Nowhere did I say they couldn't

[brandishwar]

14 minutes ago, Prysin said:

Neither is the Mythbusters. Which is a TV show, which ALSO could have gone sloppy (despite catching flak for it) and kept going. Instead, said TV show goes to great lengths to stay CONSISTENT and to ELIMINATE OBVIOUS VARIABLES.

 

none of which has been done in several of the latests LTT videos.

And at the same time, they also revisited their testing when problems have been pointed out to them or angles they didn't consider -- such as by the President of the United States. So the question is whether LTT will do so. Many look to them as an authority on PCs. That basically places an onus on them to get things right to maintain their reputation. They're good with product reviews, and seem rather thorough from the product reviews I've watched, but when it comes to these videos where they're dispensing information, frankly they're getting things wrong in rather trivial ways -- performance rather unbecoming of professionals.

 

Not thinking that a variable will matter or matter much, @Yoinkerman, does not mean the variable should not be controlled. Absolutely you should still try to control as many variables as you can account. They're called "variables" for a reason. If you can control for it, you do so. If you can't, you mention the variables you could not control, since those variables, while accounted, could still affect the results. And, again, as many have looked to LTT as an authority on PCs, that implies an obligation to get it right since their audience will be relying on that information down the line.

[Yoinkerman]

Again, sure they can make better videos.  However, nothing they have put forth with the 3 videos mentioned should be tossed and give credence to the claim they're doing it wrong.

 

There are huge myths out there regarding how many case fans people need and how air flow can get "restricted" in completely run of the mill mid tower cases.  As well as with thermal paste and cable management.

 

I haven't watched the newest delid video though, it might be different.

[Prysin]

28 minutes ago, Yoinkerman said:

Again, sure they can make better videos.  However, nothing they have put forth with the 3 videos mentioned should be tossed and give credence to the claim they're doing it wrong.

 

There are huge myths out there regarding how many case fans people need and how air flow can get "restricted" in completely run of the mill mid tower cases.  As well as with thermal paste and cable management.

 

I haven't watched the newest delid video though, it might be different.

its not only the latest videos either.

Take the RAM speed testing Linus did back in 2014/early 2015....  Then Digital Foundry, and @MageTank, within a short short timeframe BOTH disproves LInus's findings AND conclusion. Other tests also suggest RAM speed does matter in games, some games more then others.

 

Have @LinusTech, redone his video in light of this or even commented on these findings publically?? No.....

Meaning every person who watches that video (as it is readily availible to the public) will be misinformed.

 

Whilst his intentions may be good, the execution of these videos made by LTT as a company is lacking. And it may seem like the videos, whilst intended to help the general public, may in some cases end up hurting the general public by giving them the wrong answer.

[MageTank]

3 hours ago, Prysin said:

its not only the latest videos either.

Take the RAM speed testing Linus did back in 2014/early 2015....  Then Digital Foundry, and @MageTank, within a short short timeframe BOTH disproves LInus's findings AND conclusion. Other tests also suggest RAM speed does matter in games, some games more then others.

 

Have @LinusTech, redone his video in light of this or even commented on these findings publically?? No.....

Meaning every person who watches that video (as it is readily availible to the public) will be misinformed.

 

Whilst his intentions may be good, the execution of these videos made by LTT as a company is lacking. And it may seem like the videos, whilst intended to help the general public, may in some cases end up hurting the general public by giving them the wrong answer.

I've called him out on it several times. No go. Sadly, his video is the current bane of my existence at the moment, as it's everyone's go-to source to "disprove" my findings. It's not even my findings. It's not even DigitalFoundry's findings. This has been known for quite some time. The thread i keep linking on OCN was done over a year before the DF video was released. I had a friend on Skype making that claim before even that thread came out. I am just as guilty as everyone else that makes the "memory speed doesn't matter" claim, because I too made those same claims before i tested it myself.

 

That's the point of this post too. Do not just take someones word for anything. ALWAYS test it yourself. Find out for certain if it helps. If it does not work for you, request their testing methodology so that you can repeat it. If they fail to share it, then clearly they are not trustworthy. If anyone doubts the memory thing, it's easy to test. 

 

I only wish people would take the 30 or so minutes it takes to do this simple test, and help spread the word. Until @LinusTech takes another look at this subject, and does it properly (no GPU bottlenecks, monitor minimum FPS, don't have terrible memory bandwidth with quad channel, etc) then I will have to keep repeating myself. Luckily, I reached out to a few newer tech reviewers to give this test a try, so that we can get more sources on this test. I only told them to monitor minimum frame rates, and to test the impact with XMP profiles only (no manual overclocking) so that it is fair. I did not tell them which games to run either. That way, viewers will see the impact on random games, and if its worth it that way. 

[brandishwar]

5 hours ago, MageTank said:

I've called him out on it several times. No go. Sadly, his video is the current bane of my existence at the moment, as it's everyone's go-to source to "disprove" my findings. It's not even my findings. It's not even DigitalFoundry's findings. This has been known for quite some time. The thread i keep linking on OCN was done over a year before the DF video was released. I had a friend on Skype making that claim before even that thread came out. I am just as guilty as everyone else that makes the "memory speed doesn't matter" claim, because I too made those same claims before i tested it myself.

Part of the trouble here is that most people aren't going to test the claim because largely they're not interested in whether the claim is true or false, but merely what they're told. A lot of people who discuss computer topics online know just enough to be "dangerous", so to speak. They know a little and think they automatically know a lot, and think their ability to search Google is a viable substitute for the years I have actually building computers. I conversed with one such person on this forum and was able to readily show him how little he actually knew.

 

As I've already said, many see Linus as some kind of "authority" on PCs without realizing that he is one of these "dangerous" people. His latest video on YT for reflowing solder on PCBs shows that. He's a content creator. He looks good in front of the camera (to a degree). To the best of my knowledge, he was never a recognized expert with regard to PCs and electronics, though he comes across as such, and his charisma and his videos provide that air of authority as well. And Luke's videos aren't any better with regard to information. I'd love to know how they script some of those videos, and whether they consult actual professionals for doing so, or if they just search Google.

 

Their videos do, however, play to people looking for a quick answer to a complicated question. They largely don't want to think, but they will take whatever LTT says and then turn around and act like they know what they're talking about.... I've seen that time and again as well.

[royalcrown]

Okay, he didn't control for all the variables in cable management. Practically however, the majority of users aren't running cases with 80mm fans and IDE ribbon cables anymore. A couple power supply / sata cables isn't going to make any real difference in your temps unless your case sucks and temps are marginal to begin with. Better to upgrade the case at that point. It's just like overclocking, hundreds of dollars on radiators and blocks for a 15 percent overclock that makes a 5 percent difference in real life. For the majority of people, not an equitable tradeoff.

 

[theme1256]

5 hours ago, royalcrown said:

Okay, he didn't control for all the variables in cable management. Practically however, the majority of users aren't running cases with 80mm fans and IDE ribbon cables anymore. A couple power supply / sata cables isn't going to make any real difference in your temps unless your case sucks and temps are marginal to begin with. Better to upgrade the case at that point. It's just like overclocking, hundreds of dollars on radiators and blocks for a 15 percent overclock that makes a 5 percent difference in real life. For the majority of people, not an equitable tradeoff.

 

I was running just that, a couple of power and sata cables and saw an improvement after cleaning it.

[royalcrown]

On 3/1/2016 at 7:21 AM, theme1256 said:

I was running just that, a couple of power and sata cables and saw an improvement after cleaning it.

I'm not saying there isn't ANY, but really, how much improvement did you actually see ? For myself with 3 x 140 mm case fans and 120 mm cooler I seriously doubt it'd drop my temps by enough to be worth it. Now I'm more than into doing it someday for aesthetics when I install a window.

[theme1256]

8 hours ago, royalcrown said:

I'm not saying there isn't ANY, but really, how much improvement did you actually see ? For myself with 3 x 140 mm case fans and 120 mm cooler I seriously doubt it'd drop my temps by enough to be worth it. Now I'm more than into doing it someday for aesthetics when I install a window.

About 5C lower temps on the CPU. This was with no front fan and a 92mm rear fan.

[royalcrown]

14 hours ago, theme1256 said:

About 5C lower temps on the CPU. This was with no front fan and a 92mm rear fan.

I could see that working then, I just suppose most people have front fans, but there probably is a lot of hardware with no front fans. Cool, good to know that actually !

[MEC-777]

Using a hotter CPU wouldn't have made any difference on the delta T observed. It's only in the extreme cases (waaay too much or barely any TIM) that it has an effect - which was demonstrated by Luke's testing. 

 

The only problem I had with one of his experiments was the cable management with air flow blockage test. he should have used an open air cooled GPU, NOT a blower style GPU. Of course temps won't change much with the blower GPU because both the CPU and GPU are pulling their own clean air in those situations and expelling the hot air out the back. An open air GPU will expel it's hot air all around the GPU inside the case, creating a stagnant hot air pocket around the GPU - which the CPU cooler will also start pulling from. I had a Node 304 with an open air cooled R9 290 in it. I know what can happen with there's not enough airflow. lol.