Feedback

[RunDLL]

Dear Linustechtips,

 

first of all i want to say that i like your videos very much and before all else the "Big Move" videos were great.

 

The problem is that your hardware reviews, especially PC-Parts (your phone reviews are great btw.), are so superficial that i only watch them for entertainment purposes and not for actual hardware insights. In your Intel 6700K review for example you benchmark games in 4k.... You should bench CPUs in 1280x720+low settings or 800x600 so your GPU won't bottleneck at some point. You could also bench Arma III for single CPU core load to show differences between CPU generations in terms of performance per Mhz etc.

 

I think your reviews would greatly benefit from more thorough Benchmarks and actually wouldn't be much longer if you could add 2-3 more Benchmark slides. 

 

 

[Windows2000]

I think the GPU bottlenecks or any control group is not an issue on their benchmarking videos as they have used same (or similar) testbenches throughout the tests.

Also, AIDA64 or Cinebench is the INDUSTRY STANDARD. 

I don't see a problem with Linus's methods.

[BoredErica]

You should bench CPUs in 1280x720+low settings or 800x600 so your GPU won't bottleneck at some point.

I don't really agree. If we want to look at how CPU affects performance in CPU bottlenecked games we should benchmark CPUs in CPU bottlenecked games, not make a GPU bottlenecked game artificially CPU bottlenecked game. I don't know what effect that has on the results.

 

There is some talk, unproven talk, that ram speed might matter for Skylake in CPU bottlenecked games. After all, the latency on DDR4 is hilarious. It's also been suggested that for Skylake a higher ram speed might matter relative to CAS more than it used to with the new architecture. Again, I want to emphasize, unproven. But it would be nice to see somebody prove or disprove this.

 

Another way to look at CPU impact is to look at FPS differences in single vs dual GPU configs, with the latter requiring more CPU work. 1080p 144hz or 1440p 144hz is a thing, it's not always about high resolution. And minimum FPS, and frame consistency.

[Chugle]

-snip-

 

Arma III should be the new standard. 

[Picky Peep]

GPU bottlenecks aren't really a thing.

A 5960X bottlenecks 3 Titan Xs in SLI, the Titan X's can't shine because the 5960X isn't extreme enough

[BoredErica]

GPU bottlenecks aren't really a thing.

wat? Might want to elaborate, because most games are primarily GPU bottlenecked.

 

If you're saying that GPU bottlenecks aren't a thing because escalating the hardware to tri-sli Titan Xs the CPU cannot keep up, that's totally irrelevant to pretty much everybody and is still not really capturing all of the cases. In most builds the fact is that most games are GPU bottlenecked. That's a bottleneck case right there. Upgrading enough might shift the bottleneck elsewhere (at least it's conceivable) but that's besides the point. Right then and there, it's GPU bottlenecked. It might even be argued that tri sli is a worse experience than 2 way sli due to frame consistency issues.

[Picky Peep]

wat? Might want to elaborate, because most games are primarily GPU bottlenecked.

 

OP is talking about CPU benchmarks, now of course there are going to be small bottlenecks from the GPU but it's not important to see a insignificant value and go through the trouble of getting Titan Xs in SLI and overclock them to prevent the bottleneck.

[Syntaxvgm]

Benchmarking CPU performance in games is itself superficial typically 

 

And I would agree, their cpu and gpu benchmarks, though they have been getting better, have always been weak. 

[TheGeeker]

This can be debated every which way. It depends on how the viewer wants to base performance off of.

[RunDLL]

I don't really agree. If we want to look at how CPU affects performance in CPU bottlenecked games we should benchmark CPUs in CPU bottlenecked games, not make a GPU bottlenecked game artificially CPU bottlenecked game. I don't know what effect that has on the results.

 

It really is no artificially bottleneckig, it´s just removing the GPU bottleneck.  

 

Benching CPU in low res is actually common practice. Have a look at some reviews from http://www.anandtech.com/ for example 

 

 

edit: if you bench in 4k the differences in FPS will be so small thats its hard to tell the degree of performance increase between the generations. If there really is a let´s say 15% perfomnce increase you just can´t see. If the new CPU has 0.3 FPS more than the previous one you just think "meh, dont care" so to speak.

 

Let´s say you are just gaming at 1080p with you x280 or 970 with a crappy CPU an upgrade would work wonders for your FPS that does not translate in such a 4k benchmark setting

[BoredErica]

It really is no artificially bottleneckig, it´s just removing the GPU bottleneck.  

 

Benching CPU in low res is actually common practice. Have a look at some reviews from http://www.anandtech.com/ for example 

I've seen it get used before, I just don't understand the point of that when we can spend the time on cases that don't just resemble the real world cases where it matters, it literally is among the real world cases where it matters.

 

Is it possible (I'm asking because I don't know) that at artificially low resolutions, ram matters more than normal? 

[PineappleHolidays]

I have yet to see the videos, so idk about the benches

[RunDLL]

I've seen it get used before, I just don't understand the point of that when we can spend the time on cases that don't just resemble the real world cases where it matters, it literally is among the real world cases where it matters.

 

Is it possible (I'm asking because I don't know) that at artificially low resolutions, ram matters more than normal? 

 

english is not my first language so i have a hard time expressing and getting my point across.

 

It´s just that 4k isnt a real world case for most of us either. In those high resolutions the 500Mhz more or whatever doesn't make a difference. At least 1080p in addition to the 4k benches would be great.

 

Imagine you run your favorite game.

You have a crappy CPU so you got 12 FPS in every resolution you pick. Doesnt matter if 1280x720 or 4k because your GPU is not the  problem it doesnt bottleneck.

Now if you have a great CPU but a weak GPU you will notice that your games FPS will drop the more you increase the Resolution because this also increases the work the GPU has to do and at some point it just cant render fast enough to keep up with the CPU.

 

Thats the Problem right here. The GPU in this bench is not crappy but still if you put a future GPU from 2 years from now in this PC you will get much higher FPS in those games because i think the CPU could deliver more. To reduce this bottleneck you have to decrease resolution until your FPS wont improve to see the full potential of the CPU. 

[BoredErica]

english is not my first language so i have a hard time expressing and getting my point across.

 

It´s just that 4k isnt a real world case for most of us either. In those high resolutions the 500Mhz more or whatever doesn't make a difference. At least 1080p in addition to the 4k benches would be great.

 

Imagine you run your favorite game.

You have a crappy CPU so you got 12 FPS in every resolution you pick. Doesnt matter if 1280x720 or 4k because your GPU is not the  problem it doesnt bottleneck.

Now if you have a great CPU but a weak GPU you will notice that your games FPS will drop the more you increase the Resolution because at some point it just cant render fast enough.

 

Thats the Problem right here. The GPU in this bench is not crappy but still if you put a future GPU from 2 years from now in this PC you will get much higher FPS in those games because i think the CPU could deliver more. To reduce this bottleneck you have to decrease resolution until your FPS wont improve to see the full potential of the CPU. 

I don't doubt that 4k only is not really useful. I think people are running either 1080p, 1440p, 4k, 1080p 120/144hz, and 1440p 120/144hz. 4k only doesn't tell us enough. There exist games that are CPU bottlenecked, or CPU bottlenecked in some cases, at 1440p. Or even 4k. SC2, Skyrim with a ton of mods during combat, starting up Skyrim (that's right, CPU bottleneck for starting a game). I still get nasty FPS drops in Wolfenstein Enemy Territory with a lot of bots.

 

I believe I understand what you're getting at in the last paragraph there. But I would still lean towards using CPU bottlenecked games that exist today. Both would be best, but of course time is a constraint.

[RunDLL]

I believe I understand what you're getting at in the last paragraph there. But I would still lean towards using CPU bottlenecked games that exist today. Both would be best, but of course time is a constraint.

 

 

Your are right but I think most CPU bottlenecked games are not popular enough for people to care about. In the end the specifics don't even matter. They should just choose games / resolutions that illustrate the CPU impact more

[Keystone Nyan Cat]

I am inclined to agree with you- most of the recent benchmarks barely consist of 3-4 games and a few pieces of software; it would be nice to have a bit more content in each video.

 

Additionally, seeing the benefits in SLI would be nice; as someone with an Ivy Bridge 3770K, I know my CPU is a bottleneck when running in SLI, it would be interesting to see the difference with single and multiple GPUs.