Verifying The Bottlenecker

[Mira Yurizaki]

Prefix: I'm only testing one configuration here. I may go back and test others, but if you want to provide those results, feel free to do so.

 

I've been an opponent against using http://thebottlenecker.com for oversimplifying the whole bottleneck thing. If you aren't aware, it's a site where you enter your hardware configuration and it supposedly tells you how much of a bottleneck (in a percent) your system is having and what's causing it. I didn't like it because I feel it's too little information and may be taken way out of context. Like for example a setup may have a 16% bottleneck on the GPU, but it doesn't really paint the picture entirely. I'm also a bit miffed at the arbitrary decision that "anything above 10% is considered a bottleneck" (by what authority?)

 

However, I decided, why not try verifying its claim on websites that have provided the data and do the number crunching myself? So the problem here is I can't just pick any arbitrary combination of hardware because websites wouldn't have likely benchmarked that combination. Like say pairing a Core i5-2500 with a GTX 1080 Ti. I'm sure someone has, but it there likely isn't enough of them to verify a trend.

 

At first I started with an i5-8400 paired up with a GTX 1080, but The Bottlenecker only claimed 5%. I figure that's too low. So I went with the Core i3-8100 instead. The Bottlenecker claims a 16% bottleneck and so it's a bottleneck by their standards. So looking up what reviews I could find on the part...

 

https://www.tomshardware.com/reviews/intel-core-i3-8100-cpu-review,5385-4.html -  1080p on maximum details

https://www.techspot.com/review/1499-intel-core-i3-8100-i3-8350K/page3.html - 1080p on maximum details

https://www.hardwaresecrets.com/core-i3-8100-cpu-review/6/ - 1080p on medium details

https://hexus.net/tech/reviews/cpu/115481-intel-core-i3-8350k-core-i3-8100-14nm-coffee-lake/?page=7 - 1440p (in most tests) on maximum details

http://minimalisticpc.com/blog/hardware/test-intel-core-i3-8100-i5-8400/ - 1080p on maximum details

 

(These reviews had to compare the i3-8100 to a range of relevant parts and/or with the i5-8400 tested. A lot of other reviews only tested the i3-8100 against a single CPU, usually its direct competitor: Ryzen 3)

 

The verdict is for 1080p gaming, that yes, the i3-8100 does not have the chops to run a GTX 1080 at the same performance level as an i5-8400. But was there a ~10% difference that The Bottlenecker claims there would be?... I'd say based on what I did find, a good portion of the results were within that range. But also, a non-trivial amount of results (especially in DX12 titles) gave a significant edge to the i5-8400. However the only one that bucks this trend entirely were the results from Hexus, because they tested at 1440p which puts more pressure on the GPU to keep up.

 

So what's my take now on The Bottlenecker? I've warmed up to it a little, but just a little. It's a tool you can use to estimate what may happen, but it's not anywhere near an accurate tool. But my stance on bottlenecking remains: it's a symptom of a problem, not a problem itself. The problem itself is simply the PC isn't performing as well as you want it to.

 

If you're planning a new build and you're worried about bottlenecking, this is what I think you should do:

• Set a baseline of what performance you want out of the computer first with tangible figures. Like what frame rate at what resolution at what detail settings. e.g., I want 60 FPS at 1080p with maximum details.
• Look for parts that can achieve the performance requirement you want. Start high at first, then work your way down until you reach a point that balances the performance you want with the price you can afford. Care only about the part's individual performance, don't try to benchmark a system you don't own yet!
• If you want to look for bottlenecks, you can use The Bottlenecker to get a rough estimate on how much performance you can expect to lose, but note this is not a guarantee.
• Check to see if the performance loss is still tolerable or not.
  • If it is, buy the parts and enjoy your new PC.
  • If it isn't, adjust your parts list as necessary. If you're on a strict budget you only have two options: increase your budget to buy more powerful parts or lower your requirements.

EDIT: The advice above is relevant only if you're doing 1080p gaming. If you are gaming at a higher resolution, then it's less likely a bottleneck will occur and The Bottlenecker likely doesn't account for this.

[.Apex.]

you can't calculate bottleneck so vaguely... put the resolution on 720P and the percentage of the bottleneck increases... lower the graphics settings and the bottleneck increases... run it on 1440p or 4K and you basically have no bottleneck.. soooo?

[Mira Yurizaki]

4 minutes ago, syn2112 said:

you can't calculate bottleneck so vaguely... put the resolution on 720P and the percentage of the bottleneck increases... lower the graphics settings and the bottleneck increases... run it on 1440p or 4K and you basically have no bottleneck.. soooo?

The whole point of this exercise was to see how accurate the figures they tell you are. And sure while it was a sample size of one (I don't really feel like looking up other configurations but people are free to add to this post any other configurations they can verify), I like to think that some information, as long as you understand what it means, is better than no information at all. And people are going to use this website whether I like it or not, so I figure why not see how to best use it so people don't misinterpret the information they provide.

[.Apex.]

so to them the higher the core count, the less bottleneck there is? that 16 core 1.8ghz xeon will bottleneck the 1080 Ti like crazy... its just not a well made site and its retarded, no one should recommend this site to anyone.

[WereCat]

4 minutes ago, syn2112 said:

you can't calculate bottleneck so vaguely... put the resolution on 720P and the percentage of the bottleneck increases... lower the graphics settings and the bottleneck increases... run it on 1440p or 4K and you basically have no bottleneck.. soooo?

If your goal is to hit 144FPS in game and your GPU is more than capable achieving it but your CPU tops out at 120FPS then it doesn't matter whether you play at 1440p or 1080p or 720p. You still have the same bottleneck. 

Yes  the % changes as the GPU has to work harder on higher resolutions but you are still getting stuck on the same spot.