Laptops vs Desktop: How do the specs scale?

[Guest]

Hello,

 

I'm not sure if this has been covered or not, and if it has please link me to the forum/video that talks about it, but I have a question concerning hardware specs when it comes to desktops and laptops. As an example, consider my current desktop configuration (in my signature). I'm currently running an i7 4790k, a 4th generation Haswell based processor paired with a 980Ti, which is three generations removed from Nvidia's Ampere architecture. My laptop that I just acquired has an AMD 4000 series chip, the 4800H I believe, which is the latest one you can buy at the moment (I think). It also has a 1660Ti which is Pascal based. Would I expect better performance in games from the laptop versus the desktop because it has newer hardware? Or would the desktop be better because they are full size versions of the CPU and GPU?

 

I guess I'm ultimately asking at what point is it an apples to apples comparison? Can you really just compare them clock for clock, or are there other factors at play?

 

Thanks for any insight, I've always been curious about stuff like this.

[Arika]

 

In terms of your specs, I would say you should expect very vary similar performance between the laptop and desktop. Newer doesn't mean better. The 980ti is still and incredibly capable gpu, As is the devil's canyon CPU.

 

If you got the laptop as an "upgrade" you will be disappointed. Its almost a perfect side grade into a different formfactor for gaming. For content creation or more workstation like workloads, the laptop SHOULD be better if it's not thermally limited.

[Eigenvektor]

You can't really compare clock rates between generations, since there are other factors at play, like IPC increases.

 

I would always expect a laptop to be slower than a desktop with the same hardware. A desktop typically has better power delivery and better cooling. A laptop is far more likely to be thermal throttled and most of the time mobile components are weaker then desktop components of the same generation.

 

Here's a benchmark 980 Ti vs 1660 Ti:

 

Note that both of these are desktop cards, so you probably need to reduce the performance of the mobile 1660 Ti by a few more percent. The desktop variant has a power draw of around 120W while the mobile part is limited to 60W as far as I know.

[Guest]

11 minutes ago, Arika S said:

 

In terms of your specs, I would say you should expect very vary similar performance between the laptop and desktop. Newer doesn't mean better. The 980ti is still and incredibly capable gpu, As is the devil's canyon CPU.

 

If you got the laptop as an "upgrade" you will be disappointed. Its almost a perfect side grade into a different formfactor for gaming. For content creation or more workstation like workloads, the laptop SHOULD be better if it's not thermally limited.

Well the laptop was more as something on the go, not necessarily an upgrade. Just something that could play a few games when I'm away from my desk. I just wasn't sure at what point do "desktop grade" components scale with "laptop grade" ones or if that isn't even a thing? Like does a 4th gen desktop processor equal a 10th gen one assuming they have the same cores and clocks? IDK, something like that. I used my specs as an example but I guess I was thinking more broadly to encompass multiple generations of different products. Thanks for the quick reply though!

 

3 minutes ago, Eigenvektor said:

You can't really compare clock rates between generations, since there are other factors at play, like IPC increases.

 

I would always expect a laptop to be slower than a desktop with the same hardware. A desktop typically has better power delivery and better cooling. A laptop is far more likely to be thermal throttled and most of the time mobile components are weaker then desktop components of the same generation.

 

Here's a benchmark 980 Ti vs 1660 Ti:

 

Note that both of these are desktop cards, so you probably need to reduce the performance of the mobile 1660 Ti by a few more percent. The desktop variant has a power draw of around 120W while the mobile part is limited to 60W as far as I know.

I skimmed the video and that's very interesting findings, they seem to be almost neck and neck but as you say, the percentages should be dropped a bit more for the 1660Ti. Even so, good point about the cooling and power delivery, didn't think about it. And your point is valid about laptop chips generally being weaker than their desktop counterparts of the same generation, but what about earlier ones? I know you mentioned there were other factors at play like IPC increases, but could you say something like 'this generation Intel chip would be equal to this generation Intel chip" even though there's a generation gap?

[Eigenvektor]

12 minutes ago, Stylized_Violence said:

I know you mentioned there were other factors at play like IPC increases, but could you say something like 'this generation Intel chip would be equal to this generation Intel chip" even though there's a generation gap?

You'd have to look at benchmarks for the specific components in question. Performance increases between generations aren't linear, and even within the same product stack. So it's very error prone to judge relative performance simply by looking at a part's generation. Just because you know that e.g. a 10th gen mobile part is 50% faster than a 4th gen desktop part doesn't mean the same is true e.g. when comparing 11th gen to 5th gen.

 

Especially for mobile parts, even the exact same CPU/GPU can have a performance difference between different laptops. Manufacturers can impose additional power restrictions on components e.g. for ultra-mobile parts. And cooling solutions differ, meaning while performance may be equal initially it's quite possible one laptop can do this all day while the other will throttle down after 10 minutes due to heat.

[Guest]

21 minutes ago, Eigenvektor said:

You'd have to look at benchmarks for the specific components in question. Performance increases between generations aren't linear, and even within the same product stack. So it's very error prone to judge relative performance simply by looking at a part's generation. Just because you know that e.g. a 10th gen mobile part is 50% faster than a 4th gen desktop part doesn't mean the same is true e.g. when comparing 11th gen to 5th gen.

 

Especially for mobile parts, even the exact same CPU/GPU can have a performance difference between different laptops. Manufacturers can impose additional power restrictions on components e.g. for ultra-mobile parts. And cooling solutions differ, meaning while performance may be equal initially it's quite possible one laptop can do this all day while the other will throttle down after 10 minutes due to heat.

So is what you're saying is performance would really have to be judged on a case by case basis? Like there is no set standard, even generally, to say one chip will perform better than any other, regardless of generation or whether it's in a laptop or desktop?

[Eigenvektor]

2 minutes ago, Stylized_Violence said:

So is what you're saying is performance would really have to be judged on a case by case basis? Like there is no set standard, even generally, to say one chip will perform better than any other, regardless of generation or whether it's in a laptop or desktop?

Correct. There is no standard that dictates that a new generation has to be x percent faster than a previous generation or that e.g. a 10900K has to be x percent faster than a 10700K. There's also no guarantee that a 10700K is faster than a 9900K (possible, but you don't know until you test).

 

Reviews for the 3080 should be out soon. From what I've seen so far it seems to be quite a bit faster than the 2080 Ti. Compare that to the difference between 2080 Ti and 1080 Ti. If all you knew was the difference between 1080 Ti and 2080 Ti, does that lead to you a reliable prediction for a 2080 Ti to 3080 performance increase?

[Lord Vile]

5 hours ago, Eigenvektor said:

You can't really compare clock rates between generations, since there are other factors at play, like IPC increases.

 

I would always expect a laptop to be slower than a desktop with the same hardware. A desktop typically has better power delivery and better cooling. A laptop is far more likely to be thermal throttled and most of the time mobile components are weaker then desktop components of the same generation.

 

Here's a benchmark 980 Ti vs 1660 Ti:

 

Note that both of these are desktop cards, so you probably need to reduce the performance of the mobile 1660 Ti by a few more percent. The desktop variant has a power draw of around 120W while the mobile part is limited to 60W as far as I know.

Also it's only 60W when plugged in not on battery