News Razer Blade 13 Will Have Switchable TDP, Intel Says

Giroro

Splendid
I'm sure the processor will down-clock when idle, so lowering the clock/TDP will just make the processor work for a longer period of time before going idle. I think it will have a small effect on battery life, at best. There's even a chance it could lower battery life if doing the work slower keeps other parts of the system active for longer as well. Usually a switchable TDP isn't about efficiency, it is is so a designer can make the a single chip work in multiple machines/form factors without having to keep multiple parts in inventory, right?
 

Joe Braga

Reputable
Sep 4, 2019
2
0
4,510
AMD needs to get 7nm for notebooks.

They are like backward from intel.

Intel unveils new nodes to mobile first. AMD unveils new nodes to desktop first.
How so? Intel presented new mobile cpu SKUs but you need the Laptops' AIB s is still used to fab more laptops intel CPU based but 7nm mobile cpu will come in next year
 

DavidC1

Distinguished
May 18, 2006
494
67
18,860
I'm sure the processor will down-clock when idle, so lowering the clock/TDP will just make the processor work for a longer period of time before going idle.

Laptops already switch TDP depending on what you do. Convertibles switch between cTDPdown and Nominal when you switch from Tablet/Laptop mode. Nothing is special about this Razer.

Maybe its user configurable. You can use lower TDP mode in browsing/video watching to keep fan speeds down, and higher TDP for gaming and other demanding loads.

I think it will have a small effect on battery life, at best. There's even a chance it could lower battery life if doing the work slower keeps other parts of the system active for longer as well.

It depends heavily on load and the system. Higher TDP will have worse battery life on higher loads, and in workloads that needs sustained performance. In that case, faster won't help at all, as you'll use the performance to do more work anyway.

Higher TDP spec chips tend to have overall more aggressive clock ramps so that'll contribute to using more power.
 

Giroro

Splendid
Laptops already switch TDP depending on what you do. Convertibles switch between cTDPdown and Nominal when you switch from Tablet/Laptop mode. Nothing is special about this Razer.

I'm not arguing why a PC that changes configuration might need to adjust it's thermal load, for example removing your machine from a dock might remove part of the cooling solution, and the device might be otherwise too hot to hold in your hands if the CPU reaches the point of thermal throttling.

But TDP is a measurement of heat dissipation not efficiency of the CPU itself or even performance/clock rate (although they are usually closely correlated). It doesn't change the amount of work that needs to be done. If a video runs fine in 15W mode because it has low CPU requirements, the computer is going to downclock to the same level in 25W mode so there isn't going to be any difference in efficiency or performance.

The more I think about it, the best uses I can think of are "You are setting your computer up to run a workload before bed, and but the fans are annoying. So you don't care if it takes 5 hours instead of 3 to complete, you just want it quiet".
Or maybe the amount of power used by the fans is enough that, even though the computer is active and working for longer, you come out ahead by spinning the fans less.
 

GetSmart

Commendable
Jun 17, 2019
173
44
1,610
Found a video from IFA 2019 featuring both Razer Blade Stealth Mercury White Edition and that reference laptop with AMD Ryzen 7 3700U running Halo Reach.
 

DavidC1

Distinguished
May 18, 2006
494
67
18,860
It doesn't change the amount of work that needs to be done. If a video runs fine in 15W mode because it has low CPU requirements, the computer is going to downclock to the same level in 25W mode so there isn't going to be any difference in efficiency or performance.

Actually, if your load is demanding enough, yes it will directly impact your battery life.

And usually in those workloads, you can't just translate faster = finishing faster. If you are gaming for example, you won't stop gaming 30% earlier because its that much faster. You'll just play at better performance for the same amount of time.

And there are plenty of workloads that are in between. The only time where TDP doesn't factor in battery life is where its bursty enough that it can idle most of the time. And even that demand is increasing with people running more windows and tabs at the same time and with ever increasing demands brought on by the browsers and the websites themselves.
 

Giroro

Splendid
And usually in those workloads, you can't just translate faster = finishing faster. If you are gaming for example, you won't stop gaming 30% earlier because its that much faster. You'll just play at better performance for the same amount of time.

And there are plenty of workloads that are in between. The only time where TDP doesn't factor in battery life is where its bursty enough that it can idle most of the time. And even that demand is increasing with people running more windows and tabs at the same time and with ever increasing demands brought on by the browsers and the websites themselves.
I would argue that gaming is usually a bursty workload since it's lightly threaded. How often do modern games realistically get 100% CPU utilization on a laptop like this? A lot of games might fully use 1-4 threads that get bounced around the CPU, and if lowering the TDP lowers the max clockrate that could trade off gaming performance for power consumption which would increase battery life... but the CPU isn't running at it's full power consumption in either case (although it's intel, so it's possibly drawing a lot more power than it's TDP in either case - may not hit it's full boost under a sustained load like that).
If you aren't going out of your way to induce a CPU bottleneck, then usually the CPU gets done with it's work per frame before the GPU, so it's utilization has many small spikes... But yeah, I get your general point. But I don't know if that will really be a typical workload. I guess we'll just have to see how it works out in practice.
And to be clear - TDP is not a measure of how much power the CPU uses, it's a measure of how much heat your cooling system need to dissipate in order to to meet Intel's base clock on all cores. That's why it's a weird idea to make it a user-configurable setting. Especially since Windows already has power settings to passively cool and/or throttle the CPU. Turning off the fans should help efficiency at a fixed amount of work.
I'm working from assumptions because I don't even know what changing that setting does. Maybe it lowers the base clock and leaves boost the same - Maybe it just shuts off half the cores or hyperthreading - maybe it's some combination.
 

DavidC1

Distinguished
May 18, 2006
494
67
18,860
And to be clear - TDP is not a measure of how much power the CPU uses, it's a measure of how much heat your cooling system need to dissipate in order to to meet Intel's base clock on all cores. That's why it's a weird idea to make it a user-configurable setting. Especially since Windows already has power settings to passively cool and/or throttle the CPU. Turning off the fans should help efficiency at a fixed amount of work.

I'm sorry, but I think your comments need a lot of clarification, with both your points about Intel exceeding TDP specs or that TDP doesn't equal power consumption.

Point 1: In Desktops, you see a 95W specced CPU consistently using 160W. But Desktops use ridiculously sized heatsinks anyways with far more open room and additional internal case fans to help.

In Laptops though, if you really lie on TDP, you'll pay for it in performance or with the laptop running really hot. Because the manufacturer will rely on TDP specifications.

So ultimately I don't believe its them exceeding TDP as much as Desktops taking advantage of brick-sized HSFs with infinite PL2 time values. AMD does the same thing with Ryzen anyways, only to compete with Intel in performance.

Point 2: This is a natural extension to Point 1. Intel Datasheets show PL2(Turbo value) has to come down to PL1(TDP value), and over a longer time the value has to equal PL1. Because if you make the long average exceed PL1, then its the same thing as going over TDP anyways. If you are doing anything where you are running at max CPU demand for 10+ minutes, then the CPU can't exceed its set TDP values.

Also because of that point in that Datasheet, in any workloads that makes the CPU hit the TDP barrier, TDP equals the power consumption. Of course if you are web browsing or watching a video it'll be idling 95% of the time and in that case TDP doesn't equal power consumption, but why talk about TDP at all then?

I would argue that gaming is usually a bursty workload since it's lightly threaded. How often do modern games realistically get 100% CPU utilization on a laptop like this

Also, in the Razer specifically, they said the gaming oriented version with the GTX 1650 uses the 15W TDP Icelake, while the Mercury White with the Intel Iris Plus graphics uses the 25W TDP Icelake. So it seems to be since its such a small chassis, they are considering system TDP, not just the TDP of the individual components.

And for the Mercury White version, 25W TDP on the Iris Plus means either the CPU or the GPU will try to be the majority consumer of the TDP. Therefore in that case, setting it to 15W will use less power in games and thus more battery life.*

Gaming isn't considered bursty since its a very performance sensitive application and there aren't really true idle periods unlike browsing. And unless you are capping your frame rates, then the CPU and the GPU will try to run fast as possible.

*(Not that it matters anyways, since all laptops get 1.5-2 hours when running games)
 
Last edited:

Giroro

Splendid
Point 1: In Desktops, you see a 95W specced CPU consistently using 160W. But Desktops use ridiculously sized heatsinks anyways with far more open room and additional internal case fans to help.

Right. Because, like I keep saying, TDP is a thermal design recommendation, not a real spec for power draw. Which is why it's at best a meaningless setting to let the end-user change in a laptop with a fixed thermal design. It's fun to think about but I think we both know the real reason the feature is included is because it ended up being more work to remove it than to keep it in.
It's well known that overclocked processors are less efficient than stock. Does that trend meaningfully apply to the low in when the processor is running within the range it was designed for at power levels that are going to realistically end up being very similar to each other, even if power draw is hard fixed to the TDP? I don't know for sure. My best guess is that there is going to be very little difference. It's still a laptop, so I expect you're going to save more than that 6W difference just by dimming the display.
Hopefully we get to see a battery rundown and efficiency metric for both settings. Alternatively somebody can start swapping same-generation processors into a fixed laptop configuration and see what happens.