Discussion in 'Everything else' started by 3XCL4M4t10N, Dec 27, 2015.
There's probably not an awful lot that GPD themselves can actually optimize anyway.
It runs Windows 10, leaving at least one thing to improve upon.
So they added an optional fan, and then extra battery to support the extra power the fan uses.
Leaving you with one less port. I suppose its good to have options, but it means a set of them that does away with some convenience.
A further divide between power-games with mouse and keyboard, and handheldgames with just the handheld and fan.
If there wasn't a lot of handheld-games using the power before, there aren't any even more power-hungry handheld-games now. So it serves a purpose mostly in future, docked or USB-hub-like conditions.
Hmm... fan... upgraded battery... sounds to me like the Intel chip is a bit more powerhungry than they expected...
Hopefully with a performance that matches the hunger for power
Next update they'll announce that they decided to switch to a 13" screen, and a laptop case.
The fan is optionnal and as far as I know we will have to use it only for power hungry games. If GPD can make a custom one to clip on the GPD Win as a shell, it's a lot better than a simple fan that requires space and not to move. I'm quite satisfied that they balance the extra battery for the fan with more battery.
To cater for the bigger 6900mAh battery, the width of the GPD WIN has increased to 23.5 mm (a 1.5mm increase) . The bigger battery is to cater for the change to active cooling -> fan.
More on the fan to be used. this is from an interview with GPD
"In addition to updating the processor, the official also revealed that GPD WIN will be changed to active cooling, wear-resistant, oil-bearing fan a longer life than ball fan. As a single ball or two balls, the official stressed that: Although the double ball better, but the noise, we would consider striking a balance between reducing heat and noise reduction. It is understood that a single ball bearing fan The average life 40,000 hours, double ball bearing fan life in 60,000 to 100,000 hours. For small or fan noise can be tolerated, besides so small fan should be almost no noise, the official release on double ball bearing fan. Of course, not, with small series have been DIY removable disk cartridge strength, another double ball bearing fan is really not so difficult"
It's still not totally clear whether the fan is internal, or an external add on. That still needs to be confirmed. It does sound though that there is a chance it's internal now. But it's not clear. I will find out soon I'd imagine.
15% larger battery sounds pretty neat indeed. Lot of my uses for it won't stress the processor too much. I'll probably opt out of getting an external fan. The win sounds like it's turning into a really great device.
As they are already going to increase the thickness , I really rather they add an internal one. Hopefully the fan will work only under certain conditions, so the extra battery is really welcome
Wow, that really sounds like that Intel chip can heat up quite a bit... :O
I hope we won't have those infos with the OMAP5, but so far, things are looking good.
Well, I guess the fact that Windows 10 and games are made for desktop and therefore often are not very optimized to run at low speeds probably doesn't help...
All impressions I've gotten from Cherry Trails so far is that the perf/W on the CPU side is comparatively good, while the perf/W on the GPU side isn't that good. The GPU heavy tests on sites really make the thing suck power down, and that's why in tablet implementations the GPU can't really sustain anything close to its peak clocks.
This makes sense if you look at Intel's mobile history and original roadmaps. In any SoC they've ever made that was explicitly intended for phones and small tablets they've used a licensed IMG GPU, despite having their own GPU IP. Surely they've done this for reasons of power efficiency. But on the CPU side things should likely be similar to the phone designs.
Since this is a gaming device it makes sense that GPD will want to deal with this as much as possible. But if you don't need to use the GPU too heavily, like with 2D games (let's hope they're not GPU intensive anyway), certain emulators, browsing and productivity work and the like the device might have better efficiency. The situation probably won't be much better with Pyra since its peak 3D performance is much lower anyway.
Or this could be all wrong, hard to really say at this point. I'm hoping to see some power consumption tests under many scenarios.
Ah, thanks for the info!
That also explains while Intel tablets run hot shortly after running 3D games.
Well. Upgrade to 8700/8750 brings about 15 percent more speed.
I always thought this will cost battery life.
Now the battery upgrade of 15 percent.
Sounds to be a good deal.
And that's why I never buy the latest and fastest hardware and rather look at noise, power efficiency, stability, etc...
Firstly the system should work without much trouble and last a while, performance comes after that.
This said, I'd have no problem with underclocking the Pyra as long as it's still a bit faster than the Ghz Pandora and has all the nice things like keyboard backlight, more buttons, improved keyboard, GPS, etc...
I wonder whether the decision to upgrade the CPU to the same as stated for the PortableGamingScam has anything to do with mentioned project. It would be sad if a real manufacturer taking it seriously would actually worsen its product because of it.
Likely more than that for gaming. In addition to the small CPU speed increase, you also get 33% more EUs in the GPU. That's going to have much bigger impact on game performance than an extra ~150MHz. But again, more EUs means more power usage and more heat.
I still maintain that they could have avoided all this fan nonsense by putting the SoC behind the display and sinking it directly to an (aluminum) lid. It works a treat for some of the cherry trail tablets that have a similar configuration.
I'm not sure about the fanless thing... take a look at this video from PGS:
(Yes, I know it's PGS... forget about their crap with the heatsink or without, that's just fake).
This is most probably the Voyo 3, and running the 3D benchmark, you can see how fast the GPU reaches 78°C (after 2 minutes) and heavily throttles the speed.
As was mentioned before in this thread: Intel GPUs aren't that powersaving, only the CPUs are... so that would explain that.
I strongly suspect Intel GPUs look quite good in terms of perf/W against nvidia and AMD discrete cards and chips, just not at all good against anything properly low power. They were probably looking at the former when designing the things.
I don't know, nVidia's Maxwell GPU scaled down for the Tegra X1 looks like it has pretty good perf/W. Pascal in the upcoming 16nm Tegra should be better still. The thing is, power efficiency is really crucial for all GPUs these days, since even at the largest scales they're pretty power limited. nVidia made tons of changes with Maxwell that improved efficiency a lot despite staying on the same process node as its predecessor Kepler, and that is largely what enabled its performance improvements. Not sure about AMD, but they're at least working hard to stay competitive.
I think Intel has just doesn't have the same talent when it comes to GPU design. They've been doing it a long time, but for many generations it was just there to give the bare minimum for users who don't really care about 3D performance, where they got a good enough GPU bundled with the motherboard chipset or CPU. When Intel started becoming more aggressive with their IGP offerings they stayed competitive with AMD at the highest end largely because they were able to employ a specialized memory solution (large eDRAM cache) while AMD was hampered by the main memory controller. But efficiency-wise they still lagged behind the state of the art discrete solutions, despite not yet having their own exotic memory systems.
i bit the bullet and ordered one @ 330 USD. it's a punt at this point and will be a collectable item for me.
In the mobile world, tegra is a joke for power consumption.
Separate names with a comma.