The Archos Gamepad.


Boo!


oh well, some games will be ok and I'll have to get on with the analogue nub, shame though, a lot of games will be no good due to this.
 
More a problem for PS1 games than many of the older retro games I'd imagine.


I am looking forward to trying out G-Police on this though, with the 2 nubs and 4 shoulder buttons - That games always been a bit awkward on the Xperia Play and my other handhelds.
 
I still find it impressive that they've managed to sell this device for only 150$, add an keyboard and some analog nubs and you have some kind of oversized Pandora. ^^


We should find out where they build this thing and how cheap are it's components.
 
Last edited by a moderator:
Production in China, mass market, cheap SoC.


That's where the price comes from. Archos produces stuff for years.


However, if it uses one of those cheap SoCs (which I fear), I would not want to use it.


Even though our CortexA8 is old, it has some features that most of the cheap A9 SoCs don't have - which often makes it faster with proper coding.


If you check out the threads on these boards about cheap chinese A9 gaming devices (JXD, etc.) you'll see members that tell you that the emulation doesn't run as good as on the Pandora.


While it's unknown what the Archos Gamepad uses as SoC, it's assumed it's a Rockchip.


The device might be faster that a Pandora, but most probably slower than Samsung or TI CortexA9 devices, whereas the SoC is probably 20 EUR cheaper. Would that be worth it?
 
Production in China, mass market, cheap SoC.


That's where the price comes from. Archos produces stuff for years.


However, if it uses one of those cheap SoCs (which I fear), I would not want to use it.


Even though our CortexA8 is old, it has some features that most of the cheap A9 SoCs don't have - which often makes it faster with proper coding.


If you check out the threads on these boards about cheap chinese A9 gaming devices (JXD, etc.) you'll see members that tell you that the emulation doesn't run as good as on the Pandora.


While it's unknown what the Archos Gamepad uses as SoC, it's assumed it's a Rockchip.


The device might be faster that a Pandora, but most probably slower than Samsung or TI CortexA9 devices, whereas the SoC is probably 20 EUR cheaper. Would that be worth it?
the soc is said to be a rockchip rk3066, which is a proven soc. the benchmarks are close to the exynos 4210 and should beat the omap 4470.


http://mobilityupdate.com/cpu-benchmark/rockchip-rk3066-review/ (vs 4412 vs s4 vs tegra)
 
Anyone know of any reviews for this, I can't seem to find any.. I'm particularly interested in it's N64 emulator performance & some user impressions of the 'D-pad'
 
the soc is said to be a rockchip rk3066, which is a proven soc. the benchmarks are close to the exynos 4210 and should beat the omap 4470.


http://mobilityupdat...-rk3066-review/ (vs 4412 vs s4 vs tegra)

Well, yes. But benchmarks only test pure calculation speeds. That's the deal.


A benchmark doesn't take any additional features and functions into account.


E.g. if an SoC offers two hardware overlays, and another one doesn't, a game making use of that hardware overlays (for HUDs, etc.) might run faster here even though the other one (which doesn't support hardware overlays) probably has a faster CPU and benchmark.


Imagine an SoC wouldn't offer H264 hardware playback as an option: Even a fast SoC would struggle decoding FullHD, whereas a slow SoC with H264 decoding option would do it with ease. (I don't know of any current SoCs without such hardware decoder, but it's an option that's not included in the benchmark).


The OMAP4460 has 1MB L2 cache, for example, whereas the Rockchip has 512kB. This could be a difference with some stuff as well.


Besides, the 4460 is three years old, the current CortexA9 one from TI is the 4470.


Not sure how the GPU compares, but speedwise, the RK3066 runs with 250MHz GPU clock whereas the 4460 runs with 384MHz (different GPU though, so I don't know if it can be compared with that). The OMAP4470 has WAY faster GPU speeds than the OMAP4460 has, BTW.


Don't get me wrong - the Archos GamePad seems to have a good SoC (seems to be the first usable Rockchip ;) )), but a benchmark doesn't really tell much about the normal usage speed. And the chinese SoCs usually don't have much to offer except for raw CPU speed.
 
Anyone know of any reviews for this, I can't seem to find any.. I'm particularly interested in it's N64 emulator performance & some user impressions of the 'D-pad'
http://www.arctablet...1-introduction/

I just found that, but thanks anyway..


Anyone know what the benchmark tests they did: 12000 in Antutu mean for gaming/emulator & multimedia performance, how for instance does it stack up against the Nexus 7?


EDIT:


Just seen EvilD's post.. so it's unclear what the gaming performance is likely to be atm, guess we'll have to wait for some user reviews.


It's simple for me, good N64 emulator performance & half decent controls = sale.
 
Last edited by a moderator:
E.g. if an SoC offers two hardware overlays, and another one doesn't, a game making use of that hardware overlays (for HUDs, etc.) might run faster here even though the other one (which doesn't support hardware overlays) probably has a faster CPU and benchmark.

I think you'll find most SoCs offer something like this.. at least I've seen it almost everywhere except really low end and older SoCs like what you'll find in Dingoo. For instance if you look here: http://www.google.co...YSvAXLrs8zF3dRw Look under "Image post-processor" which lists a standalone scaler and "display interface" which lists 5 layers, two of which provide scaling. Easily comparable to what's on OMAP3530/DM3730.


The problem is that the OS will often fail to expose these overlays in a friendly way.. but for a Pandora-like device running Linux there could be more leverage to support this properly..

The OMAP4460 has 1MB L2 cache, for example, whereas the Rockchip has 512kB. This could be a difference with some stuff as well.

I think we've seen cut down L2 cache because of these vendors lagging process nodes, which will also give them slower and more power hungry processors. But Rockchip alleges to be using GF 28nm soon which could change things for them, while other Asian SoC vendors may not be so aggressive..

Besides, the 4460 is three years old, the current CortexA9 one from TI is the 4470.

OMAP4460 is only 1 year old :/ 4470 is a pretty modest upgrade CPU-wise.. the GPU is notably better though. But still behind what everyone else is offering. It's really sad how far TI is falling behind, and clear why they want to get out of this market along with Freescale and ST..

Not sure how the GPU compares, but speedwise, the RK3066 runs with 250MHz GPU clock whereas the 4460 runs with 384MHz (different GPU though, so I don't know if it can be compared with that). The OMAP4470 has WAY faster GPU speeds than the OMAP4460 has, BTW.

Mali-400MP4 @ 250MHz tends to be substantially faster than either SGX540 or SGX544MP1 at 384MHz.


I don't think your claim that Chinese SoCs have little to offer except CPU performance is accurate (or rather, it is for some but it's not fair to paint all of them under the same stroke), the IP for this chip is pretty competitive. Maybe they can't quite keep up with Samsung or Qualcomm but TI is another story..
 
Last edited by a moderator:
Seems to run emulators like N64 just fine form the videos of seen. Also , as monstercameron mentioned, the RK3066 is a more than capable chip, the reviews I have seen of it , have all given it glowing praise for gaming and in general. Also, from what I have read, the Mali 400 is more than a match for the PowerVRSGX540
 
I remember a while back that someone (dont remember if it was this forum or not) but it was said that the amlogic variety of cortex a9s didnt have neon support because it wasnt listed in their whitepapers. But the functionality is there it just wasnt listed. It performed like a normal a9 would. I think their documentation sucks but I dont think they are cutting corners in the core design.


True you can tell the difference between how high you can overclock, how power efficent they are, and what im guessing is accuracy of the processors. Lot more crashy and glitchy when doing the same thing on 2 comparably speced devices. Higher power drain and heat produced under similar loads.


Im not argueing that the cheaper soc isnt cheaper made, just that I dont think its fair to think of these cheaper socs as being held together with duct tape and missing core features of the arm reference design.


I could be wrong but I would like to see an example of something actually *missing* in one of these socs that comes stanard in other more reputable manufafturers.


One thing we are missing when comparing any of these socs to the pandoras is an apples to apples os to os comparison. But even then you can tell a difference between a better tweaked or optimized port of the os. So saying n64 runs better on this device than on that doesnt mean its 100% (or even mostly) hardware related.


I seen scuddlebutt on what notaz's secret side project could be that could get us closer to a "this is how bad android slows your device down" (which I havent seen 1 valid arguement other than opinion)


But put a decent amount of effort into porting angstrom onto one of these chineese game devices and run the same software and then you can say "see this soc is shit in comparison" or give the pandora a decent native android port and do the same
 
Last edited by a moderator:
The OMAP4460 has 1MB L2 cache, for example, whereas the Rockchip has 512kB. This could be a difference with some stuff as well.
I think we've seen cut down L2 cache because of these vendors lagging process nodes, which will also give them slower and more power hungry processors. But Rockchip alleges to be using GF 28nm soon which could change things for them, while other Asian SoC vendors may not be so aggressive..

Yes, as mentioned, it seems like this is the first real good Rockchip :)

Besides, the 4460 is three years old, the current CortexA9 one from TI is the 4470.
OMAP4460 is only 1 year old :/ 4470 is a pretty modest upgrade CPU-wise.. the GPU is notably better though. But still behind what everyone else is offering. It's really sad how far TI is falling behind, and clear why they want to get out of this market along with Freescale and ST..

Wow - they really took their time :eek:


I just found the news from TI where they demonstrated the first OMAP4 in February 2010... and it seems I fell for a TI bug - their archive told me a press release about the OMAP4 from February 2009, but then again, most news were from the same date, so something seems to be wrong here.


Didn't know TI was THAT far behind... I thought they were pretty advanced with the OMAP4 and only fell back recently with the OMAP5.


Sad to see, their chips were good. :(

I don't think your claim that Chinese SoCs have little to offer except CPU performance is accurate (or rather, it is for some but it's not fair to paint all of them under the same stroke), the IP for this chip is pretty competitive. Maybe they can't quite keep up with Samsung or Qualcomm but TI is another story..

Seems things have changed lately. I remember a year ago that Rockchip was terrible and Allwinner were okay, but still a bit slow, but it seems they catched up.


Thanks for the clarification :)
 
Anyone know of any reviews for this, I can't seem to find any.. I'm particularly interested in it's N64 emulator performance & some user impressions of the 'D-pad'
http://www.arctablet...1-introduction/

I just found that, but thanks anyway..


Anyone know what the benchmark tests they did: 12000 in Antutu mean for gaming/emulator & multimedia performance, how for instance does it stack up against the Nexus 7?


EDIT:


Just seen EvilD's post.. so it's unclear what the gaming performance is likely to be atm, guess we'll have to wait for some user reviews.


It's simple for me, good N64 emulator performance & half decent controls = sale.
check these videos out http://blog.thinkteletronics.com/?p=454
 
The OMAP4460 has 1MB L2 cache, for example, whereas the Rockchip has 512kB. This could be a difference with some stuff as well.
I think we've seen cut down L2 cache because of these vendors lagging process nodes, which will also give them slower and more power hungry processors. But Rockchip alleges to be using GF 28nm soon which could change things for them, while other Asian SoC vendors may not be so aggressive..

Yes, as mentioned, it seems like this is the first real good Rockchip :)

Besides, the 4460 is three years old, the current CortexA9 one from TI is the 4470.
OMAP4460 is only 1 year old :/ 4470 is a pretty modest upgrade CPU-wise.. the GPU is notably better though. But still behind what everyone else is offering. It's really sad how far TI is falling behind, and clear why they want to get out of this market along with Freescale and ST..

Wow - they really took their time :eek:


I just found the news from TI where they demonstrated the first OMAP4 in February 2010... and it seems I fell for a TI bug - their archive told me a press release about the OMAP4 from February 2009, but then again, most news were from the same date, so something seems to be wrong here.


Didn't know TI was THAT far behind... I thought they were pretty advanced with the OMAP4 and only fell back recently with the OMAP5.


Sad to see, their chips were good. :(

I don't think your claim that Chinese SoCs have little to offer except CPU performance is accurate (or rather, it is for some but it's not fair to paint all of them under the same stroke), the IP for this chip is pretty competitive. Maybe they can't quite keep up with Samsung or Qualcomm but TI is another story..

Seems things have changed lately. I remember a year ago that Rockchip was terrible and Allwinner were okay, but still a bit slow, but it seems they catched up.


Thanks for the clarification :)
in the past the rockchips didnt seem that good and allwinenr was better, but that was with rk2xxx series, single a8 and viviante gc800 gpu...was bad then.
 
Back
Top