what emulators could the pandora run that it doesn't already?


Hardware can have flaws that limits its power, those get ironed out with later revisions of the same SoC, the new Pandora contains a revised SoC of the original version.

Please don't aimlessly speculate on this and call it fact.. just admit you don't actually know of any fixes that will improve performance :p

Foxgod is well known to make overly optimistic assumptions on everything, so I wouldn't even care about what he says. Has he ever coded anything for the system to prove he knows what he is talking about ? :)
 
I am interested in knowing how well mednafen performs with gameboy advance emulation on the 1Ghz units.
Any way to get some measurements out of it, like fps counter?

Shift - F1 (Shift-Fn-1) enables the fps counter. From top to bottom - (virtual, rendered, blitted)
It's skipping frames pretty badly at 1250MHz, alternates between 30-60fps in Advance Wars intro. I guess VBA needs at least 1.5GHz on CortexA8 for stable 60fps. Well at least it can do realtime game speed.
 
Last edited by a moderator:
I am interested in knowing how well mednafen performs with gameboy advance emulation on the 1Ghz units.
Any way to get some measurements out of it, like fps counter?

Shift - F1 (Shift-Fn-1) enables the fps counter. From top to bottom - (virtual, rendered, blitted)
It's skipping frames pretty badly at 1250MHz, alternates between 30-60fps in Advance Wars intro. I guess VBA needs at least 1.5GHz on CortexA8 for stable 60fps. Well at least it can do realtime game speed.

That is interesting, thanks for testing. I tried turning on the fps counter in medafen for an nes game and noticed 60fps was consistent for virtual and rendered, but blitted tended to go back and forth between 30-60fps. CPU usage seems to be ok when watching in top- any idea on what could be causing this?
 
And for N64, we already know the bottleneck is not CPU related, so in terms of emulators, I'm not sure there is much "value" to get a 1Ghz Pandora.

I think the GPU speed was increased to 200 MHZ from 110 MHZ. The RAM is now 200 mhz instead of 166mhz. This is a lot.
 
Last edited by a moderator:
And for N64, we already know the bottleneck is not CPU related, so in terms of emulators, I'm not sure there is much "value" to get a 1Ghz Pandora.

I think the GPU speed was increased to 200 MHZ from 110 MHZ. The RAM is now 200 mhz instead of 166mhz. This is a lot.

Yup, SGX was improved too, not just the cpu.


If it indeed almost doubled it should make a major difference, if there wont be much difference then the gpu isnt the bottleneck either.
 
Last edited by a moderator:
Yup, SGX was improved too, not just the cpu.


If it indeed almost doubled it should make a major difference, if there wont be much difference then the gpu isnt the bottleneck either.

Nope, even if it shows no difference it means nothing about the GPU being or not being a bottleneck. It could mean a bandwidth bottleneck, completely unrelated to the GPU. That's what you frequently see in phones :


http://alienbabeltec...m/main/?p=17125

However, the GPU model and clock speed aren’t the only limiting factors when it comes to GPU performance. The SGX 535 for example can only put out its 28 Mt/s when used in conjunction with a device that supports the full 4.2 GB per second of memory bandwidth it needs to operate at this speed. Assume that the iPhone 3GS uses single-channel LPDDR1 memory operating at 200 MHz on a 32-bit bus (which is fairly likely). This allows for 1.6 GB/s of memory bandwidth, which is approximately 38% of what the SGX 535 needs to operate at its peak speed. Interestingly enough, 38% of 28 Mt/s equals just over 10 Mt/s… supporting Samsung’s claim (with real-world performance at 7 Mt/s being quite reasonable). While it still isn’t proof that the iPhone 3GS uses an SGX 535, it does demonstrate just how limiting single-channel memory (particularly slower memory like LPDDR1) can be and shows that the GPU in the iPhone 3GS is likely a powerful device that cannot be used to its full potential. The GPU in the Droid likely has the same memory bandwidth issues, and the SGX 530 in the OMAP3430 appears to be down-clocked to stay within those limitations.

All you need to do to know a bit more about hardware limitations is just to search and read hardware focused blogs and sites on the net.
 
Last edited by a moderator:
However, the GPU model and clock speed aren’t the only limiting factors when it comes to GPU performance. The SGX 535 for example can only put out its 28 Mt/s when used in conjunction with a device that supports the full 4.2 GB per second of memory bandwidth it needs to operate at this speed. Assume that the iPhone 3GS uses single-channel LPDDR1 memory operating at 200 MHz on a 32-bit bus (which is fairly likely). This allows for 1.6 GB/s of memory bandwidth, which is approximately 38% of what the SGX 535 needs to operate at its peak speed. Interestingly enough, 38% of 28 Mt/s equals just over 10 Mt/s… supporting Samsung’s claim (with real-world performance at 7 Mt/s being quite reasonable). While it still isn’t proof that the iPhone 3GS uses an SGX 535, it does demonstrate just how limiting single-channel memory (particularly slower memory like LPDDR1) can be and shows that the GPU in the iPhone 3GS is likely a powerful device that cannot be used to its full potential. The GPU in the Droid likely has the same memory bandwidth issues, and the SGX 530 in the OMAP3430 appears to be down-clocked to stay within those limitations.

All you need to do to know a bit more about hardware limitations is just to search and read hardware focused blogs and sites on the net.

I mentioned that the gpu wouldnt be the problem, which implies that for example, the memory bandwith could be the problem.....
 
Last edited by a moderator:
Yup, SGX was improved too, not just the cpu.


If it indeed almost doubled it should make a major difference, if there wont be much difference then the gpu isnt the bottleneck either.

Nope, even if it shows no difference it means nothing about the GPU being or not being a bottleneck. It could mean a bandwidth bottleneck, completely unrelated to the GPU. That's what you frequently see in phones :


http://alienbabeltec...m/main/?p=17125

However, the GPU model and clock speed aren’t the only limiting factors when it comes to GPU performance. The SGX 535 for example can only put out its 28 Mt/s when used in conjunction with a device that supports the full 4.2 GB per second of memory bandwidth it needs to operate at this speed. Assume that the iPhone 3GS uses single-channel LPDDR1 memory operating at 200 MHz on a 32-bit bus (which is fairly likely). This allows for 1.6 GB/s of memory bandwidth, which is approximately 38% of what the SGX 535 needs to operate at its peak speed. Interestingly enough, 38% of 28 Mt/s equals just over 10 Mt/s… supporting Samsung’s claim (with real-world performance at 7 Mt/s being quite reasonable). While it still isn’t proof that the iPhone 3GS uses an SGX 535, it does demonstrate just how limiting single-channel memory (particularly slower memory like LPDDR1) can be and shows that the GPU in the iPhone 3GS is likely a powerful device that cannot be used to its full potential. The GPU in the Droid likely has the same memory bandwidth issues, and the SGX 530 in the OMAP3430 appears to be down-clocked to stay within those limitations.

All you need to do to know a bit more about hardware limitations is just to search and read hardware focused blogs and sites on the net.

Hi, just for a layman like me, if I look at the SFC emulation of Yoshi's Island, to me it seems to have performance issues, especially when there are SuperFX enhanced features like 3D doors falling down and what not. When I compare to my SFC Cart performance, it certainly feels different. This is less apparent when I run my 256Mb unit at 750Mhz, but still not 100%.


Can one expect the performance improvements of the 1Ghz unit to help automatically in this game, or will the emulartors need to be optimized to use the faster GPU/CPU? My impression was that it would be the same software and the new OS kernal, just at a faster clock rate.


I also wonder, are there emulators that need more than 256Mb of memory today? Or is this more for "PC Type" applications like a web browser? Could developers add a "512Mb" option for devices that have the extra memory to improve loading times or whatnot?
 
However, the GPU model and clock speed aren’t the only limiting factors when it comes to GPU performance. The SGX 535 for example can only put out its 28 Mt/s when used in conjunction with a device that supports the full 4.2 GB per second of memory bandwidth it needs to operate at this speed. Assume that the iPhone 3GS uses single-channel LPDDR1 memory operating at 200 MHz on a 32-bit bus (which is fairly likely). This allows for 1.6 GB/s of memory bandwidth, which is approximately 38% of what the SGX 535 needs to operate at its peak speed. Interestingly enough, 38% of 28 Mt/s equals just over 10 Mt/s… supporting Samsung’s claim (with real-world performance at 7 Mt/s being quite reasonable). While it still isn’t proof that the iPhone 3GS uses an SGX 535, it does demonstrate just how limiting single-channel memory (particularly slower memory like LPDDR1) can be and shows that the GPU in the iPhone 3GS is likely a powerful device that cannot be used to its full potential. The GPU in the Droid likely has the same memory bandwidth issues, and the SGX 530 in the OMAP3430 appears to be down-clocked to stay within those limitations.

All you need to do to know a bit more about hardware limitations is just to search and read hardware focused blogs and sites on the net.

This article is wrong about all sorts of things, one of those really prominent sources of misinformation. You'd be better off not listening to it :/
 
However, the GPU model and clock speed aren’t the only limiting factors when it comes to GPU performance. The SGX 535 for example can only put out its 28 Mt/s when used in conjunction with a device that supports the full 4.2 GB per second of memory bandwidth it needs to operate at this speed. Assume that the iPhone 3GS uses single-channel LPDDR1 memory operating at 200 MHz on a 32-bit bus (which is fairly likely). This allows for 1.6 GB/s of memory bandwidth, which is approximately 38% of what the SGX 535 needs to operate at its peak speed. Interestingly enough, 38% of 28 Mt/s equals just over 10 Mt/s… supporting Samsung’s claim (with real-world performance at 7 Mt/s being quite reasonable). While it still isn’t proof that the iPhone 3GS uses an SGX 535, it does demonstrate just how limiting single-channel memory (particularly slower memory like LPDDR1) can be and shows that the GPU in the iPhone 3GS is likely a powerful device that cannot be used to its full potential. The GPU in the Droid likely has the same memory bandwidth issues, and the SGX 530 in the OMAP3430 appears to be down-clocked to stay within those limitations.

All you need to do to know a bit more about hardware limitations is just to search and read hardware focused blogs and sites on the net.

This article is wrong about all sorts of things, one of those really prominent sources of misinformation. You'd be better off not listening to it :/
Ill ignore it.
 
Last edited by a moderator:
For my education, care to explain what is wrong about it ? (briefly is OK).

Okay, I'll try to keep it brief: he has a major confusion between Dhrystone and real MIPS, he thinks Intrinsity optimized the instruction timing of their Cortex-A8 when they just made hard macros for it, the 28Mt/s for SGX535 that is the basis for a lot of his conjecture is pretty much completely unsubstantiated and therefore renders the whole thing meaningless, he takes at face value the completely absurd 90Mt/s SGX540 number Samsung threw around publicly for a while (which can basically be called nonsense) and his weird correlations between triangles and memory bandwidth are pulled totally out of thin air.


Honestly triangle rates are one of the most meaningless metrics in the industry, and you should be skeptical of anyone who doesn't at least realize this.
 
Back
Top