Search results

Until Tuesday i was in exams, so there wasn't much programming going on.... though I have added ucode profiling to gles2n64, here's what it spits out for super mario 64: http://pastebin.com/m113e9f90 In the last few weeks i have: 1. Implemented but not tested some optimisations to the *_TR1...

Classic example of why I've given up on that thread. I think it went public too soon.

I recently got fzerox working after adding some 64bit opcodes to the dynarec, it's about 40-50fps @ 850mhz... when it actually works. I have a big bug which i cannot put my finger on, most the time when you start fzero & 1080 you get a large black square with a few adjacent squares in the bottom...

Diddy Kong Racing is basically full speed @ 850mhz.... But your right, the others will likely be very hard to emulate fast (especially Banjo Tooie by all reports). As i said above, Conkers Bad Fur Day will not work with the current Graphics plugin (it uses custom ucodes), i wont consider adding...

glN64 has support for the Perfect Dark and GoldenEye ucodes. It doesn't support Conkers Bad Fur Day though. Starfox 64 is one of the better running games. Currently the training mission is about 12-15 fps@500mhz and 20-25 fps@850mhz with no frameskip. Games without many textures seem to have the...

Hmmm, as accurate as the PC version of glN64 anyway. It seems pretty common to use fixed function texture environments to emulate the blends modes, rice_video and glN64 use them anyway (glide64 uses the glide ~equivalent). It's so they can support OGL 1.5 cards.

MWeston's video shows it at 600mhz then at 800mhz. Fixed. Yeah shaders aren't a bottleneck yet, but mario 64 is basically the best case scenario. Games like banjo kazooie use some relatively complex shaders with multiple texture sources, simple shadows and lighting being blended together. If...

It's already purely GLES 2.0 via my wrapper. Perfect Dark & Goldeneye appear to render fine (emulating GLES2 on PC) except for the skybox not updating itself (common problem, there might be some hacks), however Ari64's dynarec doesn't like them at the moment.

Most of the stuff here will probably be available: http://www.angstrom-distribution.org/repo/?section=libs

Its rendered at 800x480, no upscaling is occuring. I don't think its the graphics plugin, i've had it run alot of games well on pc (i posted them a little while back). Heres the games i've test that appear to work (through there intros anyway): Super Mario 64, Mario Kart, 1080 Snowboarding...

In the newest build the framerate scales almost linearly with CPU speed. This is a very good thing, it means the SGX and RAM are not the bottleneck... so there's more scope for ARM/NEON optimization. From memory most the sm64 ingame stuff is about 15-17 fps @500mhz, if you enable frameskip so...

Funny how when people read "phone" they presume "iphone". Its a Nokia N95 and yes it does rock. I haven't fiddled with the sound, i haven't even tested if my board can output sound. I would presume the sound is being computed as per default. BTW: 30 fps is the maximum speed of Super mario...

Yeah I'm currently lacking all the buttons / inputs. So i was hoping something over serial was possible. And here's a shitty video taken with my phone: http://www.youtube.com/watch?v=O_EXdG5x7j0#t=0m25s Youtube Quote: I've also got 1080 Snowboarding to work occasionally and Mario Kart, both...

I've compiled the latest SVN of GemRB (with minor changes). It loads all the plugins fine but fails on the SDL_SetVideoMode() call. I've had this problem with every SDL app i've compiled, i think its due to my Frankenstein OS setup to get the SGX working on a rev4 board. At the moment I'm...

I've got mupen working on my board.... it was just as simple as removing the set video call and fixing some linking errors. I've also began debugging some of the NEON optimisations. At the moment i have neon matrix multiplication and matrix vector multiplication working. I'm running at the...

I'm pretty sure vrecpe is just: 1 / (1+m)*(2^n) = (1 / (1+m)) * (2^-n) 1 / (1+m) ~ 1.41176471 - 0.47058824 * (1+m) So you negate the exponent and do a linear expansion for the mantissa. The above values give a maximum error of ~0.0588 in the mantissa (and no error in the exponent). The step...

Sure any patchs would be most welcome. Just PM it to me or my gmail account. Also, If anyone wants SVN commit rights just contact me. Did it fail on both __NEON_OPT's? Not too surprised about the NEON not working straight away, i really need to get my board working properly so i can test this...

Yeah optimally you would be doing 4 of more at a time. This was the best i could think of for the single dotp because the vmul + vmla is a special case in the pipelining so they can be issued back to back. It can do interleaving / deinterleaving in the LS unit, but nothing as advanced as what...

I have updated the gles2n64 SVN with NEON optimisations. Just compile with -D__NEON_OPT for optimisations that i've mainly just lifted from math-neon, they should be pretty stable. Compile with -D__NEON_OPT2 for some additional optimisations i've not tested. Do you mean there's bigger...

As far as I'm aware we cannot do this. You can still do the 4 component dot product in 3 instruction though. This is pretty much identical to what you wrote: float dot4_neon(float v0[4], float v1[4]) { float r; asm volatile ( "vld1.32 {d2, d3}, [%1] \n\t" //d2={x0,y0}, d3={z0...