N64 Anyone?

[ldesnogu]

project64 (v1.6) runs at over 30fps on my XPlited eeePc 701 (upgraded to 2gb). I'm assuming that the pandora is as powerful as that and won't be having the extra weight of having XP running underneath it.

EeePC 701 is using a 900 MHz Celeron IIRC. That would make it at least twice faster than Cortex-A8 @600 MHz (and even more if project64 uses FP).

 

[midna25]

project64 (v1.6) runs at over 30fps on my XPlited eeePc 701 (upgraded to 2gb). I'm assuming that the pandora is as powerful as that and won't be having the extra weight of having XP running underneath it.

EeePC 701 is using a 900 MHz Celeron IIRC. That would make it at least twice faster than Cortex-A8 @600 MHz (and even more if project64 uses FP).

The Pandora can be overclocked to 900mhz. 600mhz is just the default speed.

 

[gruso]

Comparisons between architectures: useless.

 

[MDave]

http://www.gp32x.de/board/index.php?showtopic=45583&hl=


Claiming 4x is much more generous than 3.4x which is what's there.

QUOTE
QUOTE
To cut a long story short, I managed to port the emulator over fairly quickly, but after starting a new job development ground to a halt. A few months later I saw the buzz surrounding PSMonkey's N64 emu, and thought I'd have to finally bite the bullet and let people see what I had been working on.


http://www.emutalk.net/showthread.php?t=34897

Sounds like he got one up and running with not much effort to me!

That was his first version, which ran like crap. You must have missed the several months of blog updates he made detailing the work he put into the later versions. Even then, he still put much more work into his initial port than anyone non-Sony put into a PS1 emulator for PSP.


I did say roughly 4 times. Don't go taking things out of context!

And all I mentioned was he got it up and running quickly, nothing more. And I'll add anyway that it was still further along then ps1 or n64 emulation at the time.

 

[mali]

Celeron @900MHz:

CODE
BYTEmark* Native Mode Benchmark ver. 2 (10/95)
Index-split by Andrew D. Balsa (11/97)
Linux/Unix* port by Uwe F. Mayer (12/96,11/97)

TEST : Iterations/sec. : Old Index : New Index
: : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT : 260.26 : 6.67 : 2.19
STRING SORT : 27.4 : 12.24 : 1.89
BITFIELD : 9.8988e+07 : 16.98 : 3.55
FP EMULATION : 31.858 : 15.29 : 3.53
FOURIER : 6598.9 : 7.50 : 4.22
ASSIGNMENT : 7.1967 : 27.38 : 7.10
IDEA : 1449 : 22.16 : 6.58
HUFFMAN : 486.19 : 13.48 : 4.31
NEURAL NET : 8.706 : 13.99 : 5.88
LU DECOMPOSITION : 318.11 : 16.48 : 11.90
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX : 15.034
FLOATING-POINT INDEX: 12.004
Baseline (MSDOS*) : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
CPU : GenuineIntel Intel(R) Celeron(R) M processor 900MHz 900MHz
L2 Cache : 512 KB
OS : Linux 2.6.26-3.slh.4-sidux-686
C compiler : gcc version 4.3.2 (Debian 4.3.2-1)
libc : libc-2.7.so
MEMORY INDEX : 3.628
INTEGER INDEX : 3.847
FLOATING-POINT INDEX: 6.658
Baseline (LINUX) : AMD K6/233*, 512 KB L2-cache, gcc 2.7.2.3, libc-5.4.38
* Trademarks are property of their respective holder.

Pandora @900MHz:

CODE
TEST : Iterations/sec. : Old Index : New Index
: : Pentium 90* : AMD K6/233*
--------------------:------------------:-------------:------------
NUMERIC SORT : 363.48 : 9.32 : 3.06
STRING SORT : 49.96 : 22.32 : 3.46
BITFIELD : 1.3284e+08 : 22.79 : 4.76
FP EMULATION : 91.527 : 43.92 : 10.13
FOURIER : 395.63 : 0.45 : 0.25
ASSIGNMENT : 5.8503 : 22.26 : 5.77
IDEA : 982.24 : 15.02 : 4.46
HUFFMAN : 604.53 : 16.76 : 5.35
NEURAL NET : 1.1844 : 1.90 : 0.80
LU DECOMPOSITION : 66.92 : 3.47 : 2.50
==========================ORIGINAL BYTEMARK RESULTS==========================
INTEGER INDEX : 19.739
FLOATING-POINT INDEX: 1.437
Baseline (MSDOS*) : Pentium* 90, 256 KB L2-cache, Watcom* compiler 10.0
==============================LINUX DATA BELOW===============================
MEMORY INDEX : 4.562
INTEGER INDEX : 5.217
FLOATING-POINT INDEX: 0.797


Everything is fine, except FP.

 

[Hotcooler]

Actually INT is quite impressive! But we sure have DSP and NEON wich theoretically can help with FP. According to this DSP is quite capable and this. But I feel I'm missing something

 

[scottm]

actually I think my eeepc 701 default speed is less than 900mhz (although it can be overclocked to 900). I think the default speed is nearer 750?

Anyway it runs fine at the default clock speed

 

[MDave]

Actually INT is quite impressive! But we sure have DSP and NEON wich theoretically can help with FP.

Yes, I don't think the main cpu was designed to handle FP on its own.

I know FP helps for 3D calculations, but that is what the SGX is for right? What FP calculations would need to be done on the main cpu if the SGX is used, for your typical 3D game?

 

[Hotcooler]

According to some infos on the net. DSP alone can produce 1800Mflops at 430mhz. And Celeron 900 can do 2100 Mflops.. Yeah I know... different purposes and different architectures... but...

 

[mali]

For the interested: TMS320C64x/C64x+ DSP CPU and Instruction Set Reference Guide - no floats there.

 

[ldesnogu]

@Mali: where did you get the Celeron M and C-A8 nbench results from? One of them looks very wrong, or nbench is even worse than I thought

 

[mali]

http://sidux.com/PNphpBB2-viewtopic-t-12398-newlang-jpn.html

http://www.gp32x.de/board/index.php?s=&am...st&p=676142

 

[Hotcooler]

Oh... I guess.. there is some vierd stuff with numbers... C67 I guess in FloatP DSP.. ours is FixedP with Vector Floating Point stuff and some libs to do float on fixed

BTW using NEON should boost FP performance... but Dunno if it useful in recompiling, emulation e.t.c.

I guess we need some devs to answer this for sure..


Anyway... I want a supercomputer in my pocket!

 

[Chip]

Not to dash everybody's hopes, but there is no "supprise" N64 development in progress that we are aware of. If an accomplished emu author came forward and proposed such a project, he/she would go to the top of the dev board list, but none have. It is still a possibility that someone could take up the project, but you really shouldn't expect it soon.

 

[mali]

BTW using NEON should boost FP performance... but Dunno if it useful in recompiling, emulation e.t.c.

It should help to use NEON, but NEON itself has got problems, too. We will have to live with Cortex' weak FP performance, but after seing the Quake2 benchmark(which isn't super high, but higher than I expected) we shouldn't notice that weakness too much. It's a handheld after all and it's damn fast

 

[ldesnogu]

http://sidux.com/PNphpBB2-viewtopic-t-12398-newlang-jpn.html


If you look at the post below the one giving the Celeron M 900 MHz you'll see results for a Celeron M 1500 Mhz that are more than twice better.
CODE
CPU : GenuineIntel Intel(R) Celeron(R) M processor 900MHz 900MHz
MEMORY INDEX : 3.628
INTEGER INDEX : 3.847
FLOATING-POINT INDEX: 6.658

CODE
CPU : GenuineIntel Intel(R) Celeron(R) M processor 1500MHz 1496MHz
MEMORY INDEX : 7.841
INTEGER INDEX : 8.708
FLOATING-POINT INDEX: 15.421

Sorry but that makes no sense, something's very wrong here

 

[Hotcooler]

I'm pretty happy But there is never too much performance.

 

[scottm]

Okay then, how do we start a fund to get a seasoned emu developer on board? I'm sure a lot of people would happily contribute

Someone mentioned a N64 emu comp - this sounds like a great idea. I recently entered 2 games into the retroremakes gaming comp (total of about 20,000 lines of code for the 2 games) this lasted 3 months. Is the same timeframe feasible for a n64 comp?

 

[God Ginrai]

Okay then, how do we start a fund to get a seasoned emu developer on board? I'm sure a lot of people would happily contribute

Someone mentioned a N64 emu comp - this sounds like a great idea. I recently entered 2 games into the retroremakes gaming comp (total of about 20,000 lines of code for the 2 games) this lasted 3 months. Is the same timeframe feasible for a n64 comp?

Probably a time frame of 5-6 months would be better.

-God Ginrai

 

[Hotcooler]

To compare Nbench results with something - http://www.tux.org/~mayer/linux/results2.html