Author Topic: [Ndless] Help with bottleneck on color calcs only (Read 19228 times)

DJ Omnimaga · « **Reply #30 on:** June 27, 2014, 06:17:10 pm »

Yeah but wouldn't it still be faster to paste the data on the grayscale screen since it only supports 4 bits, such as converting the data beforehand or stripping the extra data? Not sure how the grayscale Nspire screens work, though, so maybe I'm just misunderstanding something.

Vogtinator · « **Reply #31 on:** June 27, 2014, 06:17:36 pm »

I have a theory:
The frame buffer contents mustn't be cached, as that may cause artifacts. So caching is disabled by the OS.
On GS calcs the screen buffer is in SRAM, so SRAM caching is disabled. But it's active on SDRAM writes and reads.
On CX calcs the screen buffer is somewhere in SDRAM, and the buffer set by the OS has caching disabled. So writes and reads will be slower.
What happens if you allocate two buffers for both calcs so the default one is never used?

Matrefeytontias · « **Reply #32 on:** June 27, 2014, 06:20:15 pm »

DJ : what you don't understand is that the program is already 6 times faster (according to latest aeTIos's tests) on GS calcs than on color calcs at merely the same frequency (120 MHz for GS vs 132 MHz for color).

Vogtinator : I was wondering that. Will test when I can (probably in some minutes).

Adriweb · « **Reply #33 on:** June 27, 2014, 06:22:45 pm »

Quote from: Matrefeytontias on June 27, 2014, 06:20:15 pm

DJ : what you don't understand is that the program is already 6 times faster (according to latest aeTIos's tests) on GS calcs than on color calcs at merely the same frequency (120 MHz for GS vs 132 MHz for color).

In case you didn't see :

Quote from: aeTIos on June 27, 2014, 06:07:15 pm

I decided to un-overclock my calc and re-run copyscreen1000times.12:53 seconds.

(Or well, I'm not sure where you see this '6x' faster

)

GL&HF anyway ^^

Matrefeytontias · « **Reply #34 on:** June 27, 2014, 06:24:27 pm »

12 ≈ 6 * 2 if I still know how to count. aeTIos's tests are made on a color calc and give 12 seconds, pimath's tests are done on a GS calc and gives 2 seconds.

aeTIos · « **Reply #35 on:** June 27, 2014, 06:26:42 pm »

I don't know if pimaths calc is OCed though. Even then, the max speed doesn't go much over 150MHz. So yeah something really strainge is going on.

Adriweb · « **Reply #36 on:** June 27, 2014, 06:32:10 pm »

Well, it's actually closer so 5 times, but OK, as long as tests are consistent it's good.

pimathbrainiac · « **Reply #37 on:** June 27, 2014, 06:36:13 pm »

I have not OC'd. My calc came with a non-downgrade-able OS, so I updated to 3.6, then ndless'd

Matrefeytontias · « **Reply #38 on:** June 27, 2014, 08:03:08 pm »

So I came up with this, which actually seems to give 167 FPS (I guess this is the maximum the color Nspire's LCD can give) at any frequency on a color calc ; giving a mere 6 seconds at 132 MHz as well as 262 MHz, vs 12.58 seconds at 132 MHz and a bit more than 6 seconds at 262 MHz previously (all are aeTIos's tests).

Only problem, it crashes on exit (that is, after 6 seconds), and I have no idea why.

Code: [Select]

unsigned short *BUFF_BASE_ADDRESS;
void *SCREEN_BACKUP;

void initBuffering()
{
   void *temp;
   temp = malloc(BUFF_BYTES_SIZE);
   if(!temp)
      exit(0);
   BUFF_BASE_ADDRESS = (unsigned short*)malloc(BUFF_BYTES_SIZE);
   if(!BUFF_BASE_ADDRESS)
   {
      free(temp);
      exit(0);
   }
   
   SCREEN_BACKUP = *(void**)0xC0000010;
   
   // Handle monochrome screens-specific shit
   if(is_classic)
      *(int32_t*)0xC000001C = (*(int32_t*)0xC000001C & ~0x0e) | 0x08;
   
   *(void**)0xC0000010 = temp;
}

void updateScreen()
{
   // Screen-access delays make this the fastest method apparently
   memcpy(*(void**)0xC0000010, BUFF_BASE_ADDRESS, BUFF_BYTES_SIZE);
}

void deinitBuffering()
{
   void *temp = *(void**)0xC0000010;
   // Handle monochrome screens-specific shit again
   if(is_classic)
      *(int32_t*)0xC000001C = (*(int32_t*)0xC000001C & ~0x0e) | 0x04;
   *(void**)0xC0000010 = SCREEN_BACKUP;
   free(temp);
   free(BUFF_BASE_ADDRESS);
}

int main(void)
{
   int i;
   
   initBuffering();
   clearBufferB();
   
   for(i = 0; i < 1000; i++)
   {
      memcpy(*(void**)0xC0000010, BUFF_BASE_ADDRESS, BUFF_BYTES_SIZE);
   }
   
   deinitBuffering();
   return 0;
}

Binaries attached, but it does crash your calc.

EDIT : did some tests with the current version of nKaruga (source accessible from github). The game runs at 95 FPS on my 120 MHz grayscale TI-Nspire CAS with Ndless 3.1 r914, calculated by seeing that you take 3.4 seconds to cross 320 pixels by moving one pixel by one pixel.

Streetwalrus · « **Reply #39 on:** June 28, 2014, 03:42:01 am »

I did one last experiment today. I was getting the same results as aeTIos. I tried with CPU at 246MHz and AHB at 35MMHz, the thing crashes after 12 seconds, which makes it even clearer that it's memory that causes the bottleneck.
aeTIos tried different scaling modes in gpSP and unscaled is much faster, not only because of the scaling but also for the same reason.

aeTIos · « **Reply #40 on:** June 28, 2014, 04:05:59 am »

I have not yet tested downclocking my AHB in gpSP-nspire, I'm doing that right now.

Vogtinator · « **Reply #41 on:** June 30, 2014, 05:12:21 pm »

I just tried it on my CX CAS 3.1, the program doesn't crash for me and it takes ~6 seconds = ~167 fps.
I don't know how to interpret it, I don't understand

Quote

giving a mere 6 seconds at 132 MHz as well as 262 MHz, vs 12.58 seconds at 132 MHz and a bit more than 6 seconds at 262 MHz previously

Edit: Could you post a program that writes 1000 times to something else than the screen? I don't think it could make a difference, but just in case..

Matrefeytontias · « **Reply #42 on:** July 01, 2014, 07:34:17 am »

I meant that with this version, the program executes in 6 seconds whether the calc has been clocked to 132 or 262 MHz, whereas with the previous version, the program ran in 12.58 seconds when the calc was clocked to 132 MHz and in 6 seconds when it was clocked at 262 MHz. So the newest version is better.

I'll do that in a minute.

aeTIos · « **Reply #43 on:** July 01, 2014, 07:35:09 am »

Quote from: Matrefeytontias on July 01, 2014, 07:34:17 am

I meant that with this version, the program executes in 6 seconds whether the calc has been clocked to 132 or 262 MHz, whereas with the previous version, the program ran in 12.58 seconds when the calc was clocked to 132 MHz and in 6 seconds when it was clocked at 262 MHz. So the newest version is better.

I'll do that in a minute.

Note that the 132 MHz program also ran at AHB = 33...

Matrefeytontias · « **Reply #44 on:** July 01, 2014, 07:36:25 am »

Well yeah, at this point CPU speed isn't important because it's huge compared to AHB speed (only when it comes to writing memory and doing only that of course).

Author Topic: [Ndless] Help with bottleneck on color calcs only (Read 19228 times)

DJ Omnimaga

Re: [Ndless] Help with bottleneck on color calcs only

Vogtinator

Re: [Ndless] Help with bottleneck on color calcs only

Matrefeytontias

Re: [Ndless] Help with bottleneck on color calcs only

Adriweb

Re: [Ndless] Help with bottleneck on color calcs only

Matrefeytontias

Re: [Ndless] Help with bottleneck on color calcs only

aeTIos

Re: [Ndless] Help with bottleneck on color calcs only

Adriweb

Re: [Ndless] Help with bottleneck on color calcs only

pimathbrainiac

Re: [Ndless] Help with bottleneck on color calcs only

Matrefeytontias

Re: [Ndless] Help with bottleneck on color calcs only

Streetwalrus

Re: [Ndless] Help with bottleneck on color calcs only

aeTIos

Re: [Ndless] Help with bottleneck on color calcs only

Vogtinator

Re: [Ndless] Help with bottleneck on color calcs only

Matrefeytontias

Re: [Ndless] Help with bottleneck on color calcs only

aeTIos

Re: [Ndless] Help with bottleneck on color calcs only

Matrefeytontias

Re: [Ndless] Help with bottleneck on color calcs only