Wednesday, 10 September 2014

Little gpu bits

I've mostly been taking it easy - i'm not going to be on leave forever (unfortunately) - but i've tried a couple of little things on the gpu code.

First I tried creating a tile-based implementation for the ARM/host version but this runs about 1/2 the speed of the line-oriented one. Not that I really optimised it but that's a lot to make up and i don't see the point; it's a convenient test-bed for experimenting though.

Then I tried creating tile-accurate indexing rather than using the bounding box. This improves the output a small amount on the purely arm version but takes a hit on the epiphany backend since the hit to the arm-side code exceeds the gains on the epiphany-side. It will depend on the workload and it might be worth it for larger triangles. Then again maybe the index isn't helping as much as I thought.

I also started (re)reading about some lighting stuff but didn't get very far.

Feeling pretty lazy today too.

Update: But not too lazy to poke a bit more it seems.

I made a "slight improvement" to the ARM based tile renderer and now it's a bit faster (10%) than the line-based one with a specific test-case. Being lazy the first time I was just processing the tile row by row rather than performing the rasteriser pass across the whole tile first and then processing the fragments afterwards. This just helps the compiler keep more setup data in registers for each loop and is closer to how i'm doing it on the epiphany.

Update: Haven't been able to get into it this last week. I think hayfever season is starting and even before the symptoms hit it just seems to wreck my sleep more than normal. Been really tired/lethargic and not really feeling like doing anything - it just feels like all i'm doing each day is hanging around waiting to escape from it into the unconsciousness of sleep again. Today I even feel like i'm "coming down with something" although i'm pretty sure i'm not and it's just some hayfever related nonsense. I've done a little gardening at least - preparing some garden beds, putting in a few seeds, and rejuvenating some pots.

But as a bit of a puzzle a few days ago I tried to see if i could get the rasteriser loop any faster. I think I can get the inner loop down to 8 cycles with some unrolling, double load/stores and some constant preloads. The previous best was 10 cycles but i'm not sure this new version is practical.

This came out of playing with the idea of breaking the work up into squares (4x4 or 8x8) rather than rows. This has overheads due to performing the edge tests multiple times outside of each pixel test but also reduces the overheads of calculating over the bounding box. But it's one of those things I need a solid afternoon to try out by coding it up.

These tile tests also allow one to determine full coverage outside of the loop - which removes the need for the edge testing calculations at all. So I tried to see if that could save anything in the inner loop; but so far the latency from the z buffer testing has prevented any gains being made. Even assuming I could pipeline that away I think I can only save 1 cycle.

I also toyed with creating an integer rasteriser that stores the framebuffer internally using bytes. For a flat shaded/z-buffered/non-blended triangle I think I can get that down to 7 cycles per pixel (and that's rendered, not just converted to fragments). Is that even useful? Who knows. But to test that idea out I need to work on a new design which will take another solid afternoon as well.

Boycott nvidia, cuda?

nvidia has taken the nuclear option to sue every other gpu maker in existence (apart from ati/amd with which they already have cross-licensing agreements i guess).

Patent trolling is usually the last gasp of a failing business. Which implies that despite (or because of) their overpriced hardware they are failing as a manufacturer; GPus are now commodity items and the margins no longer exist to run their type of high-margin business.

Patents are a cruel abomination which distort the workings of a "free market"; they directly codify rentier behaviour which costs society both economical and technological progress. The only beneficiaries are the unproductive leeches of society at the cost of everyone else.

If you're an engineer or scientist who is currently using or considering cuda for your work I suggest you reconsider both to protest this failure of a strategy and to protect the future value of your work.

Just for nvidia to consider this strategy shows they are not long for this world and choosing to use such a single-supplier would be foolhardy.

Saturday, 6 September 2014

Damocles is Mecenary

Oh boy ...

Seems that Just Add Water was working on a Damocles game!

"We've spent some time on pre-production, coming up with the overall direction, both visually, and as a story, as it's not a straight Damocles remake, it's using parts of the entire Mercenary story arc," JAW boss Stewart Gilray told VG247 today.

Despite the pre-production and close working relationship with original coder Paul Woakes, the project is currently on hold at JAW.

"We've had to shelve it for the moment unfortunately but it's something we are massively excited about coming back to," he confirmed.

Well I hope they don't shelve it for too long. Damocles was the only Amiga game I ever bought so it has a pretty bit spot in my heart.


I haven't been playing games much lately - hacking on code is more rewarding and satisfying and if i'm stuck or had enough or too tired I've been reading junk on the net, or watching a tiny bit of TV.

I like the new Doctor Who and i'm pleased they let him keep his "independent" accent. Although I'm sure i'm not alone in thinking of his character from The Thick of It. "Missy" is still the batshit-crazy HR chick from Green Wing too - which was a great character and it was a nice surprise to see she wasn't just a one-off for the first show (I think the whole 'heaven' and his intro as 'i'm over 2000 years old' may turn into some sort of connection with a particular fictional sandal-wearing character from that era; well probably not but it would be interesting if they did). The americanised torchwood OTOH is just not really very good ... but what can you do eh - the original wasn't really very good either if we're honest but the dumbed-down McGuyver-science stuff is a pretty shitful and unnecessary addition to the show. "Gwen" is a bit of a yummy mummy though ;-)

Back to games - since I haven't been playing much i'm kind of not sure why i'm terribly interested in these but there are some games coming up that are looking pretty sweet all the same.


Evolution make great car games and I was always a fan of the way they handled hills and long-range views adding a bit of flair from their origins in flight simulators (at least that's what i understand from reading it somewhere). The amount of processing power available now is just staggering and allowing for some really amazing graphics and world simulation.

It will be interesting to see how the social aspect works. People just seem to love that kinda shit for some reason and racing games seem like a good fit due to their competitive nature and accessibility and that repetitive play continually improves your times.

Hmmm ... I still have a copy of Motorstorm Apocalypse I haven't got around to opening yet, amongst half a dozen other games. I preferred the WRC games for the most part but the loading times were always shit - that's another big "next generation" thing they seem to have addressed in DRIVECLUB.

The Tomorrow Children

Visually stunning and aesthetically unique - something that simply wasn't possible just a few years ago because neither the hardware nor the mathematics existed. I just wish all that async compute stuff filtered down to the APUs (faster).

Like part of DRIVECLUB the multiplayer is not fully synchronous. Everyone occupies the same persistent world in real(ish?)-time but they don't have the scalability problem of trying to render 500 dolls at on screen at the same time by simply not showing other people unless they're interacting with the global state (e.g. they 'fade in' to pick up something, then fade out taking the something with them). This is a neat technological solution to the scalability issue but also addresses the confrontational aspect of most "traditional" multi-player games. Although player vs player games are quite popular a lot of people don't like them, me included, and this is one of many games adopting a different approach.

I'm not sure it will be the sort of game I would play because it looks like it will suck too much time and due to the multiplayer persistence force you to be constantly active and involved; but graphically and technically there is a lot of cool stuff going on there.

No Man's Sky. Or in Irish apparently "nomans-sky".

Technically very interesting again. This is probably something previously possible but nobody dared to try on quite this scale - or never managed to get the algorithms good enough to pull it off (assuming Hello Games can). Obviously i've been playing with noise lately which would be one of the underlying building blocks of making this work. There's absolutely no "random" in the noise algorithms although they are intended appear that way.

I think it's rather cool that although it is a persistent shared galaxy "the mathematical chances of ever meeting anyone else is approximately zero and therefore anyone you meet is simply a product of an over-active imagination" - to paraphrase a certain book.

Actually the imagination runs wild on this one with the potential scope of the game - whole galaxy which can never be fully explored, galaxy-wide civilisations, traders, pirates, conflicts, archaeological remains and relics, forgotten settlements or downtrodden settlers, whole planets to roam. Reality might not be quite so fantastical but i'm still interested to see what sort of game they come up and the universe they're algorithms will create and the potential is there for expansion toward loftier goals for years to come. I was interested to read that they have procedural room generation as well - will we be able to actually (finally) go inside every building we see? If you can do it for one there's nothing to stop you doing it for them all. That alone would be revolutionary.

I'm guessing that the main "impediment" to reaching the end-game will be the fuel required to travel from star to star and thus the main driving mechanism for the whole game will be acquiring that fuel. So my guess is the gameplay will revolve around performing tasks which earn the dough which can be used to buy the fuel to travel forward, but you're always limited in how far you can go. Although only 1 in 100 planes will have "advanced" life my guess is the barren ones will have rare/valuable minerals to be mined that will make them worth visiting too. If each solar system has one particular resource of interest and the livability of planets is based on the chemistry of the solar system it would make sense that resource abundance would also be correlated.

If they're smart they'll sell fuel canisters for real money - although I personally think it's pretty stupid to buy such "virtual goods" myself if people are dumb enough with their money then why not let them spend it? This would literally turn the game into a virtual tourism simulator which isn't such a despicable idea. Actually it will be interesting to see if they actually do this - although it could potentially imbalance most games of this type with so many planets in the galaxy it's effectively impossible to "wreck" this game that way.

Since you don't actually meet people my guess is the main 'multiplayer' aspect of the game will take place either IRL - via screenshots, blogs, faecebook, twatter, youtube, twitch and so on, or a similar in-game universal comms/atlas system. vidphones and "subspace" communications? That was definitely a mainstay in 30-60s sci-fi. Since they plan on some sort of multiplayer later on, and since you can't physically meet, my guess is it will be have to be some sort of tele-holo-deck type thing to fit in the game world.

TBH it's really hard to imagine that their geography/flora/fauna/architecture/spaceship/route algorithms will provide enough variety to satisfy punters but it's really hard to imagine the sort of big numbers they're playing with (actually it's impossible). Basing the models on the way the real world works - using an alternative but consistent chemistry and physics - at least has the potential to be just as wild and varied.

I think there were a couple of other interesting things coming up but they slip my mind for the moment.

If these more novel games get any success (or even if they don't) i'm sure more will try which will just further add to the breadth of the gameosphere.

Too noisy

Been playing a bit with simplex noise. Interesting how much you can create from the same basic function and kind of cathartic and easy on the brain.

The following were all created with the same basic noise. 4 frequencies are combined in different ways. I'm using Z as an animator so they all smoothly animate.

Blobs of liquid. This uses max(). The frequency is the same for each layer by the amplitude is altered. Note that this is purely 2D and the depth appears due to attenuation.

Smoke or writhing organic mass. This uses max(abs()) and a lower frequency.

Lava lamp ringlets. Scales to an integer and selects one of the bits from the integer. Again the depth is from attenuation and scaling in this case.

Friesian cow-hide. Or a coastline. Or a burning piece of paper. Threshold with multiple frequencies. Works very nicely as a blending mask for image transition.

I've mostly been playing with the hash function to try and create something epiphany efficient whilst still working sufficiently well. For 3D noise the current candidate uses 3 lookup tables to provide a basic hash of the x/y/z locations and they are combined using floating point multiplies and/or other bit ops. I'm only using random values which works most of the time although a better choice should be possible. It may be worth just going back to the permutation array of the original code as I realised I can implement that in only 256 bytes if i need to. I still don't know how it will run on the machine as the simplex setup code is pretty expensive too but I haven't looked at how to optimise it yet. Originally I was looking at the 2D noise because it was simpler but as 3D noise is just so much more useful I will target that instead.

I also created a 16-element spherical set of vectors for the base noise gradients. First I used an inscribed cube and some others I made up but then I finally found the code by Jon Leech (hint: it's at the bottom of the page!) which models electron repulsion to try to evenly space the points across the sphere. This does create a nicer result. 16 is used since it's a lot easier to calculate the modulus of 16 than it is for 12.

I do see patterns showing up particularly with the ringlet algorithm - lines at 45 degrees showing up as you zoom out - but this shows up for the original too. The noise is definitely not zero-mean. If I average over many frames I get fairly regular blobs at 45 degrees showing up also - although they are at 90 degrees to the ones that show up zoomed out, but again this is also present with the original Simplex noise hash function and gradient set.

Thursday, 4 September 2014

"Sentinel Saves Single Cycle Shocker!"

Whilst writing the last post I was playing with a tiny fragment to see how just testing the edge equations separate to the zbuffer loop would fare. It's a bit poor actually as even the simplest of loops will still require 9 cycles and so doesn't save anything - although one wouldn't be testing every location like this so it's pretty much irrelevant.

Irrelevant or not I did see an opportunity to save a single cycle ...

If one looks at this loop, it is performing 3 edge equations positive tests and if that fails it then has to perform a loop-bounds test.

0000015e:       orr.l     r3,r18,r19      /|                  1                                             |
00000162:       fadd.l    r18,r18,r24     \|                  1234                                          |
00000166:       add.s     r0,r0,#0x0001   /|                   1                                            |
00000168:       fadd.l    r19,r19,r25     \|                   1234                                         |
0000016c:       orr.l     r3,r3,r20       /|                    1                                           |
00000170:       fadd.l    r20,r20,r26     \|                    1234                                        |
00000174:       bgte.s    0x0000017a       |                     1                                          |
00000176:       sub.s     r2,r0,r2         |                      1                                         |
00000178:       bne.s     0x0000015e       |                       1                                        |

On in C.

  for (int x=x0; x < x1; x++) {
     if ((v0 >= 0) & (v1 >= 0) & (v2 >= 0))
       return x;
     v0 += v0x;
     v1 += v1x;
     v2 += v2x;

Problem is it needs two branches and a specific comparison check.

Can a cycle be saved somehow?

No doubt, don't need Captain Obvious and the Rhetorical Brigadettes to work that one out.

Just as with the edge tests the sign bit of the loop counter can be used too: it can be combined with these so only one test-and-branch is needed in the inner loop. After the loop is finished the actual cause of loop termination can be tested separately and the required x value recovered.

It's not a sentinel it's just combining logic that needs to be uncombined and tested post-loop in a similar way to a sentinel.

000001ba:       orr.l     r3,r18,r19      /|                  1                                             |
000001be:       fadd.l    r18,r18,r24     \|                  1234                                          |
000001c2:       sub.s     r0,r0,#0x0001   /|                   1                                            |
000001c4:       fadd.l    r19,r19,r25     \|                   1234                                         |
000001c8:       orr.l     r3,r3,r20       /|                    1                                           |
000001cc:       fadd.l    r20,r20,r26     \|                    1234                                        |
000001d0:       orr.s     r1,r0,r3         |                     1                                          |
000001d2:       blt.s     0x000001ba       |                      1                                         |

Or something more or less the same in C with the pro/epilogues and probably broken edge cases:

  int ix = x1 - x0 - 1;
  while ((v0 >= 0) & (v1 >= 0) & (v2 >= 0) & (ix >= 0)) {
    ix -= 1;
    v0 += v0x;
    v1 += v1x;
    v2 += v2x;
  if (ix >= 0) {
    return x0 + ix;

I suppose it's more a case of "Pretty Perverse Post Pontificates Pointlessly!"

Or perhaps it's just another pointless end to a another pointless day.

Might go read till I fall asleep, hopefully it doesn't take long.

ezegpu stuff

Did a bit more playing around on the ezegpu. I think i've hit another dead-end in performance although I guess I got somewhere reasonable with it.

  • I re-did the controller so it uses async dma for everything. It didn't make any difference to the performance but the code is far cleaner.

  • I tried quite a bit to get the rasteriser going a bit faster but with so few cycles to play and all the data xfer overheads with there wasn't much possible. I got about 5% on one test case by changing a pointer to volatile which made the compiler use a longword write. I got another 5% by separating that code out and optimising it in assembly language using hardware loops. This adds some extra overhead since the whole function can't then be compiled in-line but it was still an improvement. However 5% is 40s to 38s and barely noticeable.

    This is the final rasteriser (hardware) loop I ended up with. The scheduling might be able to be improved a tiny bit but I don't think it will make a material difference. This is performing the 3 edge-equation tests; interpolating, testing and updating the z-buffer; and storing the fragment location and interpolated 1/w value.

      d8:   529f 1507       fsub r2,r44,r21        ; z-buffer depth test
      dc:   69ff 090a       orr r3,r18,r19         ; v0 < 0 || v1 < 0
      e0:   480f 4987       fadd r18,r18,r24       ; v0 += edge 0 x
      e4:   6e7f 010a       orr r3,r3,r20          ; v0 < 0 || v1 < 0 || v2 < 0
      e8:   6c8f 4987       fadd r19,r19,r25       ; v1 += edge 1 x
      ec:   a41b a001       add r45,r1,8           ; frag + 1
      f0:   910f 4987       fadd r20,r20,r26       ; v2 += edge 2 x
      f4:   047c 4000       strd r16,[r1]          ; *frag = ( x, 1/w )
      f8:   69ff 000a       orr r3,r2,r3           ; v0 < 0 || v1 < 0 || v2 < 0 || (z buffer test)
      fc:   278f 4907       fadd r17,r17,r23       ; 1/w += 1/w x
     100:   347f 1402       movgte r1,r45          ; frag = !test ? frag + 1 : frag
     104:   947f a802       movgte r44,r21         ; oldzw = !test ? newzw : oldzw
     108:   90dc a500       str r44,[r12,-0x1]     ; zbuffer[-1] = zvalue
     10c:   b58f 4987       fadd r21,r21,r27       ; newz += z/w x increment
     110:   90cc a600       ldr r44,[r12],+0x1     ; oldzw = *zbuffer++;
     114:   009b 4800       add r16,r16,1          ; x += 1

    Nothing wasted eh?

    Using rounding mode there are I think 4 stalls (i'm not, but i should be). One when orring in the result of the zbuffer test and three between the last fadd and the first fsub. Given the whole lot is 10 cycles without the stalls that doesn't really leave enough room to do any better for the serial-forced sequence of 2 flops + load + store required to implement the zbuffer. To do better I would have to unroll the loop once and use ldrd/strd for the zbuffer which would let me do two pixels in 30 instructions rather than one in 16 instructions. It seems insignificant but if the scheduling improved such that the execution time went down to the ideal of 10 cycles and then an additional cycle was lopped off - from 14 to 9 cycles per pixel - that's a definitely not-insignificant 55% faster for this individual loop.

    The requirement of even-pixel starting location is an added cost though. Ahh shit i dunno, externalising the edge tests might be more fruitful, but maybe not.

  • I did a bunch more project/cvs stuff; moving things around for consistency; removing some old samples, fixing license headers and so on.

  • I tried loading another model because I was a bit bored of the star - I got the Candide-3 face model. This hits performance a lot more and the ARM-only code nearly catches up with the 16-core epiphany. I'm not sure why this is.

Although there are some other things I haven't gotten to yet i've pretty much convinced myself this design is a dead-end now mostly due to the overhead of fragment transfer and poor system utilisation.

First I will put the rasterisers and fragment shaders back together again: splitting them didn't save nearly as much memory as I'd hoped and made it too difficult to fully utilise the flops due to the work imbalance and the transfer overheads. I'm not sure yet on the controller. I could keep the single controller and gang-schedule groups of 2/3/4 cores from the primitive input - i think some sort of multiplier is necessary here for bandwidth. Or I could use 3 or 4 of the first column of cores for this purpose since they all have fair access to the external ram.

Wednesday, 3 September 2014

simplex noise, less memory

I thought i'd look at something a bit different today: noise. Something to get the fragment shaders doing some more work.

I've looked at some of this before but it's been a while and never had much use for it.

I started with "wavelet noise" but when I realised it needed big lookup tables I went back to looking at the simplex noise algorithm. It seems wavelets are being used to create bandwidth limited versions of existing noise so that it scales better; but this isn't something I need to worry about.

A paper and implementation by Stefan Gustavson and others pretty much had me covered but I wanted to try and remove the 512+256 element lookup tables used to hash the integer coordinates to save some memory on the epiphany.

I came up with two working solutions in the end.

The first one uses a 32-element lookup table of prime numbers to implement a 2D hash function. I just grabbed the first 32 primes (20 apart) for the table and fiddled with eor/mul and shift until I had something that seemed to work. I arbitrarily chose 32 because it was a nice round number.

// there's nothing particularly good or useful here

    static final int[] hasha = {
        71, 173, 281, 409, 541, 659, 809, 941,
        1069, 1223, 1373, 1511, 1657, 1811, 1987, 2129,
        2287, 2423, 2617, 2741, 2903, 3079, 3257, 3413,
        3571, 3772, 3907, 4057, 4231, 4409, 4583, 4751

    private static int hash16(int a, int b) {
        return ((((b ^ hasha[a & 31]) * (a ^ hasha[b & 31])) >> 5) & 15);

Because I was only interested in the 2D case I changed the gradient normal array to 16 elements so I didn't have to modulo the result as well. TBH it's kind of surprising it works as well as it does since hashing numbers is pretty tricky to get right and I really didn't know what I was doing.

When I started I didn't realise exactly what it was for so once I had a better understanding of why it was there I thought i'd try an existing integer hash function. In general they failed miserably but I found one that came from the h2 database which worked sufficiently well.

    private static int hash(int x) {
        x = ((x >> 16) ^ x) * 0x45d9f3b;
        x = ((x >> 16) ^ x) * 0x45d9f3b;
        x = ((x >> 16) ^ x);
        return x;

    private static int hash16(int a, int b) {
        return (hash(a * b + a + b)) & 15;

I used the (a*b+a+b) calculation to turn it into a 2D hash function.

So this final version requires no lookup table for the gradient table permute at all - nice. But it requires 3 integer multiplies - not so nice for epiphany. And even the other version needs an integer multiply and thus the same costly fpu mode changes on epiphany.

Since I only need a limited number of output bits it might (should?) be possible to change this to using float multiplies to avoid the costly mode change; but this is something for further study. The first version might make this easier.

Screenshots ... this first is a simple 4-octave fractal noise generated using the 2D Simplex Noise code from Stefan. I think the 2D noise function has a small bug because it's using the 12-point 3D gradient bases which don't always evaluate to vectors of the same length in 2D but it isn't apparent once fractal noise is generated as here.

The next one is an example using the naive hash function (it may be a different scale to the others since I ran it separately). Covering 4 octaves hides some problems it might have but I've done some very basic testing to larger scales and it seems about as stable and nicely random as the others.

And the final shot is using the M2 hash function and 16 gradients evenly spaced around the unit circle rather than 12 evenly spaced around the unit sphere as in the traditional version.

Look about the same to me?

I don't know if it's useful for anything I might do or if it is even fast enough to run in a shader on the epiphany but I learnt a couple of interesting things along the way.


I've been doing some little bits and pieces on the ezegpu code as well.

  • Changed the way async dma works in ezecore so that you can use either dma channel, enqueue your own DMA blocks and chains, and increased the queue length for more outstanding requests. I kept the old api compatible and interoperable;
  • Fixed the rasteriser to use this new queue;
  • Added some async dma to the controller but until everything uses it it wont pay off. It got messy enough that I need to redo it with the new goal in mind, it's not much code but I haven't gotten to it yet;
  • Made the backends open the framebuffer so the same 'gles2 demo' can run against different backends by just relinking them so as to simplify comparisons;
  • Added a NEON matrix multiply - it's 3x to 5x faster than a C version;
  • Added a NEON scale+clamp RGBA float to byte for the all-ARM version. This is 3x faster than the C version although i've got the channels messed up so the colours are wrong;
  • Did a bunch of cvs + code management stuff.

Together the NEON changes amount to a 6% improvement to the total runtime of the all-ARM code for my current testing case (8x8x8 stars). Nothing major, although it goes up on simpler scenes mostly due to the faster RGBA float to byte conversion.