After the bleak, joyless work of releasing software, plugging holes of preventing users from clicking buttons they should not have clicked in the first place, writing unctuous documentation and release notes and wrapping code into neat little installers with tiny bows and bells that gently tingle when touched, there are few things as profoundly satisfying as taking code that looked like line noise in the first place, applying some non-trivial transformation to it and still have it look like line noise, only now it’s twice as fast, or half as long or of some other inaccessible quality that can, on some technical level, be referred to as “cool”.

Quite some time ago, I decided that this time, “cool” will mean “runs on graphics hardware, possibly faster”. For this end, I took my trusty T61 to work, downloaded the latest CUDA toolkit, found out that nearly all examples crashed when compiled, installed the latest beta drivers from NVIDIA, noticed that everything worked now and started playing around with the examples.

After some time marveling at ball pit simulations and similarly telling examples, I printed out the CUDA programming guide and spent a considerable amount of time marveling at how well the NVidia green looked when printed with our institute’s color laser printer, before reading its more important parts.

Technical Details

Roughly said, modern graphics hardware from NVIDIA consists of an array of multiprocessors. Each multiprocessor has eight cores, some shared memory, a controller unit and units for transcendental functions. Kernels (methods that run on the device, as opposed to host code) are automatically distributed to the available multiprocessors (their number depending on the hardware, with as many as 120 on the Tesla cards).

Kernel invocations are organized in 3-dimensional thread blocks and 2-dimensional grids. A single thread block can have at most 512 threads (i.e. x × y × z ⩽ 512) and is always executed by a single multiprocessor. Thread blocks are organized in a grid (with at most 65,536 elements on each dimension), which are distributed to the available multiprocessors, and thus inherently scalable.

A multiprocessor divides the threads of a block into warps and all threads of a single warp are executed in parallel. Optimal performance is reached when the control flow in threads of the same warp does not diverge based on the input data, the cores are optimized to execute the same code in several threads on different data (therefore the name SIMT—Single instruction, multiple thread—for this architecture).

There is also a memory hierarchy, from registers to on-die shared memory to global device memory. Data has to be explicitly copied from the system RAM to device memory, which is costly and should be minimized.

In other words, if you have a computation that has to be executed many, many times on different input data, move it the GPU instead and use the freed CPU cycles for shuffling around the data. In other words, ZOOOOOOM.

Some Experiments

Back in the day, we had to write code for training Gaussian Mixture Models. At the core of the algorithm, an auxiliary vector containing each point in the training set has to be created for each Gaussian. The creation of the auxiliar vector includes exponentiation, which proved to be quite costly, since it accounted for 50% of the running time of the program¹.

Instead of doing comparedly boring optimization work for running that algorithm on a single processor, and even more boring work to have it run on two processors, I partially ported it to have the auxiliary vectors be computed on the GPU, with the result that by moving only this one part of the computation leads to the three-fold speed increase for the whole program.

On my machine (2 GiB RAM, 2.2 GHz Core 2 Duo, NVS 140M with 2 multiprocessors), running the C++-only version² takes 33s for 90,000 points and three Gaussians. Running the computation with the same data set on the GPU takes 11s.

How is it done

In order to invoke a kernel running on the GPU, the code has to be compiled with nvcc, NVIDIA’s compiler for CUDA. It separates device, which is compiled to some binary code, from host code which is transformed and handed over to the platform compiler. Therefore, it is needed to write some intermediate methods which can be called from “normal” C++ and that take care of shuffling data to the device and invoking the kernel methods³:

typedef float PROB;
 
#define BLOCK_SIZE 16
 
Matrix d_in;
Matrix d_out;
 
__device__ __constant__ KernelGaussian d_params[20];
 
extern "C"
void LoadPoints(Matrix in, size_t params) 
{
    d_in.height = in.height;
    d_in.width = in.width;    
    size_t size_in = in.width * in.height * sizeof(PROB);
 
    cutilSafeCall(cudaMalloc((void**) &d_in.elements, size_in));
    cutilSafeCall(cudaMemcpy(d_in.elements, in.elements, size_in, cudaMemcpyHostToDevice));
 
    d_out.width = in.width;
    d_out.height = params;
 
    size_t size_out = d_out.width * d_out.height * sizeof(PROB);
    cudaMalloc((void**) &d_out.elements, size_out);
}
 
__global__ void computeAuxVectors(Matrix in, Matrix out) 
{
    int p = blockIdx.y * BLOCK_SIZE + threadIdx.y;
    KernelGaussian g = d_params[blockIdx.x];
    PROB d = (in.elements[p] - g.mean) / g.stddev;
    out.elements[blockIdx.x * out.width + p] = g.factor * expf(-0.5*d*d);
}
extern "C"
void RunComputeAuxVectors(
        KernelGaussian* g, 
        std::vector<std::vector<PROB> >& out, 
        size_t points, size_t params)
{
    cutilSafeCall(
        cudaMemcpyToSymbol(
          d_params, g, 
          params*sizeof(KernelGaussian), 
          0, cudaMemcpyHostToDevice));
 
    dim3 dimGrid(params, points / BLOCK_SIZE);
    dim3 dimBlock(1, BLOCK_SIZE);
    computeAuxVectors<<<dimGrid, dimBlock>>>(d_in, d_out);
    cutilCheckMsg("Kernel execution failed");
 
    for(size_t i = 0; i < params; ++i) {
        cutilSafeCall(
            cudaMemcpy(
              &out[i][0], &d_out.elements[i * points], 
              sizeof(PROB) * points, 
              cudaMemcpyDeviceToHost));
    }
}

The matrix d_in holds the points, which are constant over the whole computation and therefore copied over to the device exactly once. d_out is the matrix that holds the auxiliary values computed for each point and each Gaussians. For faster retrieval, the Gaussians are stored in constant device memory (d_params), which is set to contain 20 elements at most, i.e. the algorithm is limited to learn mixture models with at most 20 Gaussians.

In the method LoadPoints, device memory is allocated for the points and the output vectors and the points are copied from host to device memory using cudaMemcpy.

The actual invocation of the kernel is done in the host method RunComputeAuxVectors, which first copies the parameters into the constant storage and invokes the kernel computeAuxVectors using the new <<>> syntax. grid holds the dimensions of the grid, which is number of gaussians× points / BLOCK_SIZE. The block size is the width of a thread block and somewhat arbitrarily chosen to be 16⁴, currently this also limits the number of points to be a multiple of 16⁵. After the blocking kernel invocation, the results are copied back into host memory.

Important:

• My card only supports single-precision floats, newer ones also have double-precision
• blockIdx and threadIdx are global variables provided by CUDA compiler
• __global__ methods are run on the device and can be called from both host and device code
• Currently, the device memory for d_in and d_out is never freed, the program simply exits

Results

Working with CUDA is definitely fun (if rewarding), the system itself is quite unforgiving though. Wrapping each invocation of a CUDA method into cutilSafeCall therefore is strongly recommended, in case of an error it will print a (not always helpful) error message and exit. Though in this case the code running on the GPU itself is quite trivial, it can be hard to figure out what is going wrong, because the GPU itself is a black box, and it’s not possible to simply print out a message⁶. It’s possible to emulate the code on the CPU, and there are other tools available which I haven’t tried out.

I’ve also started toying around with my original idea, having the constraint checks for the TIGER query evaluation run on the graphics hardware and initial results are mixed. Still, if nothing at all, this gives me justification to spent a lot of money on expensive graphics hardware for my next computer—it’s all for science, this harsh and demanding mistress.

1. There might be a smart way of getting rid of the exp() call, though I haven’t found it yet
2. compiled using g++ 4.4.0 -O3
3. looking at the SDK examples was quite helpful here
4. I didn’t try out other block sizes
5. this is a restriction of the implementation, one could simply pad the points with however many zeros as needed and compute a little more
6. or maybe it is, and I don’t know how to do it…

• +2 GB RAM (4 GB overall)
• 3G card
• USB Port Replicator
• DisplayPort->DVI converter

Being in Switzerland, I had the choice between CH and US keyboard layout. Since CH is physically the same as the German layout (105 keys), I took it and haven’t really noticed that the keys have different symbols (as far as most of the special characters are concered) printed on that what appears on the screen when I hit them. I should really get Das Keyboard after all.

The X301 is not quite the workhorse the T61 is, which is especially noticeable from graphics speed, although that might be due to an Intel driver being in several different transitions right now. The SSD compensates for that, booting and starting up is a breeze. Fortunately, I don’t have to do much booting these days, because unlike other major graphics hardware creators, Intel’s developers are able to support powersave modes on Linux hardware, both Suspend-to-RAM and Suspend-to-disk.

Everything else works more or less, even the DisplayPort, dutyfully serving my 24″ screen at the university. The pièce de résistance is definitely the 3G card. I had to install it myself, which was quite a bit more involved than what I would have liked, but it worked on the second try. I got myself a prepaid data contract which allows me to surf the net for 3 CHF/h, which is a good complement to free wireless access at most public hotspots in Switzerland, which is due to SWITCH. The wireless became especially hand today, when I was shopping for a new router without having a clue about what’s good. Standing in the Media Markt (yes, this very pillar of the German retail market also metastasised into Switzerland) network hardware aisles browsing the web with a notebook probably didn’t look as superior as doing the same thing with a smartphone, but it helped me finding a router.

What really sets the X301 apart is the weight. I’m quite used to it by now, since I’ve had it for over a month, but it’s usually the first response I get when I hand it over to somebody else. It is, however, very noticeable in direct comparison, and the T61 feels clunky and hugely oversized when I carry it around or use it on my lap. The battery lifetime is okay, usually I get 3h under Linux. There’s probably some more running time to be gotten, but I’m more or less done tweaking the system for now and using it for actual work; or for blogging, emailing and chatting on my nice, new, comfy chair.

Upsides

For such an early release and given my kind-of new and kind-of exotic hardware (Quadro NVS 140M, a G86), it worked quite well. XRandR 1.2 is a breeze, and 2D performance was ok. No hard lockups, either.

So, in short terms, what does work:

• XRandR 1.2
• External Output
• Somehow fast 2D

Downsides

I couldn’t put my finger on it, but I think the picture quality on the VGA port was lower than with the Nvidia drivers (I have an external screen with a 1680×1050 resolution). Also, video overlay doesn’t work yet on the newest cards, so the video output was very grainy.

I also saw a lot of wrongly rendered pixmaps in FF3. No real showstopper, but also not very nice. Also, suspend/resume didn’t work yet.

I have changed back to the Nvidia driver for now.

Nicht abgebildet: Das T60 der Dozentin.

Im Oktober 2004 fing so auch bei mir das Zeitalter des mobilen Rechnens an, und seitdem war ich selten länger als eine Woche von meinem Notebook getrennt. Ob ich beim ersten Praktikum bei FAST zur Schande der Firma an meinem eigenen Rechner gearbeitet, in Schweden mit mehreren MBit aus dem Internet gesaugt, in Bochum die Nächte durchzockt, in Heidelberg am PhraseHunter gehackt, in Tübingen in der großen Datenpromiskuität teilgenommen oder in Gladbeck damit meinen BA geschrieben habe – mein T41p hat mir immer treue Dienste geleistet.

Eine kleine Zusammenfassung:

• mehrere hundert Stunden Spielspaß
• einige Notabschaltungen beim Erstellen von FSTs im Hitzesommer 06
• 12 Kernelversionen (2.6.9 – .21, mit Ausnahme von 2.6.11)
• 4 Umzüge
• 1 Reparatur
• 0 zufriedenstellende Grafiktreiber für Linux

Was bleibt? Auf jeden Fall eine Vorliebe für ThinkPads und die Frage, ob Desktop-Rechner noch Sinn machen. So, genug Reflexion… Ab jetzt wirklich nur noch Substanz.

In einer nächtlichen Aktion habe ich gestern auch tatsächlich Windows XP installiert, nachdem Vista nach ein paar Stunden auf wundersame Art und Weise mitsamt der Wiederherstellungspartition seinen Hut nahm und zuletzt auf dem Weg nach Kuba gesichtet wurde, um einige Exportrestriktionen ad absurdum zu führen.

Bisher bin ich eigentlich auch recht zufrieden, abgesehen von einigen kleineren Unschönheiten (Hardware-Mixer und Lautstärkeregler gehen nicht etc.), man hätte es schlimmer erwarten können. So kann ich mich also daran erfreuen, ganze drei ThinkPad-Generationen (T42, T43, T60) übersprungen zu haben und mich überall dort, wo man sich ausstattungsmäßig verbessern konnte, auch verbessert habe.

So, genug der Weihrauchverteilung, ab jetzt nur noch Substanz.

Torsten meint: Da wartet man doch gerne eine Woche länger!

Um die Hardwarebasis für die nächsten 3 Jahre zu schaffen, habe ich mir in den letzten Wochen und Monaten das Gel aus den Haaren gespart und seit Monaten Hersteller- und Hardwareseiten durchwühlt, mit mir selbst gerungen und gewartet – auf die Erleuchtung, auf neue Modelle, auf eine Entscheidung.

Aber nun ist es soweit! Die Lieferzeiten für meinen Laptop der Wahl, ein T61 ND114GE sind auf lediglich eine Woche gefallen und nur noch die furchtsame Krämerseele in mir hielt mich für einige Stunden ab, die Bestellung abzuschicken. Dann aber verlor ich die Selbstkontrolle und unterlag den Lockungen der Technologie.

Zusammen mit der vielen Hardware kaufe ich mir natürlich auch mal wieder eine Herausforderung, um mein durch Vernachlässigung angerostetes Geek-Image wieder aufzumöbeln. Eine kleine Zusammenfassung der Treibersituation unter Linux:

• Grafik: im neuesten Nvidia-Treiber unterstützt
• Sound: ALSA-Treiber in Entwicklung
• WLAN: Treiber von Intel in Entwicklung
• Intel Turbo Memory: noch nicht unterstützt

Gute Zeiten kommen auf mich zu.

This does not make my search for a new laptop easier. I had more or less settled for another ThinkPad, being quite satisfied with my almost 3 years old T41p, but now the ATI laptops are an alternative again. Moreover, I am not completely convinced that Lenovo keeps IBM’s quality standards.