Reverend at The Pulpit #6

[Reverend]

Here we go again, and probably the most boring Pulpit musing so far!

Site Stuff
Nothing much has been happening on my end where the site is concerned. That's because of a priority change forced upon me (a bit more later).

The main thing is to spend as much time as I can in finishing up my pixel shader project/research. It's about 80% done right now I'd say. It's taken an unusually long time (it's taken about the length of this year, although definietly not a year's worth of work since on average I spend only about an hour on it per day ) because I simply haven't enough time dedicated to it and the fact that I started out making it DX9HLSL-exclusive and then decided around April to add in OpenGL as well (no HLSL here for rather obvious reasons). During its course, I had, quite a few times, decided to dump it because it was too huge a project for myself to undertake alone but then decided against this everytime because it would mean wasted time. So, it will be completed. Just don't ask when! Of course, it can be benchmarked (otherwise Dave will be annoyed with me!).

One sticking point is whether I should let the public have the source. I haven't been keen on the idea (still isn't) because this usually means additional time post-release (feedback and stuff). But then, as I asked John Carmack for his opinion on some structural details, he said :

One point I would make is that it is valuable to have the source to the synthetic benchmark avaialble. Driver vendors will be doing complete call traces anyway, you might as well make the source public so developers can see exactly which paths get what performance, which isn't always clear from a one line title on a performance graph.

I do worry about the incredible number of variables there are to test for over something as simple as texture access -- texture size, texture filtering, texture compression / depth, etc, especially when permuted through large numbers of texture units. Still, synthetic benchmarks do have value, and an open one would be a benefit over commercially supported ones.

John Carmack

So... well, I'm still undecided at the moment but making it open source seems a better idea now. Maybe I'll have my own definition of "open source" though. Wonder what John meant by his last sentence...

Also, I had asked Tim Sweeney about the possibility of having an arrangement whereby Beyond3D can have early looks at his UnrealEngine3. His reply :

Hi Anthony,

It's too early to start thinking about this. Let's talk again in a year. (This isn't a "yes" or even a "maybe", it's just an "it's way too early.") Our goal in facilitating benchmarks is to incentivise hardware vendors to optimize for our performance patterns and feature usage, and though lots of stuff is up and running, we're a ways away from having nailed down our final feature set.

UnrealEngine3 is the name for the next major step in Unreal technology, aimed at DirectX9 hardware as the absolute minimum spec, and scaling (way) upwards from there.

-Tim

Well, let's see what happens one year from now.

A little while back, I was pleasantly surprised to receive an email from someone who basically "kick-started" me in the video card reviewing business/hobby. He was a NVIDIA DevRel manager at the time but he's now (his position) "Lead Evangelist, Windows 3rd Party Gaming and Entertainment, Microsoft Corporation". I'm thinking of interviewing him. He hasn't agreed yet (only asked him yesterday) but it should be interesting if it happens.

Personal Stuff
Well, it has been a rather hectic time for me. Going down to Melbourne/AUS and coming back within the day isn't much fun (you want to talk about jet-lag? How about 14 hours flight-time to-and-fro with a 11 hour break between flights but not being able to get any shuteye the whole while?). Had to do it to bring my parents home from my sister's place in Melbourne. Mom's 72, dad's 82 and has Parkinson's. Needless to say, my time management gets shot to hell with them around and the need to send my old man to a nursing home is both sad but also unavoidable given my personal life arrangements.

S'pose life's like that and many have to go through the same thing...

Oh, and try the following :

While sitting at your desk make CLOCKWISE circles with your RIGHT FOOT. While doing this, draw the number "6" in the air with your RIGHT HAND. Can you do that without having your foot change direction? Don't cheat and try to do it in really-slow motion.

 

[Joe DeFuria]

Actually, I'd say this is one of the more interesting ones so far.

Glad to see you're sticking with the project. I'm really not sure why an "open source" synthetic benchmark would be more valuable than a closed one.

Perhaps Carmack is thinking purely in terms of "IHVs being able to optimize for your benchmark?" Well, what if one of the goals is NOT to have IHVs "optimize" for it?

Pesronal Stuff....must be really hard to do what you think is best for your Dad given your situation. Good luck.

Oh and that foot - 6 thing is f'd up! I'll take solace in the fact that I can at least pat my head and rub my stomach at the same time. (Or is that pat my stomach and rub my head....)

 

[K.I.L.E.R]

You were in Melbourne?

Goodness, why didn't you stop over at my place?

How do you like the wheather there? Pretty hot compared to where you live eh?

Thunderstorms are occuring now.

Putting a family member in a home is never easy but sometimes there's no other option.

BTW: Your little game was easy.

 

[Sage]

Bah, that silly spin-the-chair-draw-a-six thing was not easy, it took no more mental effort than using a mouse. But then I've always been VERY talented with my hands and have recieved many emphatic compliments on my coordination . Try this- (not suggested for die-hard christian fanatics, or in their presence) make a fist then stick out your pointer and pinky and move your pointer straight up and down while moving your pinky in a circle. It's supposed to put the hex on whoever you're pointing it at, and it is pretty hard until you embrace the Dark Master ( ). I got it down pretty fast....



edit:

KILLER: I just read your signature, it's hysterical!!!!!! Thank you so much for ruining the LOTR movies for all the fools who can't pick up a damned book and read a classic!

 

[Reverend]

Glad to see you're sticking with the project. I'm really not sure why an "open source" synthetic benchmark would be more valuable than a closed one.

Perhaps Carmack is thinking purely in terms of "IHVs being able to optimize for your benchmark?" Well, what if one of the goals is NOT to have IHVs "optimize" for it?

I think he meant developers would be able to look at my source and correlate that to the performance results, i.e. be able to see the behaviour characteristics of the hardware being tested for a given feature. It is far more easier to deduce that if the source is available (which means anyone can mess around with the app, change things, see the difference, etc).

I think it also means he is open to the idea that whatever I come up with may be useful to him, hehe

 

[Tim Murray]

I said a few months back that open-source benchmarks were the way of the future, but nobody listened to me. Oh well.

Perhaps Carmack is thinking purely in terms of "IHVs being able to optimize for your benchmark?" Well, what if one of the goals is NOT to have IHVs "optimize" for it?

See, that's the beauty of an open-source benchmark. How does one cheat in it? Answer: They can't. Well, sure, a driver vendor could just do what they do what they are doing now with application-specific optimizations, but that would work for exactly one version of the source. Jimbo Johnson down in Mississippi reorders a few shaders, changes very minor things that have no impact on performance on whatsoever. By doing this, he defeats the detection mechanism. Cheats are then exposed.

It's the only way to make a fair benchmark, really.