Visitors can check out the Forum FAQ by clicking this link. You have to register before you can post: click the REGISTER link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below. View our Forum Privacy Policy.
Want to receive the latest contracting news and advice straight to your inbox? Sign up to the ContractorUK newsletter here. Every sign up will also be entered into a draw to WIN £100 Amazon vouchers!
It requires decent 3D graphics. It's been designed to require a dedicated card from nVidia/ATI but business machines tend not to have this. It's a question of both feature-set and raw grunt power... some integrated chipsets expose features but are still dog-slow, we're talking 10-50X times slower than a mid-range dedicated card.
I'm a bit out of date on all this, but DirectX used to have a HEL (Hardware Emulation Layer) that would do things in software that the hardware couldn't do, which means everything would work but of course would be much much slower. But there was a way of initialising it to use hardware only. If it's 10-50x slower, then it's probably not a case of performance of the card that's lacking, but that the card lacks hardware support for what you need to do and the software is filling in.
There must be 3D performance benchmarking tools out there you could try.
Integrated graphics hardware reports benchmarks 10-50X slower than half-decent dedicated hardware, using hardware support... it's basically a case of lack of transistors even when the integrated hardware has proper support. Modern graphics hardware is massively parallelised and integrated chips just have far fewer parallel execution pipelines (amongst other things).
I don't actually know if the HEL still exists, or was dropped once shaders got so much more complicated... you can run a program 100s of lines long for every single pixel that's rendered, whereas in the old days as many cycles would be a terrible implementation, you wanted to squeeze a pixel every 10 cycles or less. It's really quite staggering how much graphics has changed in the last 10-15 years, raw graphics power is probably 1000X faster if not more.
I don't actually know if the HEL still exists, or was dropped once shaders got so much more complicated.
You're probably right. I did quite a lot with DirectDraw (that's 2D), and some with Direct 3D - 3 I think, but they changed the whole API after that, and again, and quite possibly again.
What's the bandwidth on of an ExpressCard? ca 2Gbit/s in PCIE mode or 1 PCI Express 1.0 Channel.
You might think that's a lot, but modern GPU come with 16x PCIE 2.0 interface, which is 64 Gbit so 32 times more.
So you'll find the bw of the ExpressCard to be a problem.
Looking at the prices is seems like such a pointless exercise for the time being.
I would consider:
1. A dedicated mid-range desktop PC with a decent GPU, these can be had for ca $500 nowadays. Form factors can be surprisingly tiny at a cost of $$$ and noise (or lower performance).
2. A replacement laptop with a dedicated GPU already built in. They will be a bit more expensive than a basic laptop, but should offer similar performance to the ViDock solution in a much better (more portable) form factor.
In general, unless you're doing math on GPU, which would be double backwards in this case, you're dealing with graphics and that graphics needs to presented, it's much, much better to do this on a PC connected to a huge, color-calibrated display than a tiny laptop with a dim screen.
Portability is a key client requirement - salesmen travelling to customers' sites to demo the software, and apparently most end users have laptops. Requiring all customers to buy a desktop is not seen as a sensible business move (not that I have any involvement on the business side, but I argued at length all the technical reasons and that was the response I got).
Comment