The GT300/Fermi Thread - Part 2!

[saaya]

Well that's an interesting comparison but your only calculating the effect of chip size. But bigger dies also run hotter so there is also increased cost from components, cooling and possibly also PCB quality.

yeah i mentioned that somewhere in that huge post
i think it wont have too much of an impact on the actual cost of the card, i calculated 25$ extra for the GF100 cards vs rv870 cards...
this makes sense especially because most gf100 cards will be 470s which have 1.5gb of memory while competing 5870 and 5850 cards will probably have 2gb, cause 1gb makes them look too whimpy and 2gb is the next step up for them... plus they will be 320bit vs 256bit for rv870 which isnt a huge difference, the added pcb complexity should make them that expensive... an additional pcb layer for a full size atx motherboard is only around 5-10$, which is 3x the size of a vga, so... and a heatsink isnt that expensive either, going from an rv870 to a more powerful heatsink probably costs you a few extra dollars, not that much more... heatpipes costs less than 1$ a piece these days...

"Again"? No, I didn't want to make fun of you! The post was a very good read, thx

P.S. The trains are stopped in whole NRW because we have a freaking storm here O.O There are many big trees right in front of my flat...

hehehe ok then

damn... do you think train traffic from frankfurt to hannover and hannover to hamburg will be blocked or restriced next week? :S

creating an efficient engine will pay off more in the end; Unreal engine has proven this, runs great on low end hardware, can look pretty spliffy on high end hardware.

thats true... flexibility and efficiency are very important... i think the main 2 reasons why the u3 engine has been so successful are something else though...
1. gears of war - this game still looks awesome today compared to the latest games... this made everybody realize that the engine can be used to create awesome graphics
2. very good dev tools and dev support/relations

of course an engine has to be flexible and efficient, but that alone wont get you attention from anybody, and it wont help you sell your code..
you need a flagship product, an example of what your code can do when pushed to the limit, ie crysis, gears of war, stalker...

doom3 and hl2 failed to impress, and as a result their engines failed to capture market share as well, no matter how efficient or flexible they are...

JC building such an engine for their Tech 5 would increase chances of a commercial success, freeing up funds for future developments.

we will see... they are working on it for a while and what ive seen so far doesnt look all that impressive and revolutionary, it kinda looked like borderland before they went cell...

obligatory "saaya" multi-quote

why are you quoting random posts?

[FischOderAal]

hehehe ok then

damn... do you think train traffic from frankfurt to hannover and hannover to hamburg will be blocked or restriced next week? :S

Cebit? Today it's still a bit chaos at the DB but the storm is gone. No tree hit me

doom3 and hl2 failed to impress, and as a result their engines failed to capture market share as well, no matter how efficient or flexible they are...

Doom 3 failed to impress, I agree, but HL2? I think it looked darn good back when I played it with a 9600 Pro. If I remember correctly my jaw dropped But it didn't scale up that well.

The Source-Engine is rather conservative on it's features. Although it makes heavily use of multiple Cores, more than any other engine I've seen (or I didn't notice it that well...). In TF2 I get 150 FPS on 1680x1050 (max details, max AA, AF, Bloom etc) and as soon as I disable multicore rendering it chokes.

The Source-Engine has proven it's worth imho and is used in a variety of games, although it's limited to games developed by Valve iIrc.

Hopefully Episode 3 (or HL3) will be out this year (or early next) and maybe it will use an updated Engine with DX11.