What's the situation regarding the 64bit executable?

I'd like DirectX 11 up and running before they even start to worry about the 64bit exe. From what I have heard and read, DX11 is a much more efficient version that DX10 is and this will usually result in better performance with the same or even more eye-candy enabled. (confirm/deny)

[CaptainDoggles]

Confirm. DX11 supports tessellation shaders, which make graphical transformations a snap.

[NedLynch]

confirm

My own experience, after supporting DX11 with a patch in Shogun2 the game looked better and framerates went up. The general consensus seems to be you can just do more with each pixel more efficiently.

[Buchon]

Originally Posted by speculum jockey

I'd like DirectX 11 up and running before they even start to worry about the 64bit exe. From what I have heard and read, DX11 is a much more efficient version that DX10 is and this will usually result in better performance with the same or even more eye-candy enabled. (confirm/deny)

DX11 contain AA processing optimizations, witch mean better performance with AA enabled and use better HDR precision witch mean better HDR lighting.

Also can do high quality shadows processing without performance loss (no aliased shadows)

[TUSA/TX-Gunslinger]

Maybe that's why AA is wierd in current DX10 release

That would certainly make every competitor squirm and explain the length of development time.

Heheh.... this could be very interesting come monday, except I'm going overseas for a week.

S!

Gunny

Originally Posted by Buchon

DX11 contain AA processing optimizations, witch mean better performance with AA enabled and use better HDR precision witch mean better HDR lighting.

Also can do high quality shadows processing without performance loss (no aliased shadows)

This is correct, although some here during development thought that DX11 would be a hindrance to performance, it won't it will improve FPS and the graphic detail, If a game is coded correctly in DX10 then applying DX11 shouldn't be a whole big deal, (so I've been told).

[Skoshi Tiger]

Back to the original topic, when DCS A10 was released there was a major issue with Track IR not being recognised in the 64bit version, so until Natural Point got around to fixing their software I was forced (Seriously , who could go back to not having a trackIR?) to use the 32bit version of the game.

Since the Track IR software was fixed and I was able to use the 64bit version I haven't really noticed any boost in performance on my system.

64 bit is good to have, but the software has to take advantage of the extra memory space.

Cheers!

[louisv]

That's just it, because a lot of people still run in 32bit (even if the hardware has been 64bit since the later models of Pentium IVs), the devs have to keep the two versions very similar and will not really take advantage of the wider data path until a large majority have made the move.

Then programmers will be able to make bigger programs.

Remember the days of 16bit ? The 286, 386 ?

Programs were quite a bit smaller then, with an address space of 2 to the 16th power being 64K on the 8088, the first PCs. The 286 and 386 had a bigger space of 1MB, or 20bit of address space.

So to recap,

16bit: 64KB of memory
32bit: 4GB
64bit: 18.4 X 10^9 GB or about 18.4 Giga GB or 18.4 Exabytes

18.4EB is a big number, I wonder when we will go 128 !

[ZaltysZ]

Originally Posted by louisv

Remember the days of 16bit ? The 286, 386 ?

Programs were quite a bit smaller then, with an address space of 2 to the 16th power being 64K on the 8088, the first PCs. The 286 and 386 had a bigger space of 1MB, or 20bit of address space.

8088 had 16-bit registers, so only 16-bit long addresses were possible, what gave those 64KB. However, there were also possible to use segmented memory access, which combined segment selector and offset to allow access more memory than 64KB. Hardware had means to use 20-bit address space (1MB), which could be accessible by software via segmented access. Usually 640KB were available to user, and upper region were used by BIOS.

268 and 386 added 24-bit and 32-bit protected modes respectively, whose extended available address spaces to 16MB and 4GB.

[Igo kyu]

Originally Posted by ZaltysZ

8088 had 16-bit registers, so only 16-bit long addresses were possible, what gave those 64KB. However, there were also possible to use segmented memory access, which combined segment selector and offset to allow access more memory than 64KB. Hardware had means to use 20-bit address space (1MB), which could be accessible by software via segmented access. Usually 640KB were available to user, and upper region were used by BIOS.

268 and 386 added 24-bit and 32-bit protected modes respectively, whose extended available address spaces to 16MB and 4GB.

That's almost right, but probably due to language differences, it doesn't read quite correctly to me.

Segmented memory addressing was standard on the early IBM compatible PCs.

The Intel 8086 (16 bit) started segmented addressing which gave it one megabyte of address space, then Intel made the 8088 (which was in some ways an 8 bit chip though it used 16 bit registers, as the 8086 and the earlier "8 bit" chips had). Because the 8088 was sort of 8 bit, though it had a one megabyte address space like the 8086, it used cheaper 8 bit support chips, and IBM chose the 8088 for their PC, presumably because the support chips (which wouldn't necessarily come from Intel in the case of either CPU) for the 16 bit 8086 were more expensive.

Segmented memory addressing was such a mess, it gave Intel a legitimate six month lead over the Motorola 68000, but that mess kept running for five or ten years due to "IBM compatibility".