Wow! AMD have alot of games supporting 3DNow!
- Thread starter SSP
- Start date
I'd rather have a t-bird than a P3, but to play devil's advocate games are optimized for sse too.
Yeah, but the problem is, what kind of 3DNow optimizations?
It's very simple for a lazy game developer to detect the instruction set and tweak a few token routines to make trivial use of it. It's another matter entirely to rewrite a significant portion of the game engine and spend time doing some true optimizations. Game developers still seem extremely reluctant to put serious effort into 3DNow, while they fall ass backwards for Intel's SSE. It's funny, because the one time somebody put in the time to really code for 3DNow -- in AMD's special Quake 2 patch -- the results were phenomenal.
To really show that most games don't make good use of 3DNow, just look for an option to disable it. On the majority of games, you won't find one, because they don't want people to benchmark and see that their claimed "support" of 3DNow is usually just enough to get the sticker on the box.
Modus
It's very simple for a lazy game developer to detect the instruction set and tweak a few token routines to make trivial use of it. It's another matter entirely to rewrite a significant portion of the game engine and spend time doing some true optimizations. Game developers still seem extremely reluctant to put serious effort into 3DNow, while they fall ass backwards for Intel's SSE. It's funny, because the one time somebody put in the time to really code for 3DNow -- in AMD's special Quake 2 patch -- the results were phenomenal.
To really show that most games don't make good use of 3DNow, just look for an option to disable it. On the majority of games, you won't find one, because they don't want people to benchmark and see that their claimed "support" of 3DNow is usually just enough to get the sticker on the box.
Modus
I believe what Modus says. Think about it, it makes good sense. Play Quake II with the patch, and you'll see. Play other games, where they 'claim' to have 3DNow support, and watch how your Intel counterparts get similar or better results. 
-RSI
-RSI
On my old system K6-2 300mhz with a voodoo2 the patch for quake2 gave me an 40% performance increase! Luckily that was the only game I played ;-)
MODUS is exactly correct. programmers are REALLY lazy. Quake 3 should run WAYY better on a K6-2 then it does normally.
my opinion on programmers has never been a good one. they can be extremely innovative, and extremely lazy at the same time!
Quake 2 required AMD to make a patch to show of what the K6-2 was made of.
ALOT of games claim 3DNow! support, but I really really doubt that they tried to use it in any useful way (Quake 3 is a perfect example of that, which is why people are releasing special DLL files for 3DNow! optimization).
I'd like to see the performance of QUake 3 after those DLL files are finished on an Athlon..
my opinion on programmers has never been a good one. they can be extremely innovative, and extremely lazy at the same time!
Quake 2 required AMD to make a patch to show of what the K6-2 was made of.
ALOT of games claim 3DNow! support, but I really really doubt that they tried to use it in any useful way (Quake 3 is a perfect example of that, which is why people are releasing special DLL files for 3DNow! optimization).
I'd like to see the performance of QUake 3 after those DLL files are finished on an Athlon..
Intel has 80% of CPU market share right now.Programmers arent lazy, they work within business constraints.
They have limited resources (both time and money).
Situation will only change if AMD grabs 50%. Maybe a year or two from now they will have it...
They have limited resources (both time and money).
Situation will only change if AMD grabs 50%. Maybe a year or two from now they will have it...
"Intel has 80% of CPU market share right now. Situation will only change if AMD grabs 50%. Maybe a year or two from now they will have it..."
yeah but that's just instruction sets, not even talking about optimizations for the CPU's quirks.
the only way to get programmers to become better at what they do is to give them less leeway on the amount of computer power they can use.
ie, if the computer hardware industry wasn't going so fast, we might still see the same graphics quality we see today, but with a lower requirements list, becuase they're forced to squeeze more speed out of the same components.
yeah but that's just instruction sets, not even talking about optimizations for the CPU's quirks.
the only way to get programmers to become better at what they do is to give them less leeway on the amount of computer power they can use.
ie, if the computer hardware industry wasn't going so fast, we might still see the same graphics quality we see today, but with a lower requirements list, becuase they're forced to squeeze more speed out of the same components.
Everything is BLOATware today. Little or no time is spent optomizing... the goal is to rush games to market, so why spend 3 months getting a game to run well on a 600mhz CPU, when in 3 months your target hard core gamers will all be running 1ghz+? It's always been that way. Cheaper to waste CPU cycles when CPUs have them to spare, than to spend precious time optimizing. That's ignoring optimizing for specific instruction sets, which takes even more time.
Most programmers today write routines that call functions from libraries. Those libraries were written with generic x86 in mind. One could go through and rewrite many of the functions in assemby, taking advantage of modern instructions (additional to x86)... but it is sooooo tedious... and programmers that really know how to do it are few and far in between.
Most programmers today write routines that call functions from libraries. Those libraries were written with generic x86 in mind. One could go through and rewrite many of the functions in assemby, taking advantage of modern instructions (additional to x86)... but it is sooooo tedious... and programmers that really know how to do it are few and far in between.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 24K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 20K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter