Memory size

[pepokun]

Hi
is there a difference between a graphic card with 128Mb or 256Mb, for example in the FX5900 or Radeon 9800?
For playing games, of course.
Thanks in advance

 

[cleeve]

256 megs will give a very, very small performance advantage when running at unholy high resolutions with Antialiassing.

It also gives you the option of using high resolution textures in next-generation games, but the only game in existance that I've heard of making use of more than 128 megs of textures is half-life 2.

Basically, 256 megs is a complete waste right now. 128 megs is more than ample, hell, 64 megs will do just fine for some time...

------------------
Radeon 9500 (modded to PRO w/8 pixel pipelines)
AMD AthlonXP 2000+
3dMark03: 3529

 

[ufo_warviper]

Yeah, get a 128 meg card for now. THe new chips aren't powerful enough to make good use of the extra memory. Besides, when 256 megs of textures are actually used in a game, you wil have already upgraded.

My OS features preemptive multitasking, a fully interactive command line, & support for 640K of RAM!

 

[Spitfire_x86]

256 MB VRAM is a BIG waste of money. It's only useful for marketing and if you are running current games at 1600 x 1200 with 4x or 6x AA and stuck with AGP 4x. But no current card is capable of delivering playable framerate in current games with such settings.

64 MB is OK for now if you don't use anti-aliasing with modern games. 128 MB is enough.

----------------
<b><A HREF="http://geocities.com/spitfire_x86" target="_new">My Website</A></b>

<b><A HREF="http://geocities.com/spitfire_x86/myrig.html" target="_new">My Rig & 3DMark score</A></b>

 

[Guest]

Another important point is that they may in fact run slower and lower resolution(lower being 1024 and 1280) because of the higher memory latency caused by the big amount of memory.

Question why the hell would they ship that Volari DUo with 512 meg??

 

[Ion]

Marketing!!! :tongue:

 

[TheGreatGrapeApe]

Question why the hell would they ship that Volari DUo with 512 meg??

Multi-monitor and High RES. support. Dual VPUs is great for things like multi monitor, and in that case the more th better. And who knows how it's designed, perhaps it's inefficient to EVER go to system memory, so they build it to have MORE than enough onboard to do the job, so as to not suddenly gor from 50FPS to 7FPS because of overloading a smaller amount.
It's likely cheaper memory so don't expect wicked speeds out of the 512, and therefore the timings are also less important (when it's that slow).

We shall see, remember this is not the same architecture, and I have a feeling that we will see some things that occur (problems / benifits) that are unique to the design.

We shall see.



- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <A HREF="http://www.redgreen.com" target="_new"><font color=green>RED</font color=green> <font color=red>GREEN</font color=red></A> GA to SK

 

[cleeve]

From what I've read, the Volari duo needs dedicated memory for each GPU.

Remember the old Rage 128 MAXX? it had 64 megs... but really, it only had 32 megs of memory per GPU.

Well, the same with the Volari. It has 256 megs for each GPU, they don't share memory. It's a pretty inefficient way to deal with the memory, like the MAXX did... because duplicates of the same textures have to be loaded into each memory bank.

------------------
Radeon 9500 (modded to PRO w/8 pixel pipelines)
AMD AthlonXP 2000+
3dMark03: 3529

 

[TheGreatGrapeApe]

That makes alot of sense. I knew it was likely architecture specific.
Even the Matrox 512mb (the only 512 card I had ever heardof) appears to be a variant of the new HR256, not the Parhellia it was rumoured to be.

I'm still very intrigued, but not overly optimistic about the Volari cards.

Thanks for clearing that up.


- You need a licence to buy a gun, but they'll sell anyone a stamp <i>(or internet account)</i> ! <A HREF="http://www.redgreen.com" target="_new"><font color=green>RED</font color=green> <font color=red>GREEN</font color=red></A> GA to SK

 

[Guest]

Ya I saw somewhere afterwards that each gpu would have his dedicated memory...
Maybe its ennificient to load the textures twice but isnt it faster afterward? One gpu doesnt have to wait to access memory because the other one is using it?
Just a tought. Anyway if the card perform well and its priced aggresivly, it could be an intersting product.
I also think that for a first product its very interseting...