-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
maybe a bit of a noob question but
- Thread starter ethan
- Start date
Soldato
- Joined
- 28 Nov 2005
- Posts
- 13,002
it doesnt really need to ...
it will have 256mb.... or 512mb of EVEN faster memory
it will have 256mb.... or 512mb of EVEN faster memory
It's a pity system ram isn't fast enough then because if it was, it might cut down on the expense of some of the high-end gpu's : Also it would have a benefit on system speed as well. Hopefully, this is the way it might go in the future 
Last edited:
They do when they run out of onboard texture memory.
The problem is that system memory is much slower than that on graphics cards.
I've sometimes wondered why we don't see graphics card memory used as system memory. It's expensive, but not hideously so. If you could hook it up right, surely 512meg of GDDR3 at 1ghz would give a significant boost over standard DDR2 at 667 or whatever
The problem is that system memory is much slower than that on graphics cards.
I've sometimes wondered why we don't see graphics card memory used as system memory. It's expensive, but not hideously so. If you could hook it up right, surely 512meg of GDDR3 at 1ghz would give a significant boost over standard DDR2 at 667 or whatever
My 6800nu performs horribly on 3dmark 06 - I think I must have changed something in the past where my GFX card doesnt tap into my system ram - because the graphics card goes from generating 15-20 fps then to 1fps suddenly barely moving lowering my score to 800.
ethan said:Why can't current high-end cards tap into extra system ram if needed? Isn't system ram fast enough because it is in connection with current configerations, plentiful
because system bandwidth is about 5,000-6,000MB/s in todays high-end systems, whereas the newest graphics card have bandwidth of about 30,000-40,000MB/s.
U can already get DD2-800 Ram for Intel skt775 mobos and its possible to run it @ the native 800 (1:1 FSB ratio). Would need to lower the multiplier and have some serious phase cooling though 
Recent skt775s can run allday on DDR2-667 with a 1333FSB to give some insane clock speeds
Back in the 386/486 days you had to buy some mega expensive CPU cache seperately to plug the cache card into the mobo. Cost was extreme. Something like £150-200+ for a 128-256kb L2 cache. Early CPUs did not always come with any L2 cache or very little (16k was a common amount).
Once there is a new mobo architecture which ditches all the useless legacy stuff like PCI,Serial + Parallel + PS2 Ports etc etc then there is no reason why you cannot have high 5 figure mem bandwidth on the mobo which will then shift the bottleneck back to the CPU.
Recent skt775s can run allday on DDR2-667 with a 1333FSB to give some insane clock speeds
Back in the 386/486 days you had to buy some mega expensive CPU cache seperately to plug the cache card into the mobo. Cost was extreme. Something like £150-200+ for a 128-256kb L2 cache. Early CPUs did not always come with any L2 cache or very little (16k was a common amount).
Once there is a new mobo architecture which ditches all the useless legacy stuff like PCI,Serial + Parallel + PS2 Ports etc etc then there is no reason why you cannot have high 5 figure mem bandwidth on the mobo which will then shift the bottleneck back to the CPU.
Last edited: