Technical question!

My question has to do with amount of memory on my IGP and bottlenecks. I have ATI 1150. In the catalyst I can set aside 256MB system RAM for GPU. Shared but dedicated, the card can then draw as much as an additional 256MB shared from general RAM. I am convinced that because of things like Buss size and speed issues that this goes beyond the useful abilities of the card and therefor does not return performance. I set my UMA at 128MB so total is 256MB as explained above. I have read about discrete card family's having the MFG add additional memory beyond the functional abilities of the system because of the bottlenecks mentioned. Can anyone help explain these limitations and quantify so I can get an Idea of what amount of memory to set aside so as to not be useless? I do not run Vista but with it's 128MB requirement setting UMA at 64MB would register as 128MB. If I have not laid this question out well please give input and I will try and clarify. Thanks much!

 

Bumped. Please help.

 

Regardless of at which level you set your shared VRAM at, the performance difference while gaming is going to be negligible as the X1150 is a weak card by its own right.
If you are using the computer not for gaming, consider disabling Turbocache altogether so that your computer can use all of your RAM effectively.
In case you're wondering, 64MB of VRAM is sufficient to run Vista using your card.

 

I am very happy you responded but you talk Turbocache which is Nvidia not ATI which is Hyper Memory does not really matter and as you pointed out 64MB would have been enough. Turbo cache is considered better but I have ATI so I am stuck. You do not seem to understand the strategy of the card, but that is OK I am looking for that and no one seems to know! The diff so what gives?

 

let me clarify that. turbocache is more effective of an implementation of shared memory than ati's hyper memory. turbocache is hardware based (fast) and hyper memory is software based (not as fast).

however, this means that all ati cards can use hypermemory, while only cards with the turbocache feature can use it. (at this point basically all the cards nvidia ships out can enable this feature)

also- just because nvidia's cache system is better than ati's doesn't mean that EITHER of them are good. on the contrary- in both cases they are much too slow to render themselves effective for any practical gaming use. games will slow down tremendously if you have a 256 meg video card and run an application that is actually using 512 megs of memory - and that is true both of ati and nvidia. (however, in the past, had a game demanded 512 megs of vram, it would not have been able to run at all on a 256 meg card)

just set aside 128 megs of shared ram for your ati integrated card. its not correct to call it "dedicated" - that is the term for onboard ram, but i suppose that you can say you have "dedicated" that ram to the purpose of fulfilling your gpu. doing general desktop work, watching a movie, etc, its unlikely that the gpu will decide to borrow more ram from the system, even though it could choose to take an additional 128 megs on its own whim.

 

I know what you mean when you say do not call it dedicated, but do not know if you are correct. It is in fact a hybrid and is “dedicated” system RAM. So the GPU goes to a certain part of RAM instead of searching all RAM also Info is not broken apart and spread throughout RAM for this reason in theory is faster. And therefor different than traditional shared memory. As such deserves distinction. I do set at 128MB but does anyone know some #'s based on core 400Mhz and RAM 533Mhz of when I might saturate it's capabilities?