Turbo cache Vs. dedicated memory

Hi, I have Acer 5920G wich has 256MB $video memory running on 400MHz (no more than 500MHz with OC)
And the 2GB system memory of Santa Rosa that runs on 667MHz.

Now I wonder is the shared system memory really faster than dedicated for videocard use???!

btw is there a way to disable or limit Turbo cache on Vista 32bit?

 

Which specific video card do you have?

 

Turning Turbo Cache off is only going to hurt performance. Without it on your total graphics memory will be 256mb plus whatever is allocated via Turbo Cache. With it off your total graphics memory will be 256mb plus nothing.

 

Never mind. I decided to not be lazy and actually look up the specs. Looks like a sweet gaming rig

 

even though it runs at 667mhz it still is going to have MUCH lower bandwidth because it has to travel along the data bus as opposed to the GPU's memory bus.

 

You always want dedicated memory for the GPU.

 

Ofcourse dedicated is much faster.

Although don't bother disabling turbo cache.It's used only when it's needed and believe me,that's rare..

 

You messed something up here... the video memory runs at 400mhz which is double pumped so effectifly it runs at 400mhz x2 = 800mhz. The System memory runs at 333mhz double pumped to 333mhz x2 = 667mhz

 

just want to put it out there an tell all of you that ATI's version of TurboCache (HyperMemory), is currently inferior to nVidia's technology, so if you're looking for a mainstream card with some dedicated and some shared memory, shoot for nVidia

 

Not much of a difference anyways.

 

I say keep the turbocache enabled as it can help at some stages but quite rare

 

PCI-e 1.1 allows up to 250MB/s bi-directional through a single lane. Assuming that the video card is running on a PCI-e X16 interface, that gives a max of 4GB/s bi-directional, which is lower than most dedicated graphic's memory bandwidth.

The maximum theoretical bandwidth provided by dual channel DDR2 667 is 10.67GB/s. However, you have to remember that this bandwidth is shared among the entire system - it is not completely dedicated to the video card in TurboCache/HyperMemory.

So, in all but the rarest of cases, TurboCache/HyperMemory will be slower than dedicated video buffer.

 

I disagree if it was, you would read in the specs of GPU. The dedicated memory does not use system bus which is why so much faster, Turbo cache and Hyper memory use system memory so they operate at system memory speed nothing to do with dedicated memory speed in anyway.

The poll is funny, very much like asking if the Earth is flat, in your poll, answer is so obvious and non debatable that it is not a matter of opinion. It will yeild no useful results.

 

To say, shared system memory will have a bottleneck which will generally impact the performance regardless of speed. Dedicated memory is much faster.

 

Thanks everyone, I totally forgot about memory badwidth as some point, of course dedicated is unbeatable because of bandwidth in the first place, I belive moderators could close this thread, it is pretty much finished.

Not quite funny, I asked a simple question because I couldn't figure out this "obvious" answer, and it is much easyer and convinient to vote in a poll than post a whole comment. As you can see there are 35 votes and only 13 comments so far to back up my point.

 

I apologize I did not mean to offend but I guess I did. Sorry.

 

The nice thing about Turbo Cache is that it is hardware-based. For lower-end cards like the Go 7400, it can make a big difference. However for higher end cards not so much. Regardless, I would suggest leaving it enabled; the card will only take RAM as it needs it.

 

The nice thing about HyperMemory (to me) is that you can actually adjust the vram to know what the card is using, inferior or not. With the TurboCache on my Go 7200, Windows reports that I have 1071MB vram (depending upon which driver version I have installed) while gpu bench programs like 3DMark06 can tell me I have 61MB available vram (while notifying that the current card is not even Vista-capable due to it's low vram).

In other words, I have not found an app yet that can accurately quantify TurboCache availability. This was not a problem with HyperMemory.

(End rant.)

 

In my view, Turbo Cache is a safety net that was implemented by graphic card manufacturers when they realized that games were starting to use exponentially larger amounts of video ram. So, instead of saying, "Oh my lord, we need some sort of method to supplement our video ram or it is going to fill up and cause issues!" they said, "With Turbo Cache, your video card can now utilize system memory for increased performance!"

Thus, its great to have turbo cache as a safety net when you are running out of video memory. You don't have your computer crash or your game slow down to 1 FPS, you just have a little hiccup as the RAM trades current memory data with your dedicated vRam. Its a logical solution.