Share system RAM with my GPU VRAM

[chrvontab]

Hi

I have 2 gb VRAM on my GTX 770. If i wanted to share some of my 16 gb of system RAM with my GPU, how would i do it?

 

[IInuyasha74]

There would be a setting in BIOS but really you don't want to in honest. While in some situations you could make use of more than 2GB vRAM, the drastically slower speed of that RAM would be a massive bottleneck to performance if it was used.

 

[chrvontab]

There would be a setting in BIOS but really you don't want to in honest. While in some situations you could make use of more than 2GB vRAM, the drastically slower speed of that RAM would be a massive bottleneck to performance if it was used.

Oh okay. Thanks. But won't the system automatically compensate for the loss of speed? I'm on windows 8.1 btw and my RAM is 2400Mhz (though there is an error where it writes wrong sometimes and creates corrupt data so i only run on 2200 Mhz which works fine) if it makes any difference. As far as i can see the GPU memory clock is running 1700-1800 mhz (during MSI Kombuster test + GPU-Z stats)

 

[rgd1101]

You can't use system memory as gpu memory, if something is required more then 2gb of vram, replace your gpu.

 

[chrvontab]

You can use system memory as gpu memory, if something is required more then 2gb of vram, replace your gpu.

I'm not rich...

 

[IInuyasha74]

There would be a setting in BIOS but really you don't want to in honest. While in some situations you could make use of more than 2GB vRAM, the drastically slower speed of that RAM would be a massive bottleneck to performance if it was used.

Oh okay. Thanks. But won't the system automatically compensate for the loss of speed? I'm on windows 8.1 btw and my RAM is 2400Mhz (though there is an error where it writes wrong sometimes and creates corrupt data so i only run on 2200 Mhz which works fine) if it makes any difference. As far as i can see the GPU memory clock is running 1700-1800 mhz (during MSI Kombuster test + GPU-Z stats)

Yes it will compensate by throttling yoru GPU clock down to save power, because it is being starved by the slow system RAM speed.

I will teach you a little about RAM while we are here. First things first, the width of the data interface. You probably saw when you bought your GTX 770 that it has a 256-bit memory bus. What this means is there are 256 wires inside of your GPU that are used for transferring data between the GPU core itself and the RAM of the GPU clock. These wires effect RAM performance as you might think, by increasing by the same amount with each additional wire.

For example, lets say it can transfer 30GB of data every second on 64-wires. It would be able to transfer 60GB of data every second using 128-wires, 90GB of data every second using 192-wires, and 120GB of data every second using 256-wires.

So this is the first big difference between your system and the GPU, as your system RAM only has a 128-bit connection to the rest of the system, half as much as the GPU RAM to the GPU itself.

Then there is the RAM type. your system RAM is DDR3, what this means is in short that it transfers information twice every time it is able to. Your RAM actually runs at 1100Mhz, but because it can send data twice it has the same performance as RAM that is running at 2200Mhz so we often simplify things and write it as such because it makes the calculation for determining total amount of data transferred every second simpler.
Your GPU on the other hand is using GDDR5 RAM. This is twice as good, and sends information four times every time it is able to. So while you read the RAM on your GPU is 1500Mhz, its doing the same work as RAM running at 6000Mhz.

So with this we can get a rough calculation of the bandwidth of the two RAM types in use.

Your system RAM has a max bandwidth of:
2200 x 128/8 = 35.2GBps

Your GPU RAM max bandwidth:
6000 x 256/8 =192GBps

So your GPU RAM transfers 540% more data than your system RAM. On top of that your system RAM has other demands on it such as the needs of the CPU, and there are other factors too. Long story short, if your GPU was using that RAM instead, your GPU performance would be limited to about the same as a GTX 730.

 

[chrvontab]

There would be a setting in BIOS but really you don't want to in honest. While in some situations you could make use of more than 2GB vRAM, the drastically slower speed of that RAM would be a massive bottleneck to performance if it was used.

Oh okay. Thanks. But won't the system automatically compensate for the loss of speed? I'm on windows 8.1 btw and my RAM is 2400Mhz (though there is an error where it writes wrong sometimes and creates corrupt data so i only run on 2200 Mhz which works fine) if it makes any difference. As far as i can see the GPU memory clock is running 1700-1800 mhz (during MSI Kombuster test + GPU-Z stats)

Yes it will compensate by throttling yoru GPU clock down to save power, because it is being starved by the slow system RAM speed.

I will teach you a little about RAM while we are here. First things first, the width of the data interface. You probably saw when you bought your GTX 770 that it has a 256-bit memory bus. What this means is there are 256 wires inside of your GPU that are used for transferring data between the GPU core itself and the RAM of the GPU clock. These wires effect RAM performance as you might think, by increasing by the same amount with each additional wire.

For example, lets say it can transfer 30GB of data every second on 64-wires. It would be able to transfer 60GB of data every second using 128-wires, 90GB of data every second using 192-wires, and 120GB of data every second using 256-wires.

So this is the first big difference between your system and the GPU, as your system RAM only has a 128-bit connection to the rest of the system, half as much as the GPU RAM to the GPU itself.

Then there is the RAM type. your system RAM is DDR3, what this means is in short that it transfers information twice every time it is able to. Your RAM actually runs at 1100Mhz, but because it can send data twice it has the same performance as RAM that is running at 2200Mhz so we often simplify things and write it as such because it makes the calculation for determining total amount of data transferred every second simpler.
Your GPU on the other hand is using GDDR5 RAM. This is twice as good, and sends information four times every time it is able to. So while you read the RAM on your GPU is 1500Mhz, its doing the same work as RAM running at 6000Mhz.

So with this we can get a rough calculation of the bandwidth of the two RAM types in use.

Your system RAM has a max bandwidth of:
2200 x 128/8 = 35.2GBps

Your GPU RAM max bandwidth:
6000 x 256/8 =192GBps

So your GPU RAM transfers 540% more data than your system RAM. On top of that your system RAM has other demands on it such as the needs of the CPU, and there are other factors too. Long story short, if your GPU was using that RAM instead, your GPU performance would be limited to about the same as a GTX 730.

Thanks alot! That was really a nice and understandable explanation 😉. i will figure sometimes out. Thank you again!!

BTW i think there is a ghost breakdancing besides me :/

 

[IInuyasha74]

Haha well happy to help.

Btw what game you trying to play? You can probably get your desired effect by carefully tweaking the settings. The two biggest vRAM users are increased resolution and any time of AA (AA, FXAA, MSAA, SSAA, ect.). If you notice your vRAM maxing out, just drop AA a little will probably fix it as that literally multiplies the amount of space used in vRAM.

Need any more help feel free to ask

 

[Mark_37]

There would be a setting in BIOS but really you don't want to in honest. While in some situations you could make use of more than 2GB vRAM, the drastically slower speed of that RAM would be a massive bottleneck to performance if it was used.

Oh okay. Thanks. But won't the system automatically compensate for the loss of speed? I'm on windows 8.1 btw and my RAM is 2400Mhz (though there is an error where it writes wrong sometimes and creates corrupt data so i only run on 2200 Mhz which works fine) if it makes any difference. As far as i can see the GPU memory clock is running 1700-1800 mhz (during MSI Kombuster test + GPU-Z stats)

Yes it will compensate by throttling yoru GPU clock down to save power, because it is being starved by the slow system RAM speed.

I will teach you a little about RAM while we are here. First things first, the width of the data interface. You probably saw when you bought your GTX 770 that it has a 256-bit memory bus. What this means is there are 256 wires inside of your GPU that are used for transferring data between the GPU core itself and the RAM of the GPU clock. These wires effect RAM performance as you might think, by increasing by the same amount with each additional wire.

For example, lets say it can transfer 30GB of data every second on 64-wires. It would be able to transfer 60GB of data every second using 128-wires, 90GB of data every second using 192-wires, and 120GB of data every second using 256-wires.

So this is the first big difference between your system and the GPU, as your system RAM only has a 128-bit connection to the rest of the system, half as much as the GPU RAM to the GPU itself.

Then there is the RAM type. your system RAM is DDR3, what this means is in short that it transfers information twice every time it is able to. Your RAM actually runs at 1100Mhz, but because it can send data twice it has the same performance as RAM that is running at 2200Mhz so we often simplify things and write it as such because it makes the calculation for determining total amount of data transferred every second simpler.
Your GPU on the other hand is using GDDR5 RAM. This is twice as good, and sends information four times every time it is able to. So while you read the RAM on your GPU is 1500Mhz, its doing the same work as RAM running at 6000Mhz.

So with this we can get a rough calculation of the bandwidth of the two RAM types in use.

Your system RAM has a max bandwidth of:
2200 x 128/8 = 35.2GBps

Your GPU RAM max bandwidth:
6000 x 256/8 =192GBps

So your GPU RAM transfers 540% more data than your system RAM. On top of that your system RAM has other demands on it such as the needs of the CPU, and there are other factors too. Long story short, if your GPU was using that RAM instead, your GPU performance would be limited to about the same as a GTX 730.

Whoah... I just googled difference b/w RAM and VRAM and ended up here.
I must say,
that
was
sexy.

And super interesting.

 

[Dimi516]

There would be a setting in BIOS but really you don't want to in honest. While in some situations you could make use of more than 2GB vRAM, the drastically slower speed of that RAM would be a massive bottleneck to performance if it was used.

Oh okay. Thanks. But won't the system automatically compensate for the loss of speed? I'm on windows 8.1 btw and my RAM is 2400Mhz (though there is an error where it writes wrong sometimes and creates corrupt data so i only run on 2200 Mhz which works fine) if it makes any difference. As far as i can see the GPU memory clock is running 1700-1800 mhz (during MSI Kombuster test + GPU-Z stats)

Yes it will compensate by throttling yoru GPU clock down to save power, because it is being starved by the slow system RAM speed.

I will teach you a little about RAM while we are here. First things first, the width of the data interface. You probably saw when you bought your GTX 770 that it has a 256-bit memory bus. What this means is there are 256 wires inside of your GPU that are used for transferring data between the GPU core itself and the RAM of the GPU clock. These wires effect RAM performance as you might think, by increasing by the same amount with each additional wire.

For example, lets say it can transfer 30GB of data every second on 64-wires. It would be able to transfer 60GB of data every second using 128-wires, 90GB of data every second using 192-wires, and 120GB of data every second using 256-wires.

So this is the first big difference between your system and the GPU, as your system RAM only has a 128-bit connection to the rest of the system, half as much as the GPU RAM to the GPU itself.

Then there is the RAM type. your system RAM is DDR3, what this means is in short that it transfers information twice every time it is able to. Your RAM actually runs at 1100Mhz, but because it can send data twice it has the same performance as RAM that is running at 2200Mhz so we often simplify things and write it as such because it makes the calculation for determining total amount of data transferred every second simpler.
Your GPU on the other hand is using GDDR5 RAM. This is twice as good, and sends information four times every time it is able to. So while you read the RAM on your GPU is 1500Mhz, its doing the same work as RAM running at 6000Mhz.

So with this we can get a rough calculation of the bandwidth of the two RAM types in use.

Your system RAM has a max bandwidth of:
2200 x 128/8 = 35.2GBps

Your GPU RAM max bandwidth:
6000 x 256/8 =192GBps

So your GPU RAM transfers 540% more data than your system RAM. On top of that your system RAM has other demands on it such as the needs of the CPU, and there are other factors too. Long story short, if your GPU was using that RAM instead, your GPU performance would be limited to about the same as a GTX 730.

Whoah... I just googled difference b/w RAM and VRAM and ended up here.
I must say,
that
was
sexy.

And super interesting.

Dude you blow my mind!!!!i think you might be a teacher or something!!!!EXCELLENT EXPLANATION!!!!

 

[mah_1013812]

There would be a setting in BIOS but really you don't want to in honest. While in some situations you could make use of more than 2GB vRAM, the drastically slower speed of that RAM would be a massive bottleneck to performance if it was used.

Oh okay. Thanks. But won't the system automatically compensate for the loss of speed? I'm on windows 8.1 btw and my RAM is 2400Mhz (though there is an error where it writes wrong sometimes and creates corrupt data so i only run on 2200 Mhz which works fine) if it makes any difference. As far as i can see the GPU memory clock is running 1700-1800 mhz (during MSI Kombuster test + GPU-Z stats)

It only runs as fast as its slowest part. Much like having people driving 40 MPH slower than everyone else on the highway. You are comparing GDDR5 to DDR3/4. GDDR5 still runs much faster.