Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
9600M GT -512MB vs 256MB
- Thread starter thirtytwopaths
- Start date
- Sort by reaction score
I personally went with the 2.4 256mb vram model. I just think that it's a lot better value. I ordered 4gb of memory and a 500gb@7200rpm hdd for it and it still came cheaper than the 2.53 base model. I don't think laptops are for gaming anyways, that's why I have my beloved PS3.
500$ is a significant sum. You could justify upgrading to a new computer ~25% sooner, and I think having a new computer 1 year earlier or so is a far larger performance boost.
GTAIV will use as much as you can throw at it (Or at least, any card I've seen.) So much texture maps... ridiculous.
Who knows as as far as Snow Leopard, but hopefully with today's keynote you'll have some idea.
Who knows as as far as Snow Leopard, but hopefully with today's keynote you'll have some idea.
When D3 and SC2 comes out, ppl who own the 2.4ghz mbp will be moaning and groaning....just watch
When it comes to investing in big ticket items, most people neglect to calculate for the future.
But questions pertaining to the 2.4 or 2.5ghz version should be:
1. How much gaming will I be doing in the next 2-4 years
2. How much more $ can I get from it when I sell it
3. How much longer can I keep this machine as opposed to the 2.4ghz
Considering you're already spending 80% ($1999/$2499) of the cost of the lower-end MBP, that 20% isn't really that much more.
Buying the 2.4ghz mbp is kinda a middle-of-the-road decision. If you're going Pro, go Pro w/ the more $ one. Otherwise, the MB is a better investment. The only people I can see the lower-end mbp catering to, are those who want SOME ability to play newer games, but only to a small extent 2 years down the line.
When it comes to investing in big ticket items, most people neglect to calculate for the future.
But questions pertaining to the 2.4 or 2.5ghz version should be:
1. How much gaming will I be doing in the next 2-4 years
2. How much more $ can I get from it when I sell it
3. How much longer can I keep this machine as opposed to the 2.4ghz
Considering you're already spending 80% ($1999/$2499) of the cost of the lower-end MBP, that 20% isn't really that much more.
Buying the 2.4ghz mbp is kinda a middle-of-the-road decision. If you're going Pro, go Pro w/ the more $ one. Otherwise, the MB is a better investment. The only people I can see the lower-end mbp catering to, are those who want SOME ability to play newer games, but only to a small extent 2 years down the line.
Just don't think too much about it.
The way I use to calculate these kind of things is how long I'll be using it for. I think I use MBP on average for 4 hours per day everyday of the year for 3 years so that's 4300+ so it's only 2 dollars+ a day for using the laptop. Not so bad right?
The way I use to calculate these kind of things is how long I'll be using it for. I think I use MBP on average for 4 hours per day everyday of the year for 3 years so that's 4300+ so it's only 2 dollars+ a day for using the laptop. Not so bad right?
Shifting comparisons I think that is called. Check out Dan Gilbert, he puts new perspectives on things like that.
If you were buying a screen for $500 and the next model was $1000, you'd probably think i don't want to pay twice as much for the better model that i don't really need.
However in your situation, your saying, well I've already spend $1999, what's another $500.
That $500 doesn't know where it's being spent though. It's $500 regardless
Perhaps I should get the 17" now then
Indeed you cant even run as well as you would think with an overclocked 8600GT with 512mb of RAM. It is good for about half the max detail available.
If you are looking at gaming you will want to go to go as max'd as possible
You will notice it, irrespective of what anyone else says.
That's complete ********. Granted, coming from someone who misspelled "theoretically" more than twice in one explanation, I'm not surprised they got it so horribly wrong, but it's still awful.
I don't understand why I keep hearing this. It doesn't even make sense.
The card can address quite a bit more than 256 MB of RAM, and transfer more than 256 MB at a time. What the hell does the width of its memory bus have to do with anything? That has to do with throughput, not address space.
In instances where no more than 256 MB is used, obviously there will be no improvement. In instances where the card is so fillrate and/or bandwidth-bound, a larger frame buffer probably won't help much, even when called for. However, this explanation is nonsense.
Most computers being sold for the mainstream are still using 64-bit memory paths... higher-end are 128-bit. Does that mean you shouldn't have more than 256 MB of RAM? I think not.
i found this thread while searching google for 9600m gt reviews. seems to me that the issue of memory size/bandwidth/bus rate isn't too well understood by most people. So I'll give a shot to explain it.
#1. the most important thing regarding a video card's memory (also known as frame buffer) is bandwidth, or how much data can be transferred per second.
that being said, there are two determinants to bandwidth, memory speed and bus-width. with faster memory, the data moves through quicker. with wider bus, more data can be processed simultaneously.
in an analogy, consider a one road connecting point A and B. Point A and B are 1 mile apart. If I want to have a "bandwidth" of 1 car per minute, then the car has to travel 1 mile per minute. If I want to have a bandwidth of 2 cars per minute, then the car has to travel 2 miles per minute. This is the same thing regarding memory speeds.
now, there is another way to obtain a "bandwidth" of 2 cars per minute. Rather than make the cars move faster, I can add another road. Now even though the cars are moving at 1 mile per minute, because i have two roads, the rate is 2 cars per minute, because i added a road. This is the same thing as using a wider bus.
as one can see, there are multiple ways to obtain the same bandwidth, by altering the memory speed or, increasing the bus width. so far, so good?
now it gets tricky. Where memory size becomes important is the application of the frame buffer. video card memory has and needs very high bandwidth in comparison to regular RAM and your hard drive. For example, the 9600m gt's stock bandwidth is 25.6Gb/s. The new macbook pro's ram at 1066mhz gives 8.5Gb/s (hence PC8500 rating). To maximize your video card's performance, you want to keep everything in video memory. As soon as you overflow into regular RAM, performance will drop.
In an analogy, imagine you have a parking lot to a diner, with one space available. One car comes in every minute, one car leaves every minute. You're good to go. Now comes two cars at the same time. You only have space for one of them, but luckily, your friend "Computer RAM" can take the other car, but his parking lot is further away, so one guy is going to take longer to park his car so he can eat at the diner. In order to keep all the cars parked near the diner, you need more spaces. This is the same thing as increasing the memory size.
Finally, we reach what we wanted to know in the first place. How does memory size relate to bandwidth and thus video card performance? Bandwidth determines the amount of useful video card memory.
Back to our diner parking lot analogy, if there's only one car coming in per minute, and one car leaving per minute (bandwidth of 1car/min), does it make sense to make two parking spots? It doesn't. The second spot will go unused. I could add 1000 more spots, and my diner would still only have 1 customer per minute. This is why people commonly say the 9600m gt cannot use more than 256mb memory.
What might be misleading to those who don't know that much about hardware is the bandwidth of the video card. Come on! 25.6 gigabits per second? and it can't use more than 256 megabytes of memory? Although 25.6GB/s of bandwidth sounds high, it is actually quite low. Current generation desktop video cards have bandwidths of 141Gb/s (Nvidia's GTX 280). There's no point discussing here how exactly video cards compute the data, but the point is that 25.6GB/s is a extremely low bandwidth, which is why more powerful laptop gaming video cards like the 9800m gt don't just increase the core mhz and shader clock, but double the bus-width( to double the bandwidth), and add in more shaders to utilize the increased bandwidth.
this is why I bought the 256mb 2.4ghz unibody without much hesitation. the slight fps increases for the 2.53ghz version in the benchmarks from macworld that was posted are most likely due to the faster cpu with larger cache. Especially at a low resolution like 1280x800, that 512mb is extra useless.
hope this explains everything. and kudos to anyone that read through it all
-nemesis730 from ocforums
#1. the most important thing regarding a video card's memory (also known as frame buffer) is bandwidth, or how much data can be transferred per second.
that being said, there are two determinants to bandwidth, memory speed and bus-width. with faster memory, the data moves through quicker. with wider bus, more data can be processed simultaneously.
in an analogy, consider a one road connecting point A and B. Point A and B are 1 mile apart. If I want to have a "bandwidth" of 1 car per minute, then the car has to travel 1 mile per minute. If I want to have a bandwidth of 2 cars per minute, then the car has to travel 2 miles per minute. This is the same thing regarding memory speeds.
now, there is another way to obtain a "bandwidth" of 2 cars per minute. Rather than make the cars move faster, I can add another road. Now even though the cars are moving at 1 mile per minute, because i have two roads, the rate is 2 cars per minute, because i added a road. This is the same thing as using a wider bus.
as one can see, there are multiple ways to obtain the same bandwidth, by altering the memory speed or, increasing the bus width. so far, so good?
now it gets tricky. Where memory size becomes important is the application of the frame buffer. video card memory has and needs very high bandwidth in comparison to regular RAM and your hard drive. For example, the 9600m gt's stock bandwidth is 25.6Gb/s. The new macbook pro's ram at 1066mhz gives 8.5Gb/s (hence PC8500 rating). To maximize your video card's performance, you want to keep everything in video memory. As soon as you overflow into regular RAM, performance will drop.
In an analogy, imagine you have a parking lot to a diner, with one space available. One car comes in every minute, one car leaves every minute. You're good to go. Now comes two cars at the same time. You only have space for one of them, but luckily, your friend "Computer RAM" can take the other car, but his parking lot is further away, so one guy is going to take longer to park his car so he can eat at the diner. In order to keep all the cars parked near the diner, you need more spaces. This is the same thing as increasing the memory size.
Finally, we reach what we wanted to know in the first place. How does memory size relate to bandwidth and thus video card performance? Bandwidth determines the amount of useful video card memory.
Back to our diner parking lot analogy, if there's only one car coming in per minute, and one car leaving per minute (bandwidth of 1car/min), does it make sense to make two parking spots? It doesn't. The second spot will go unused. I could add 1000 more spots, and my diner would still only have 1 customer per minute. This is why people commonly say the 9600m gt cannot use more than 256mb memory.
What might be misleading to those who don't know that much about hardware is the bandwidth of the video card. Come on! 25.6 gigabits per second? and it can't use more than 256 megabytes of memory? Although 25.6GB/s of bandwidth sounds high, it is actually quite low. Current generation desktop video cards have bandwidths of 141Gb/s (Nvidia's GTX 280). There's no point discussing here how exactly video cards compute the data, but the point is that 25.6GB/s is a extremely low bandwidth, which is why more powerful laptop gaming video cards like the 9800m gt don't just increase the core mhz and shader clock, but double the bus-width( to double the bandwidth), and add in more shaders to utilize the increased bandwidth.
this is why I bought the 256mb 2.4ghz unibody without much hesitation. the slight fps increases for the 2.53ghz version in the benchmarks from macworld that was posted are most likely due to the faster cpu with larger cache. Especially at a low resolution like 1280x800, that 512mb is extra useless.
hope this explains everything. and kudos to anyone that read through it all
-nemesis730 from ocforums
i found this thread while searching google for 9600m gt reviews. seems to me that the issue of memory size/bandwidth/bus rate isn't too well understood by most people. So I'll give a shot to explain it.
#1. the most important thing regarding a video card's memory (also known as frame buffer) is bandwidth, or how much data can be transferred per second.
that being said, there are two determinants to bandwidth, memory speed and bus-width. with faster memory, the data moves through quicker. with wider bus, more data can be processed simultaneously.
in an analogy, consider a one road connecting point A and B. Point A and B are 1 mile apart. If I want to have a "bandwidth" of 1 car per minute, then the car has to travel 1 mile per minute. If I want to have a bandwidth of 2 cars per minute, then the car has to travel 2 miles per minute. This is the same thing regarding memory speeds.
now, there is another way to obtain a "bandwidth" of 2 cars per minute. Rather than make the cars move faster, I can add another road. Now even though the cars are moving at 1 mile per minute, because i have two roads, the rate is 2 cars per minute, because i added a road. This is the same thing as using a wider bus.
as one can see, there are multiple ways to obtain the same bandwidth, by altering the memory speed or, increasing the bus width. so far, so good?
now it gets tricky. Where memory size becomes important is the application of the frame buffer. video card memory has and needs very high bandwidth in comparison to regular RAM and your hard drive. For example, the 9600m gt's stock bandwidth is 25.6Gb/s. The new macbook pro's ram at 1066mhz gives 8.5Gb/s (hence PC8500 rating). To maximize your video card's performance, you want to keep everything in video memory. As soon as you overflow into regular RAM, performance will drop.
In an analogy, imagine you have a parking lot to a diner, with one space available. One car comes in every minute, one car leaves every minute. You're good to go. Now comes two cars at the same time. You only have space for one of them, but luckily, your friend "Computer RAM" can take the other car, but his parking lot is further away, so one guy is going to take longer to park his car so he can eat at the diner. In order to keep all the cars parked near the diner, you need more spaces. This is the same thing as increasing the memory size.
Finally, we reach what we wanted to know in the first place. How does memory size relate to bandwidth and thus video card performance? Bandwidth determines the amount of useful video card memory.
Back to our diner parking lot analogy, if there's only one car coming in per minute, and one car leaving per minute (bandwidth of 1car/min), does it make sense to make two parking spots? It doesn't. The second spot will go unused. I could add 1000 more spots, and my diner would still only have 1 customer per minute. This is why people commonly say the 9600m gt cannot use more than 256mb memory.
What might be misleading to those who don't know that much about hardware is the bandwidth of the video card. Come on! 25.6 gigabits per second? and it can't use more than 256 megabytes of memory? Although 25.6GB/s of bandwidth sounds high, it is actually quite low. Current generation desktop video cards have bandwidths of 141Gb/s (Nvidia's GTX 280). There's no point discussing here how exactly video cards compute the data, but the point is that 25.6GB/s is a extremely low bandwidth, which is why more powerful laptop gaming video cards like the 9800m gt don't just increase the core mhz and shader clock, but double the bus-width( to double the bandwidth), and add in more shaders to utilize the increased bandwidth.
this is why I bought the 256mb 2.4ghz unibody without much hesitation. the slight fps increases for the 2.53ghz version in the benchmarks from macworld that was posted are most likely due to the faster cpu with larger cache. Especially at a low resolution like 1280x800, that 512mb is extra useless.
hope this explains everything. and kudos to anyone that read through it all
-nemesis730 from ocforums
Very well written, good stuff. So I'm guessing if i run stuff on an external monitor with higher resolution (ACD LED), it might be worth it for the 512 config?
the extra memory is useful for running external monitors because the 25.6Gb/s bandwidth is more than enough for 2D desktop situations. The limiting factor becomes having enough video memory so no video data needs to go to slower RAM. That's why most people say that the 512mb video card feels "snappier" than the 256mb when running externals, especially at higher resolutions.
i found this thread while searching google for 9600m gt reviews. seems to me that the issue of memory size/bandwidth/bus rate isn't too well understood by most people. So I'll give a shot to explain it.
#1. the most important thing regarding a video card's memory (also known as frame buffer) is bandwidth, or how much data can be transferred per second.
that being said, there are two determinants to bandwidth, memory speed and bus-width. with faster memory, the data moves through quicker. with wider bus, more data can be processed simultaneously.
in an analogy, consider a one road connecting point A and B. Point A and B are 1 mile apart. If I want to have a "bandwidth" of 1 car per minute, then the car has to travel 1 mile per minute. If I want to have a bandwidth of 2 cars per minute, then the car has to travel 2 miles per minute. This is the same thing regarding memory speeds.
now, there is another way to obtain a "bandwidth" of 2 cars per minute. Rather than make the cars move faster, I can add another road. Now even though the cars are moving at 1 mile per minute, because i have two roads, the rate is 2 cars per minute, because i added a road. This is the same thing as using a wider bus.
as one can see, there are multiple ways to obtain the same bandwidth, by altering the memory speed or, increasing the bus width. so far, so good?
now it gets tricky. Where memory size becomes important is the application of the frame buffer. video card memory has and needs very high bandwidth in comparison to regular RAM and your hard drive. For example, the 9600m gt's stock bandwidth is 25.6Gb/s. The new macbook pro's ram at 1066mhz gives 8.5Gb/s (hence PC8500 rating). To maximize your video card's performance, you want to keep everything in video memory. As soon as you overflow into regular RAM, performance will drop.
In an analogy, imagine you have a parking lot to a diner, with one space available. One car comes in every minute, one car leaves every minute. You're good to go. Now comes two cars at the same time. You only have space for one of them, but luckily, your friend "Computer RAM" can take the other car, but his parking lot is further away, so one guy is going to take longer to park his car so he can eat at the diner. In order to keep all the cars parked near the diner, you need more spaces. This is the same thing as increasing the memory size.
Finally, we reach what we wanted to know in the first place. How does memory size relate to bandwidth and thus video card performance? Bandwidth determines the amount of useful video card memory.
Back to our diner parking lot analogy, if there's only one car coming in per minute, and one car leaving per minute (bandwidth of 1car/min), does it make sense to make two parking spots? It doesn't. The second spot will go unused. I could add 1000 more spots, and my diner would still only have 1 customer per minute. This is why people commonly say the 9600m gt cannot use more than 256mb memory.
What might be misleading to those who don't know that much about hardware is the bandwidth of the video card. Come on! 25.6 gigabits per second? and it can't use more than 256 megabytes of memory? Although 25.6GB/s of bandwidth sounds high, it is actually quite low. Current generation desktop video cards have bandwidths of 141Gb/s (Nvidia's GTX 280). There's no point discussing here how exactly video cards compute the data, but the point is that 25.6GB/s is a extremely low bandwidth, which is why more powerful laptop gaming video cards like the 9800m gt don't just increase the core mhz and shader clock, but double the bus-width( to double the bandwidth), and add in more shaders to utilize the increased bandwidth.
this is why I bought the 256mb 2.4ghz unibody without much hesitation. the slight fps increases for the 2.53ghz version in the benchmarks from macworld that was posted are most likely due to the faster cpu with larger cache. Especially at a low resolution like 1280x800, that 512mb is extra useless.
hope this explains everything. and kudos to anyone that read through it all
-nemesis730 from ocforums
So with the 25.6 GB, you can still utilize the 512 mb?
I'm confused... so the reason you went with the 2.4 is because the increases are not significant?
i found this thread while searching google for 9600m gt reviews. seems to me that the issue of memory size/bandwidth/bus rate isn't too well understood by most people. So I'll give a shot to explain it.
#1. the most important thing regarding a video card's memory (also known as frame buffer) is bandwidth, or how much data can be transferred per second.
that being said, there are two determinants to bandwidth, memory speed and bus-width. with faster memory, the data moves through quicker. with wider bus, more data can be processed simultaneously.
in an analogy, consider a one road connecting point A and B. Point A and B are 1 mile apart. If I want to have a "bandwidth" of 1 car per minute, then the car has to travel 1 mile per minute. If I want to have a bandwidth of 2 cars per minute, then the car has to travel 2 miles per minute. This is the same thing regarding memory speeds.
now, there is another way to obtain a "bandwidth" of 2 cars per minute. Rather than make the cars move faster, I can add another road. Now even though the cars are moving at 1 mile per minute, because i have two roads, the rate is 2 cars per minute, because i added a road. This is the same thing as using a wider bus.
as one can see, there are multiple ways to obtain the same bandwidth, by altering the memory speed or, increasing the bus width. so far, so good?
now it gets tricky. Where memory size becomes important is the application of the frame buffer. video card memory has and needs very high bandwidth in comparison to regular RAM and your hard drive. For example, the 9600m gt's stock bandwidth is 25.6Gb/s. The new macbook pro's ram at 1066mhz gives 8.5Gb/s (hence PC8500 rating). To maximize your video card's performance, you want to keep everything in video memory. As soon as you overflow into regular RAM, performance will drop.
In an analogy, imagine you have a parking lot to a diner, with one space available. One car comes in every minute, one car leaves every minute. You're good to go. Now comes two cars at the same time. You only have space for one of them, but luckily, your friend "Computer RAM" can take the other car, but his parking lot is further away, so one guy is going to take longer to park his car so he can eat at the diner. In order to keep all the cars parked near the diner, you need more spaces. This is the same thing as increasing the memory size.
Finally, we reach what we wanted to know in the first place. How does memory size relate to bandwidth and thus video card performance? Bandwidth determines the amount of useful video card memory.
Back to our diner parking lot analogy, if there's only one car coming in per minute, and one car leaving per minute (bandwidth of 1car/min), does it make sense to make two parking spots? It doesn't. The second spot will go unused. I could add 1000 more spots, and my diner would still only have 1 customer per minute. This is why people commonly say the 9600m gt cannot use more than 256mb memory.
What might be misleading to those who don't know that much about hardware is the bandwidth of the video card. Come on! 25.6 gigabits per second? and it can't use more than 256 megabytes of memory? Although 25.6GB/s of bandwidth sounds high, it is actually quite low. Current generation desktop video cards have bandwidths of 141Gb/s (Nvidia's GTX 280). There's no point discussing here how exactly video cards compute the data, but the point is that 25.6GB/s is a extremely low bandwidth, which is why more powerful laptop gaming video cards like the 9800m gt don't just increase the core mhz and shader clock, but double the bus-width( to double the bandwidth), and add in more shaders to utilize the increased bandwidth.
this is why I bought the 256mb 2.4ghz unibody without much hesitation. the slight fps increases for the 2.53ghz version in the benchmarks from macworld that was posted are most likely due to the faster cpu with larger cache. Especially at a low resolution like 1280x800, that 512mb is extra useless.
hope this explains everything. and kudos to anyone that read through it all
-nemesis730 from ocforums
Wow! Sticky please
This thing will stop many threads like these (which there are already too many of them). I have seen some explanations before, but this is the first one I actually understood! the short answer is 512mb can be used for 2D (like external monitors), not for 3D. Going back to my diner parking lot example, i'll try to explain why.
Let's start by simplifying the 512mb memory to 50 car spaces and the 25.6Gb/s bandwidth means you can take 25 cars per minute in, and 25 cars per minute out. With gaming, every individual pixel changes per second, there are physics and motion calculation, etc. Because there is new data to be calculated, the old data must be discarded. In the analogy, old customers must leave to make space for new ones. So the reason why 25 cars leave per minute is because 25 new cars are about to enter.
"Wait a minute!", you say. instead of being discarded, can't you let them stay if you have more memory? Meaning, that the extra 25 spaces can be used to let the old data "park" instead of forcing them to leave?
While this is true, think about the usefulness of this. These 25 customers have already entered, eaten at the diner, and paid. Do you make more money (ie gain performance) if you let them park their car longer after they've paid? No. After eating at the diner (data processing), there is no gain to let them park longer. Or think of it this way- after your graphics card calculated all the data to show 1 frame of you defeating the boss at the end of the fire-world level, is it useful to hold on to that data when you move on to the ice-world level?
With 2D, or running an extra monitor, there isn't much data movement, because most of the data is the same. Doesn't require massive bandwidth to show the same black background pixel over and over again. In this analogy, there are no new customers, just the same ones over and over. Like 3D, you could kick them out and have them come back in (refreshing the same background repeatedly), or you can let them stay (just show the background constantly). So here, there is a gain to letting them park longer. You save yourself the time it takes them to leave the diner, leave through the exit, and circle back around to the parking lot entrance. And the more memory you have, the more spaces you can give to the same customers so they can stay. This is why people say 512mb of video memory makes running external monitors "snappier".
Finally, i got the 2.4ghz because the gain in gaming performance with the 2.53ghz version was not due to more memory on the video card. It's due to the faster cpu. Would it make sense to believe that a ferarri is faster than a civic, not because its engine has more horsepower, but because it can hold 20 gallons of gas instead of 10?
Therefore 512MB wouldn't help much with 3D processing?
Awww that's very disappointing to know. But at the same time it's good to know that my 2.93GHz will be utilized for 3D at some stage.
Awww that's very disappointing to know.
But at the same time it's good to know that my 2.93GHz will be utilized for 3D at some stage.I'm sorry, but I have both the 2.4 and the 2.53 at the current moment and they both are about the same heat and battery. so......
I made a bad purchase decision also. I went to 2.53 for addition 256 MB on the memory, now that I read it is useless - the only thing useful from $500 upgrade is the marginal CPU and 4GB memory.
I just got a 4GB memory from OWC for $67 and the price difference between 2.53 and 2.8 17" is marginal... should either go to 2.4 or 17".
I just got a 4GB memory from OWC for $67 and the price difference between 2.53 and 2.8 17" is marginal... should either go to 2.4 or 17".
Due to what this truth says, the only question is, would you get the 2.53GHz model based on the L2 cahce increase and already pumped up RAM?
After all its a $500 difference for 3MB L2 cache many average users won't use. Not neglect the memory increase as you can get DDR3 for your MB(P) cheaply for less than $100. That means you are basically paying ~$400 for 3MB L2 cache.
I like this very well laid out explanation. Kudos to you sir and +2 internets...
there will be no noticeable difference UNLESS the game/app you are using tries to use more than 256/512 mb of graphics textures. So..if your playing WoW..you'll prob get the same FPS. If you're playing crysis..there might be a difference. Once you turn down the textures to 256 levels..the performance would be similar.