GeForce GTX 750 Ti Review: Maxwell Adds Performance Using Less Power

[Lascar]

Also,the reviewer goes on about the R7 265 being a 150W GPU and the GTX750TI being a 60W one??However,it seems he has not read his own article:http://media.bestofmicro.com/3/3/422607/original/06-Power-Consumption-Gaming.pngThe R9 270X and HD7870 are consuming under 130W,and the R7 265 is likely to consume less power if the HD7850 is anything to go by.Yes,the GTX750TI has decent performance for a bus powered card,but the reviewer seems to be overstating the power consumption of the R7 265 and understating its performance at the same time. Why??

Because those are the rated TDP's.

Also I have no idea how you come away from this thinking the 750 Ti merely has "decent" performance per W. The performance it offers at its rated TDP is unprecedented, all the more impressive considering it's still 28nm. You usually expect this sort of efficiency improvement out of a die shrink.

Well again i cannot stress this more. "WAIT"!!!! The technology has barely been implemented. Soon like i said AMD will choke Nvidia as they always do with a better and less costly card. I bet you 100 bucks they will do that 2 months from the 750Ti and sell the card 120 something. AMD Always wins the price performance war no two ways around it.

This response is so opinionated, and so unrelated to your previous comment, or my comment for that matter (which I'm assuming you're responding to) that I'm not even sure what to make of it.

I guess you could argue in a very general generational sense that AMD tends to offer the better value, speaking strictly in terms of price to performance ratios. But to say that AMD "always" wins can be so easily countered that I'm not sure why even the most adamant of AMD fanboy's would go through the trouble, and resulting humiliation of making such a claim, especially now.

When arguing with fanboys...

Sorry dude u are the fanboy raging over the obvious, AMD always wins the price performance competition. Simply because nvidia sticks to crazy prices that can never be justified. You own nvidia good for you, you paid twice as much as AMD would give you, good for you. You have money to waste good for you. If nvidia beat AMD on price performance i would totally buy but it never happens and it never will.

 

[Lascar]

I have a really old PC and right now a full upgrade is out of the question.Would this card work with my ASUS M2A-VM motherboard? I have a 9800 GT right now.

Yes friend it would, since the mobo has the PCI-E x16 to power it.

 

[Lascar]

Might have to get one of these in the future to go with my undervolted & underclocked 0.752 V E7200 still using IGP with it finally a low wattage GPU that really performs.

As i said few times above, do not rush. This is new tech that is being sold crazy expensive seriously. Wait 2 months and see new lineups with better prices appear.

 

[dragonsqrrl]

Also,the reviewer goes on about the R7 265 being a 150W GPU and the GTX750TI being a 60W one??However,it seems he has not read his own article:http://media.bestofmicro.com/3/3/422607/original/06-Power-Consumption-Gaming.pngThe R9 270X and HD7870 are consuming under 130W,and the R7 265 is likely to consume less power if the HD7850 is anything to go by.Yes,the GTX750TI has decent performance for a bus powered card,but the reviewer seems to be overstating the power consumption of the R7 265 and understating its performance at the same time. Why??

Because those are the rated TDP's.

Also I have no idea how you come away from this thinking the 750 Ti merely has "decent" performance per W. The performance it offers at its rated TDP is unprecedented, all the more impressive considering it's still 28nm. You usually expect this sort of efficiency improvement out of a die shrink.

Well again i cannot stress this more. "WAIT"!!!! The technology has barely been implemented. Soon like i said AMD will choke Nvidia as they always do with a better and less costly card. I bet you 100 bucks they will do that 2 months from the 750Ti and sell the card 120 something. AMD Always wins the price performance war no two ways around it.

This response is so opinionated, and so unrelated to your previous comment, or my comment for that matter (which I'm assuming you're responding to) that I'm not even sure what to make of it.

I guess you could argue in a very general generational sense that AMD tends to offer the better value, speaking strictly in terms of price to performance ratios. But to say that AMD "always" wins can be so easily countered that I'm not sure why even the most adamant of AMD fanboy's would go through the trouble, and resulting humiliation of making such a claim, especially now.

When arguing with fanboys...

AMD always wins the price performance competition... If nvidia beat AMD on price performance i would totally buy but it never happens and it never will.

Umm, not sure if you got this earlier (I guess I was being a little vague) but it's sort of happening right now. It's also happened many times in the past...

I'm also ready to get back on the original topic, whenever you're done trolling.

 

[Lascar]

Also,the reviewer goes on about the R7 265 being a 150W GPU and the GTX750TI being a 60W one??However,it seems he has not read his own article:http://media.bestofmicro.com/3/3/422607/original/06-Power-Consumption-Gaming.pngThe R9 270X and HD7870 are consuming under 130W,and the R7 265 is likely to consume less power if the HD7850 is anything to go by.Yes,the GTX750TI has decent performance for a bus powered card,but the reviewer seems to be overstating the power consumption of the R7 265 and understating its performance at the same time. Why??

Because those are the rated TDP's.

Also I have no idea how you come away from this thinking the 750 Ti merely has "decent" performance per W. The performance it offers at its rated TDP is unprecedented, all the more impressive considering it's still 28nm. You usually expect this sort of efficiency improvement out of a die shrink.

Well again i cannot stress this more. "WAIT"!!!! The technology has barely been implemented. Soon like i said AMD will choke Nvidia as they always do with a better and less costly card. I bet you 100 bucks they will do that 2 months from the 750Ti and sell the card 120 something. AMD Always wins the price performance war no two ways around it.

This response is so opinionated, and so unrelated to your previous comment, or my comment for that matter (which I'm assuming you're responding to) that I'm not even sure what to make of it.

I guess you could argue in a very general generational sense that AMD tends to offer the better value, speaking strictly in terms of price to performance ratios. But to say that AMD "always" wins can be so easily countered that I'm not sure why even the most adamant of AMD fanboy's would go through the trouble, and resulting humiliation of making such a claim, especially now.

When arguing with fanboys...

AMD always wins the price performance competition... If nvidia beat AMD on price performance i would totally buy but it never happens and it never will.

Umm, not sure if you got this earlier (I guess I was being a little vague) but it's sort of happening right now. It's also happened many times in the past...

I'm also ready to get back on the original topic, whenever you're done trolling.

Huh! nevermind the childish retort then, If for you price performance is not a factor when dealing with graphic cards then by all means suit yourself. I do not wish to discuss price performance matter with someone who owns a gtx 480 and sees only performance.

 

[smeezekitty]

Might have to get one of these in the future to go with my undervolted & underclocked 0.752 V E7200 still using IGP with it finally a low wattage GPU that really performs.

That would be a waste. A STOCK E7200 will bottleneck it in some cases. Underclocked that far would be totally pointless

 

[Lascar]

Might have to get one of these in the future to go with my undervolted & underclocked 0.752 V E7200 still using IGP with it finally a low wattage GPU that really performs.

That would be a waste. A STOCK E7200 will bottleneck it in some cases. Underclocked that far would be totally pointless

As in Nvidia architectures prior, ROP partitions and L2 cache slices are aligned. And like the GeForce GTX 650 Ti’s GK106 processor, GM107 sports two partitions with eight units each, giving you up to 16 32-bit integer pixels per clock. Where the two GPUs really diverge is their L2 cache capacity. In GK106, you were looking at 128 KB per slice, adding up to 256 KB in an implementation with two ROP partitions. GM107 appears to wield 1 MB per slice, yielding 2 MB of memory used for servicing load, store, and texture requests. According to Nvidia, this translates to a significant load shifted away from the external memory system, along with notable power savings.

Going easy on memory bandwidth is smart, since GM107 exposes a pair of 64-bit memory controllers to which 1 or 2 GB of 1350 MHz GDDR5 DRAM is attached. Peak throughput is, interestingly, exactly what we got from GeForce GTX 650 Ti: 86.4 GB/s. The memory is feeding fewer CUDA cores, but they’re managed more efficiently. So, the big L2 is supposed to play an instrumental role in preventing a bottleneck.

 

[average joe]

So in AMD terms this thing performs like a 7790 with the power draw of a 7750 without a die shrink. That's impressive. AMD is screwed, they quit making FX processors and an I3 with this in a laptop is going to rape the Kaveri.

 

[Mousemonkey]

So in AMD terms this thing performs like a 7790 with the power draw of a 7750 without a die shrink. That's impressive. AMD is screwed, they quit making FX processors and an I3 with this in a laptop is going to rape the Kaveri.

I have a 7790 and am considering getting one of these to see if it can fold and run other software at the same time.

 

[average joe]

I just bought an xfx 7870 which dropped down under 200 briefly this week. There's a mail in rebate too but I never expect those to pan out. it's stock clock but has the dual fans i doubt ill get much OC over it these must be the rejects from the OC version... I haven't had an nvidia card since the 8800gt. I was really planning on the 760 but I needed a PSU for the AMD or Nvidia and I could get the 7870 and PSU briefly for the cost of the 760.

 

[jrlawson]

This make me wonder when we can expect higher end maxwell cards (i.e. 800 series)

 

[average joe]

I think the higher end cards will come out after the die shrink to 20nm. Right now this chip is crippled by the memory bus and a locked multiplier. I appear like it is going to overclock like the core 2 intel chips did a few years ago. Anandtech said they could over clock every card they had for review up to max voltage and multiplier without even reaching thermal limits where throttling would slow the card down. Well imagine 6 ghz ram and a 192 bit bus or 256. These things are going to be amazing. During the benchmarks they never topped 60 Celsius and throttling starts at 80.

 

[average joe]

I was disappointed in the Kaveri and upset that AMD dumped the FX series to make APU's exclusively. But I love me 7770 which played pretty good at 1080 and they always had the better price. What Maxwell seems to be at first glance is a major game changer. Like the Core 2 Duo on the processor side. This thing is better performance per watt than anything AMD has to offer. So they are being squeezed from both sides. The CPU side is abysmal vs Intel. AMD shugs it off because of the IGPU advantage which is not growing but shrinking with Intel. AMD then claims its focus on portables and tablets not desktops. The Samsung and Nvidia Tegra 4 in the tablet space it clobbering them. AMD shrugs this off because they have the GPU market value play. Well unless they start selling R9 270's for $149.99 instead of 239.99 they aren't that good a deal anymore. I don't know what the CEO is doing but you can't survive turning out 2 tier product in every category you compete in. This thing and an i3 in a tablet would destroy an A10 and the battery would last twice as long.

 

[dragonsqrrl]

Also,the reviewer goes on about the R7 265 being a 150W GPU and the GTX750TI being a 60W one??However,it seems he has not read his own article:http://media.bestofmicro.com/3/3/422607/original/06-Power-Consumption-Gaming.pngThe R9 270X and HD7870 are consuming under 130W,and the R7 265 is likely to consume less power if the HD7850 is anything to go by.Yes,the GTX750TI has decent performance for a bus powered card,but the reviewer seems to be overstating the power consumption of the R7 265 and understating its performance at the same time. Why??

Because those are the rated TDP's.

Also I have no idea how you come away from this thinking the 750 Ti merely has "decent" performance per W. The performance it offers at its rated TDP is unprecedented, all the more impressive considering it's still 28nm. You usually expect this sort of efficiency improvement out of a die shrink.

Well again i cannot stress this more. "WAIT"!!!! The technology has barely been implemented. Soon like i said AMD will choke Nvidia as they always do with a better and less costly card. I bet you 100 bucks they will do that 2 months from the 750Ti and sell the card 120 something. AMD Always wins the price performance war no two ways around it.

This response is so opinionated, and so unrelated to your previous comment, or my comment for that matter (which I'm assuming you're responding to) that I'm not even sure what to make of it.

I guess you could argue in a very general generational sense that AMD tends to offer the better value, speaking strictly in terms of price to performance ratios. But to say that AMD "always" wins can be so easily countered that I'm not sure why even the most adamant of AMD fanboy's would go through the trouble, and resulting humiliation of making such a claim, especially now.

When arguing with fanboys...

AMD always wins the price performance competition... If nvidia beat AMD on price performance i would totally buy but it never happens and it never will.

Umm, not sure if you got this earlier (I guess I was being a little vague) but it's sort of happening right now. It's also happened many times in the past...

I'm also ready to get back on the original topic, whenever you're done trolling.

Huh! nevermind the childish retort then, If for you price performance is not a factor when dealing with graphic cards then by all means suit yourself. I do not wish to discuss price performance matter with someone who owns a gtx 480 and sees only performance.

I guess that's it then...

Although I will say a few things. You realize you diverted to the whole price performance discussion, right?

You seem to be under the assumption that I bought that GTX480 at or around MSRP. Not sure why you would assume that. But even if I had, I'm not sure why you think I paid twice as much for it compared to what AMD could give me...

Also I apologize for the tone of my previous comment, but you seemed to be unaware of (or indifferent to) the fact that you were contradicting yourself. But looking at it from your perspective I guess that wouldn't matter much to me either.

You really should be careful about making blanket statements, like assuming I only value performance simply because I own a 480, or asserting that AMD is always the better value, without actually knowing how much I paid for my card or having any way of backing up what you're saying.

 

[smeezekitty]

Might have to get one of these in the future to go with my undervolted & underclocked 0.752 V E7200 still using IGP with it finally a low wattage GPU that really performs.

That would be a waste. A STOCK E7200 will bottleneck it in some cases. Underclocked that far would be totally pointless

As in Nvidia architectures prior, ROP partitions and L2 cache slices are aligned. And like the GeForce GTX 650 Ti’s GK106 processor, GM107 sports two partitions with eight units each, giving you up to 16 32-bit integer pixels per clock. Where the two GPUs really diverge is their L2 cache capacity. In GK106, you were looking at 128 KB per slice, adding up to 256 KB in an implementation with two ROP partitions. GM107 appears to wield 1 MB per slice, yielding 2 MB of memory used for servicing load, store, and texture requests. According to Nvidia, this translates to a significant load shifted away from the external memory system, along with notable power savings.

Going easy on memory bandwidth is smart, since GM107 exposes a pair of 64-bit memory controllers to which 1 or 2 GB of 1350 MHz GDDR5 DRAM is attached. Peak throughput is, interestingly, exactly what we got from GeForce GTX 650 Ti: 86.4 GB/s. The memory is feeding fewer CUDA cores, but they’re managed more efficiently. So, the big L2 is supposed to play an instrumental role in preventing a bottleneck.

That is just all fine and dandy but that doesn't change the fact a 1 GHz CPU doesn't have chance with keeping up with even a low end card

 

[jin_mtvt]

dragonsqrrl: and by half of amd's offering you mean 2 cards right ? The only real new gpus of the last few months are the 290's ...this is still same old stuff reworked fore better efficiency etc.. not much different from a rebadged + tweaked product to fill up a current market . Power consumption is only good on mobile situation. Again, i do not see anything bad about this new offering, lower consumption has its place and it needs to be implemented a some point, but nothing much revolutionary here. I hope green has more surprises in the coming months. Meanwhile on my 250W 290 in BF4

 

[Lascar]

Again still people are not even taking time to look in front at all i mean seriously at all. Stop comparing new tech to rebrands. AMD is fucked? where? They just have to implement new tech all the same. AMD always does this, wait for nvidia to release something good and expensive, then step up their tech and release at lower price

I was disappointed in the Kaveri and upset that AMD dumped the FX series to make APU's exclusively. But I love me 7770 which played pretty good at 1080 and they always had the better price. What Maxwell seems to be at first glance is a major game changer. Like the Core 2 Duo on the processor side. This thing is better performance per watt than anything AMD has to offer. So they are being squeezed from both sides. The CPU side is abysmal vs Intel. AMD shugs it off because of the IGPU advantage which is not growing but shrinking with Intel. AMD then claims its focus on portables and tablets not desktops. The Samsung and Nvidia Tegra 4 in the tablet space it clobbering them. AMD shrugs this off because they have the GPU market value play. Well unless they start selling R9 270's for $149.99 instead of 239.99 they aren't that good a deal anymore. I don't know what the CEO is doing but you can't survive turning out 2 tier product in every category you compete in. This thing and an i3 in a tablet would destroy an A10 and the battery would last twice as long.

About that, kaveri is new APU update. But if u followed the FM2+ path, u would notice that they discontinued AM3+ to make all future chips FM2+ not completely dissolving the CPU only chip. They have not yet entered production phase probably still planning new architecture

 

[Kingpin007]

Woah.. this card is going to be a best seller i know atleast 30 people who would buy this right away....

 

[HideOut]

I didnt notice. SLI capable? less than 150 and sli scaled. Could be interesting...

 

[smeezekitty]

I didnt notice. SLI capable? less than 150 and sli scaled. Could be interesting...

No SLI capability. Not surprising from NV

 

[Lascar]

Might have to get one of these in the future to go with my undervolted & underclocked 0.752 V E7200 still using IGP with it finally a low wattage GPU that really performs.

That would be a waste. A STOCK E7200 will bottleneck it in some cases. Underclocked that far would be totally pointless

As in Nvidia architectures prior, ROP partitions and L2 cache slices are aligned. And like the GeForce GTX 650 Ti’s GK106 processor, GM107 sports two partitions with eight units each, giving you up to 16 32-bit integer pixels per clock. Where the two GPUs really diverge is their L2 cache capacity. In GK106, you were looking at 128 KB per slice, adding up to 256 KB in an implementation with two ROP partitions. GM107 appears to wield 1 MB per slice, yielding 2 MB of memory used for servicing load, store, and texture requests. According to Nvidia, this translates to a significant load shifted away from the external memory system, along with notable power savings.

Going easy on memory bandwidth is smart, since GM107 exposes a pair of 64-bit memory controllers to which 1 or 2 GB of 1350 MHz GDDR5 DRAM is attached. Peak throughput is, interestingly, exactly what we got from GeForce GTX 650 Ti: 86.4 GB/s. The memory is feeding fewer CUDA cores, but they’re managed more efficiently. So, the big L2 is supposed to play an instrumental role in preventing a bottleneck.

That is just all fine and dandy but that doesn't change the fact a 1 GHz CPU doesn't have chance with keeping up with even a low end card

wait what!!!! 1ghz, dude core 2 DUO E7200 is not 1GHZ. Go look at specs before saying those random stuff.

 

[fbbam]

Imagine this kind of efficiency in a laptop

 

[dragonsqrrl]

dragonsqrrl: and by half of amd's offering you mean 2 cards right ? The only real new gpus of the last few months are the 290's ...this is still same old stuff reworked fore better efficiency etc.. not much different from a rebadged + tweaked product to fill up a current market . Power consumption is only good on mobile situation. Again, i do not see anything bad about this new offering, lower consumption has its place and it needs to be implemented a some point, but nothing much revolutionary here. I hope green has more surprises in the coming months. Meanwhile on my 250W 290 in BF4

That's not a bad place to be with Mantle.

I would hardly call GM107 "not much different from a rebadged". While I don't have a problem with rebadges if it makes sense (price/performance fits into the new lineup), how different would it have been if something like a slightly tweaked 650 Ti Boost came to market as the 750 Ti at $150?

You're right, I was exaggerating about AMD's overcrowded sub $200 lineup, but it was only to make a joke about AMD cannibalizing their own products, which they seem to have gotten quite good at lately.

 

[Someone Somewhere]

Might have to get one of these in the future to go with my undervolted & underclocked 0.752 V E7200 still using IGP with it finally a low wattage GPU that really performs.

That would be a waste. A STOCK E7200 will bottleneck it in some cases. Underclocked that far would be totally pointless

As in Nvidia architectures prior, ROP partitions and L2 cache slices are aligned. And like the GeForce GTX 650 Ti’s GK106 processor, GM107 sports two partitions with eight units each, giving you up to 16 32-bit integer pixels per clock. Where the two GPUs really diverge is their L2 cache capacity. In GK106, you were looking at 128 KB per slice, adding up to 256 KB in an implementation with two ROP partitions. GM107 appears to wield 1 MB per slice, yielding 2 MB of memory used for servicing load, store, and texture requests. According to Nvidia, this translates to a significant load shifted away from the external memory system, along with notable power savings.

Going easy on memory bandwidth is smart, since GM107 exposes a pair of 64-bit memory controllers to which 1 or 2 GB of 1350 MHz GDDR5 DRAM is attached. Peak throughput is, interestingly, exactly what we got from GeForce GTX 650 Ti: 86.4 GB/s. The memory is feeding fewer CUDA cores, but they’re managed more efficiently. So, the big L2 is supposed to play an instrumental role in preventing a bottleneck.

That is just all fine and dandy but that doesn't change the fact a 1 GHz CPU doesn't have chance with keeping up with even a low end card

wait what!!!! 1ghz, dude core 2 DUO E7200 is not 1GHZ. Go look at specs before saying those random stuff.

Note where it says underclocked and undervolted, and the CPU-Z validator saying 995MHz.

 

[Lascar]

Might have to get one of these in the future to go with my undervolted & underclocked 0.752 V E7200 still using IGP with it finally a low wattage GPU that really performs.

That would be a waste. A STOCK E7200 will bottleneck it in some cases. Underclocked that far would be totally pointless

As in Nvidia architectures prior, ROP partitions and L2 cache slices are aligned. And like the GeForce GTX 650 Ti’s GK106 processor, GM107 sports two partitions with eight units each, giving you up to 16 32-bit integer pixels per clock. Where the two GPUs really diverge is their L2 cache capacity. In GK106, you were looking at 128 KB per slice, adding up to 256 KB in an implementation with two ROP partitions. GM107 appears to wield 1 MB per slice, yielding 2 MB of memory used for servicing load, store, and texture requests. According to Nvidia, this translates to a significant load shifted away from the external memory system, along with notable power savings.

Going easy on memory bandwidth is smart, since GM107 exposes a pair of 64-bit memory controllers to which 1 or 2 GB of 1350 MHz GDDR5 DRAM is attached. Peak throughput is, interestingly, exactly what we got from GeForce GTX 650 Ti: 86.4 GB/s. The memory is feeding fewer CUDA cores, but they’re managed more efficiently. So, the big L2 is supposed to play an instrumental role in preventing a bottleneck.

That is just all fine and dandy but that doesn't change the fact a 1 GHz CPU doesn't have chance with keeping up with even a low end card

wait what!!!! 1ghz, dude core 2 DUO E7200 is not 1GHZ. Go look at specs before saying those random stuff.

Note where it says underclocked and undervolted, and the CPU-Z validator saying 995MHz.

http://www.hardwaresecrets.com/fullimage.php?image=10725
According to benchmarks this CPU does not bottleneck a HD 7770 i don't see why it would bottleneck this card. But i'd rather let that up to CPU experts. But i am pretty confident in what i was saying, it will most certainly not bottleneck the 750Ti. Well in his case he is doing the undervolting and underclocking, however its his own setup, talking about a normal setup there should be no problem. But btw why you undervolt and underclock ur cpu?