Upcoming GTX 750 Ti tested

[SiliconWars]

Yup - tbh Techpowerup makes a lot of these errors. They bench too many games at too many settings (automatically) and don't double check for errors. Clearly something has gone wrong when a 7770 outperforms the 260X by such a huge margin - but all that is seen is the end totals.

I like Techpowerup but their results warrant very close examination, especially when they just don't add up.

As for that Toms graph, 38% efficiency over the 260X seems about right based on what I suggested earlier with 25%-30% efficiency over the 7790.

Honestly it's a good card with amazing performance for the power draw, but AMD will likely match or exceed this at 28nm if they try to. It might take a few months, or maybe they are just waiting on 20nm (seems likely), but the 60%+ perf/W claims based on techpowerup's flawed benchmarks are completely bogus.

 

[scubadunc]

Sorry for the dumb question...but if I were to place the 750 ti into a PCI-E 4x slot, will it be highly bottle-necked?

My mobo is an old Gigabyte GA-EG45M-DS2H that I use as a HTPC

 

[Roland00Address]

Sorry for the dumb question...but if I were to place the 750 ti into a PCI-E 4x slot, will it be highly bottle-necked?

My mobo is an old Gigabyte GA-EG45M-DS2H that I use as a HTPC

Your 775 cpu would be as much a bottleneck as your 1st gen pci express 4x slot

 

[OCGuy]

What do you play, what is your resolution, what CPU?

 

[el etro]

As for that Toms graph, 38% efficiency over the 260X seems about right based on what I suggested earlier with 25%-30% efficiency over the 7790.

That's because the techpowerup error in test.

Honestly it's a good card with amazing performance for the power draw, but AMD will likely match or exceed this at 28nm if they try to. It might take a few months, or maybe they are just waiting on 20nm (seems likely), but the 60%+ perf/W claims based on techpowerup's flawed benchmarks are completely bogus.

IMO Amd don't want to put other architecture to fight Maxwell... until Maxwell don't show in the High-End or 20nm won't be ready, AMD's new arch will be not shown. HD 7790/290x is not Maxwell competitor.

 

[ams23]

Erm, the "whole system" remains the same except the graphics card. That means the difference in the graphics cards is....?

You also realise that Bonaire is only 8% (12mm2) bigger right? And it has the DSP's for TrueAudio taking up at least 10mm2?

Nvidia basically has a chip that is 25%-30% more efficient - their next gen chip that was built for this purpose vs a year-old AMD chip that was never even designed for efficiency in the first place.

Use some common sense. 750 Ti is ~ 20% faster than 7790 in Crysis 3, while the 7790 consumes ~ 20% more power at load in the game (ie. 12 watts more GPU power consumption). If you do the math, the 750 Ti has nearly 50% better (!) perf. per watt in this game, on a 28nm HP fab. process too! And this is just load power in one game where 750 Ti is bandwidth-limited.

Tom's Hardware did an analysis of avg. power consumption and performance across multiple games using 750 Ti, and the results were insanely good:

And here is a look at GPU power consumption in Metro LL:

The differences in power consumed and power efficiency is just staggering.

 

[OCGuy]

The overclocking ability is amazing. 1400mhz seems easily attainable.

 

[Homeles]

I don't really like Nvidia, but Maxwell looks like a home run for them. With the delay of Broadwell especially, they should be very good on discrete graphics shipments this year.

AMD really needs to go "mobile first" with graphics as well.

 

[Mondozei]

AMD really needs to go "mobile first" with graphics as well.

Yeah. They probably are already for their next architecture since Hawaii takes up a lot of power.

I don't really like Nvidia, but Maxwell looks like a home run for them. With the delay of Broadwell especially, they should be very good on discrete graphics shipments this year.

I'm neutral towards Nvidia as a company, although their pricing and greed in general(Linus Torvalds called them the single worst company he has ever worked with, so there's that) hasn't helped their reputation.

That being said, I doubt that R9-390X will be able to beast GTX 880 if they both come out Q4 of this year. Nevertheless, in Sweden at least, 8 out of the 10 top-selling cards are AMD cards. Last year, the ratio was reversed. Probably in large part because while many buy them for mining, the etailers here know that if they jacked up the prices like Newegg, they would be blown away by other competitors(there are about 6-7 major etailers nationwide, so that kind of terrorbalance prevents massive pricing increases).

I'm not sure how it looks in the rest of the world, but miners alone probably drive a lot of GPU for AMD. And overall GPU sales are still much larger outside of North America in total so my guess is that AMD probably is going to have a much better year this year than they would of had otherwise if we'd gotten 20 nm ready already by this March.

Once GTX 880 is out, I don't think we'll see anything from AMD that can beat it until they get a new architecture, so that means that 2015 will be the year of Nvidia if 2014 will be the year of AMD in the GPU market.

 

[OCGuy]

Yeah. They probably are already for their next architecture since Hawaii takes up a lot of power.

Hawaii was supposed to be 20nm, it will be interesting to see what they do in 2014, because a Hawaii shrink doesn't look like it will be enough to compete with GM2xx.

 

[psolord]

GTX 750 Ti reviews

http://www.tomshardware.com/reviews/geforce-gtx-750-ti-review,3750.html

http://www.techpowerup.com/reviews/MSI/GTX_750_Ti_Gaming/

http://www.guru3d.com/articles_pages/nvidia_geforce_gtx_750_and_750_ti_review,1.html

http://www.bit-tech.net/hardware/graphics/2014/02/18/nvidia-geforce-gtx-750-ti-review/1

http://www.anandtech.com/show/7764/the-nvidia-geforce-gtx-750-ti-and-gtx-750-review-maxwell

http://www.pcper.com/reviews/Graphi...750-Ti-Review-Maxwell-Architecture-debuts-150

http://www.techspot.com/review/783-geforce-gtx-750-ti-vs-radeon-r7-265/

http://www.gamespot.com/articles/gtx-750-ti-review-minimal-power-maximum-performance/1100-6417810/

http://hexus.net/tech/reviews/graphics/66145-nvidia-geforce-gtx-750-ti-28nm-maxwell/

http://www.legitreviews.com/nvidia-geforce-gtx-750-ti-2gb-video-card-review_135752

http://hothardware.com/Reviews/NVIDIA-GeForce-GTX-750-Ti-Maxwell-GPU-Review/

http://www.hardwarecanucks.com/foru.../65440-gtx-750-ti-review-maxwell-arrives.html

http://benchmarkreviews.com/12744/nvidia-geforce-gtx-750-ti-benchmark-performance/

http://www.tweaktown.com/reviews/60...b-reference-card-video-card-review/index.html

http://www.hardwareheaven.com/revie...dition-graphics-card-review-introduction.html

http://www.pcgameshardware.de/Nvidi...-255598/Videos/GTX-750-Ti-Test-Video-1109873/

http://www.computerbase.de/artikel/grafikkarten/2014/nvidia-geforce-gtx-750-ti-maxwell-im-test/

http://www.hardware.fr/news/13568/nvidia-lance-geforce-gtx-750-ti-750-maxwell.html

 

[SPBHM]

comparing this card with the GTX 480 is quite funny
http://www.anandtech.com/bench/product/1135?vs=1130

 

[SiliconWars]

Use some common sense. 750 Ti is ~ 20% faster than 7790 in Crysis 3, while the 7790 consumes ~ 20% more power at load in the game (ie. 12 watts more GPU power consumption). If you do the math, the 750 Ti has nearly 50% better (!) perf. per watt in this game, on a 28nm HP fab. process too! And this is just load power in one game where 750 Ti is bandwidth-limited.

Yep and what if you measure efficiency during Company of Heroes 2 instead of Crysis 3? This is the problem with only having one measurement of one game then attempting to take efficiency from that (which is exactly what TPU does).

Or what if you use Guru3d's power numbers?

That doesn't look very good does it?

Tom's Hardware did an analysis of avg. power consumption and performance across multiple games using 750 Ti, and the results were insanely good:

As I mentioned earlier, that is 38% better than the 260X over an average of 5 games (so a much better sample). The 260X is an overclocked version of an already hard pushed and not very efficient card (the 7790).

The 260X gains only 5% performance over the 7790 but loses a lot more in efficiency. It's not hard to see that the actual efficiency over the 7790 (an already inefficient card - and I remember plenty of people here saying that Bonaire was nothing special in efficiency in the past) will only be around 25%-30%.

So there's your "common sense". Maybe try using it yourself before suggesting others do?

 

[Keysplayr]

Yep and what if you measure efficiency during Company of Heroes 2 instead of Crysis 3? This is the problem with only having one measurement of one game then attempting to take efficiency from that (which is exactly what TPU does).

Or what if you use Guru3d's power numbers?

That doesn't look very good does it?

As I mentioned earlier, that is 38% better than the 260X over an average of 5 games (so a much better sample). The 260X is an overclocked version of an already hard pushed and not very efficient card (the 7790).

The 260X gains only 5% performance over the 7790 but loses a lot more in efficiency. It's not hard to see that the actual efficiency over the 7790 (an already inefficient card - and I remember plenty of people here saying that Bonaire was nothing special in efficiency in the past) will only be around 25%-30%.

So there's your "common sense". Maybe try using it yourself before suggesting others do?

Not sure what you're attempting to convey.. Are you saying the 750Ti isn't power efficient? Is power efficient? Good performance per watt? Bad performance per watt?
The general consensus of conclusions from all these reviews seems to say "Excellent" or "Innovative". I haven't seen one that says "Meh". Unless you direct me to the ones you found.
Also not sure why you are putting some effort into making sure everyone thinks that the 750Ti isn't as good as it appears. Why can't you just look at it for what it is? I think saying it's impressive is a totally fair and not over the top descriptor. Wouldn't you?

 

[Aeiou]

I hope this isn't a sign that new high end will just be super power efficient for a sweet FA jump in power.

I want a huge jump.

 

[SiliconWars]

Not sure what you're attempting to convey.. Are you saying the 750Ti isn't power efficient? Is power efficient? Good performance per watt? Bad performance per watt?

No if you'd actually read my posts you'd see I have said...

Honestly it's a good card with amazing performance for the power draw

Sure it's a good card and Maxwell shows a lot of promise

The 750 Ti is a nice card

obviously if staying around 65W total card draw is a must then the 750 Ti is the best choice by far

Is that enough or need I be more flattering in order to meet some standard considered worthwhile for this forum?

The general consensus of conclusions from all these reviews seems to say "Excellent" or "Innovative". I haven't seen one that says "Meh". Unless you direct me to the ones you found.

Maybe you can start by directing me to where I said the card was "Meh", and we'll go from there.

Also not sure why you are putting some effort into making sure everyone thinks that the 750Ti isn't as good as it appears. Why can't you just look at it for what it is? I think saying it's impressive is a totally fair and not over the top descriptor. Wouldn't you?

Yes believe it or not that's pretty much what I've been saying. You could always "put some effort" into reading my posts first. Properly this time.

 

[Keysplayr]

No if you'd actually read my posts you'd see I have said...

Is that enough or need I be more flattering in order to meet some standard considered worthwhile for this forum?

Maybe you can start by directing me to where I said the card was "Meh", and we'll go from there.

Yes believe it or not that's pretty much what I've been saying. You could always "put some effort" into reading my posts first. Properly this time.

Thank you!! I just wanted to be clear because the mixed signals were flying. Much obliged.

 

[TeknoBug]

Step up?! I was thinking it would be a step down or a sidegrade at best

Edit: Did some diggin around and found a benchmark that features the GTX 560, GTX 650Ti, and R260x. 650Ti/R260x should be on par with the performance of the 750Ti, so it should be a valid comparison.

Oh thanks, nice find. With the 560Ti there's no access to Shadowplay and such, and the 750Ti would have less power consumption which is the motive here and still being able to play games casually (like Planetside 2 or F1 or NASCAR series games hooked to the TV as an HTPC).

 

[Magic Carpet]

Is this going to feature ARM processors?

 

[ShintaiDK]

Is this going to feature ARM processors?

ARM is most likely Telsa only if any at all. No ARM in GM107.

Not that it mattered anyways. And I dont think I have seen anything official related to Maxwell dGPUs supposed to have ARM chips. Might simply be a misunderstanding with the Tegra chips.

 

[coffeejunkee]

This is the problem with only having one measurement of one game then attempting to take efficiency from that (which is exactly what TPU does).

Now that you mention it:

"Average: Crysis 2 at 1920x1080, Extreme profile, representing a typical gaming power draw. Average of all readings (12 per second) while the benchmark was rendering (no title/loading screen)."

is what they use for power consumption. Sadly there are no Crysis 2 benchmarks but if it's anything like Crysis 3 the Radeon cards perform pretty poorly at that (or rather the 750Ti is really good at it, faster than GTX 660 according to tpu). Obviously 750Ti still has much better perf/watt but on average it shouldn't be so unreal.

Btw, Bonaire is/was actually pretty efficient. The 260X is just clocked a bit too high and also uses higher voltage than normal 7790.

 

[SiliconWars]

Bonaire isn't "inefficient" as such I suppose - but obviously a part like this is a bit top-heavy and with a 1 GHz clock it's approaching the area where AMD cards seem to fall off the cliff a little in efficiency terms. That's why the 7750 and 7850 showed better efficiency characteristics than the 7770 and 7870, even though the former were salvage parts.

An 800-900 MHz Bonaire would be more efficient again, I guess. Also even slower of course, but it would still narrow the efficiency gap more to this 750 Ti.

Obviously AMD would have some work to do to catch Maxwell, but a year is a long time on this node as we've seen previously with Hawaii. That's the important thing, the extra year on the process.

It's a shame but I don't believe we'll see any more 28nm chips from AMD so it will remain an interesting "what if".

 

[sushiwarrior]

I think what Silicon is trying to say is that yes, the card that was designed from the ground up to be power efficient is power efficient, however considering that is the main goal, it makes the achievement somewhat less impressive? AKA. either company could make a card that efficient if they wanted to. You don't get FPS/$, you get FPS/W... while AMD is arguably more focused on FPS/$

 

[coffeejunkee]

Well, disregard what I said about Crysis 3 performance, the tpu perf/watt figures are based on all games average performance compared to wattage during Crysis 2. I actually thought they measured power draw for all games (not that hard with a good voltmeter which can show average power for a particular benchmark).

The 7790/260X get screwed a bit by the bad Batman scores, but for the other cards it should still be valid. Still, the difference can be much smaller for other games than Crysis 2. There's also the question of total system power consumption. I read for Maxwell scheduling is done more on the cpu than gpu, so that would obviously make the gpu consume less power. TPU uses only card power consumption for their perf/w figures.

For mining it's best to use 830Mhz for 7790 so what you say sounds right.

I think what Silicon is trying to say is that yes, the card that was designed from the ground up to be power efficient is power efficient, however considering that is the main goal, it makes the achievement somewhat less impressive? AKA. either company could make a card that efficient if they wanted to. You don't get FPS/$, you get FPS/W... while AMD is arguably more focused on FPS/$

I think we are trying to say the same: that 750 Ti is really very efficient, just not as much as the tpu bar would lead you to believe.

 

[B-Riz]

comparing this card with the GTX 480 is quite funny
http://www.anandtech.com/bench/product/1135?vs=1130

Funny and impressive on the 750 ti part! :biggrin: