NVIDIA GeForce GTX 650 Ti Boost SLI

[EarthDog]

No clue.. that sounds like it can be a source of the poor scaling though.

 

[0x4452]

Great FPS scaling indeed. Could you add a section in your reviews about latency & stutter though?

At best, there should be no stutter and latency should stay the same as the single GPU.

PS First time poster here - thanks for the great reviews and please keep taking pictures of the PCBs - my favorite part of the reviews

 

[Ikaruga]

PS First time poster here

Welcome to TPU

 

[Nortrop]

Looks impressive. Here where I live the projected price for two of these is actually way less than that of a mid range 670. On the other hand though the 7850s are quite impressive and boast ridiculous OC potential...

But, what about temperatures and noise? Current gen cards are very power efficient and cool, but those two things are still keeping me at bay ever since I parted ways with my first and only SLI setup (GTX 460s).

 

[Melvis]

This card is very very good, and i find it a strange and odd move by Nvidia as it made the rest of there range (GTX660 and above) pointless for the price/performance. I can get GTX 680 performance for $150 LESS then a GTX680, crazy!

 

[kroks]

sli vs single card.. show us crossfire hd7870/hd7850

 

[Sasqui]

sli vs single card.. show us crossfire hd7870/hd7850

I thought there were 7870 and 7850 CF reviews, but can't find...

7950 CF review: http://www.techpowerup.com/reviews/AMD/HD_7950_CrossFire/

 

[EarthDog]

I will have that direct comparison up in a couple weeks.

 

[Fluffmeister]

I will have that direct comparison up in a couple weeks.

The important bit will be the frame rate latencies.

 

[EarthDog]

Which we do not test for... FRAPS is borderline useless for that testing according to some articles and our site doesnt have the money to buy the proper equipment to test with either. Perhaps another site then... sorry.

 

[Fluffmeister]

Yeah that's fair enough, it's just that comparing FPS alone doesn't tell the whole story either.

Although it's commendable what you are doing.

 

[EarthDog]

Guru3D actually went and bought the equipment and didnt have much good to say...

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

Basically at the end he says 'we are doing it, cant see it anyways in most cases, so its more or less pointless'

 

[Fluffmeister]

It's hardly pointless when AMD have confirmed things aren't as they should be, and as such are working on a fix.

It benefits the users at the end of the day, although I appreciate a lot of people would rather stick their head in the sand.

 

[EarthDog]

No.. the FRAPS testing is kind of pointless... If we could afford it, we would do it. However we agree with guru3d's assessment of the testing as well.

 

[Ikaruga]

Guru3D actually went ....

Btw, Guru3D's EVGA GTX650ti boost SC review has an interesting point about the OC. They actually found that the Power Target is most efficient at 105% (with that particular sample ofc).
http://www.guru3d.com/articles_pages/evga_geforce_gtx_650_ti_boost_sc_review,22.html

I still wonder is this whole boost thing is there to limit the OC capabilities of the cards, so that a 650 user would not go much above 660s, and they would need to buy a better card if they want to?

 

[EarthDog]

That doesnt make sense to me.... I will have to read that article.

EDIT: I dont think it means what you took from it? That is one heck of a core clock though. I have 3 650ti Boosts and they all, no matter what power limit top out at 1202 or 1209 (or w/e the next jump up is). They are calling it a 'tweak' for some odd reason when its just a setting.

 

[Fluffmeister]

No.. the FRAPS testing is kind of pointless... If we could afford it, we would do it. However we agree with guru3d's assessment of the testing as well.

Agreed, FCAT moves things forward and has given benchmarking a much needed kick in the rump.

 

[Ikaruga]

That doesnt make sense to me.... I will have to read that article.

EDIT: I dont think it means what you took from it? That is one heck of a core clock though. I have 3 650ti Boosts and they all, no matter what power limit top out at 1202 or 1209 (or w/e the next jump up is). They are calling it a 'tweak' for some odd reason when its just a setting.

No, you misunderstood me. The interesting part for me in the story is that they still get around 10% gains like in all the other 650ti reviews everywhere else, regardless of the OC amount or the brand/sample (and the OC they got there is quite nice indeed). I wonder, is it possible that Nvidia just forcing a "maximum possible performance gain" on the users with the introduction of this new boost scheme?

 

[EarthDog]

Not sure... I doubt it though? Its tough to compare gains across different sites due to different settings used for the games. For example, if one site tests at 1080p Ultra while another does High the scaling from overclocking will be different.

 

[Ikaruga]

Not sure... I doubt it though? Its tough to compare gains across different sites due to different settings used for the games. For example, if one site tests at 1080p Ultra while another does High the scaling from overclocking will be different.

I'm usually also against conspiracy theories, but the plot thickens for me a bit here tbh:

Other GTX 650 Ti Boost cards that we tested before delivered much higher frequencies. What I find interesting though is that all three cards top out at the same maximum real-life performance levels—looks like each card uses a different boost configuration.

 

[EarthDog]

Remember though, boost depends on a couple of factors, power used and heat. So long as you are under 70C and under the power limit you should get the full boost bins (7?). I'm not sure how the power consumption is measured but what if its a leaky chip that uses more power so it will hit the limit earlier? Or the inverse, what if its not a leaky chip and you have more headroom?

I dont know.. just guessing a bit.

 

[Ikaruga]

Remember though, boost depends on a couple of factors, power used and heat. So long as you are under 70C and under the power limit you should get the full boost bins (7?). I'm not sure how the power consumption is measured but what if its a leaky chip that uses more power so it will hit the limit earlier? Or the inverse, what if its not a leaky chip and you have more headroom?

I dont know.. just guessing a bit.

You are right but everybody everywhere is getting the same gains regardless of the OC, like it's a hard wired limit that you can gain X percent with this card and no more whatever you do. You have to admit this is not the usual stuff. We use to have different performances with different samples and OC figures, yet - with this card - , 100Mhz difference gives you nothing, and I find that interesting and also a little frightening tbh

 

[Ders33]

Excellent

If you compare gtx 650ti sli and titan, sli combination is gangbang for the buck ) and performance is very close to titan and cost waay more less than overpriced titan so considering to this test i should say that gtx 660 sli performs like titan and again with sli you get high performance like you get with 1000$ dollar titan

 

[Ikaruga]

If you compare gtx 650ti sli and titan, sli combination is gangbang for the buck ) and performance is very close to titan and cost waay more less than overpriced titan so considering to this test i should say that gtx 660 sli performs like titan and again with sli you get high performance like you get with 1000$ dollar titan

Comparing the awesome 650TIB's SLI performance with the Titan's is good, but comparing the price is not imo.
Titan is for wealthy enthusiasts, who doesn't need to worry about money. If you take the cost out of the evaluation, you can see that it's really something special. It's a single GPU, so its power will show in every little good game and application and not just in the "mainstream" titles. I recently had the opportunity to spend some time with that beast and I fell in love in seconds.

 

[Dave65]

Good review..
Wish there would have been a comparison to 660s in SLI..
I got 2 of them and am quite impressed.