GeForce GTX 650 Ti Review: Kepler Hits $150

Julio Franco

Posts: 9,090   +2,042
Staff member
Read the full article at:
[newwindow=https://www.techspot.com/review/583-nvidia-geforce-gtx-650-ti/]https://www.techspot.com/review/583-nvidia-geforce-gtx-650-ti/[/newwindow]

Please leave your feedback here.
 
I dont think that there is any need of such a cooling solution for a sub $200 card.
its pointless............
 
Surely that's not a VGA port??

The 7850 has also come down in price recently, although it's still a bit more expensive than this 650Ti. But given it's overclocking potential I think that would be a better buy. The 7770 was never really a great performance/price card.
 
Hmmmm, no GTX 650 included, let me go dig that review up and compare numbers.

Hard to believe the card directly under it in the same series was not included. Anyone looking at benchmarks for the GTX 650Ti, does not really care how it compares to the 680. There was a huge gap in performance between the GTX 650 and the GTX 660, where does the 650Ti fall in?
 
I'm thinking about replacing my PNY NVIDIA 250 GTS (1GB GDDR3 DX10) with something soon. I really want DX11 and the fan on my card is insanely loud. Maybe I'll get the 650 Ti...
 
Nice review but pretty useless benchmarks:

Where is the GTX 650 or even GTX 560/460 (previous budget cards)?

The GTX 650Ti is always on the botton of the graphs, so this information is useless for anyone who owns a previous generation card like the GTX 460 looking for an affordable upgrade. You are just showing that it's slower than the superior models? Yeah that's useful... :/
 
The reviewer mentioned that MSRP for HD7770 is $139 but it has been dropped to $119 last week:
http://www.xbitlabs.com/news/graphi...ng_of_Radeon_HD_7700_7800_Graphics_Cards.html

HD7770 = $119 MSRP
HD7850 1GB = $169 MSRP ($159 with MIR by partners)
HD7850 2GB = $189 MSRP

GTX650Ti 1GB = $149 MSRP

However, 2GB of the 650Ti are going for $160-170.

This card is overpriced in relation to HD7770 that despite $119 MSRP can be regularly found for $105-115 and in relation to the 25-30% faster HD7850 cards that for $20-30 more deliver exponential increase in performance/$.

Other than the gaming bundle, NV should have launched this at $129 or at best $139.

Take 7850's overclocking into account and it ~ HD7870/GTX660/GTX580 in performance. GTX650Ti is outclassed and $150-160 market prices for it are too high at the moment.
 
U guys should do these tests with vsync off. my hd5770 hits better frames in some of the games youve tested compared to higher end cards. vsync and shadows are fps rapists.
 
HD7770 = $119 MSRP
HD7850 1GB = $169 MSRP ($159 with MIR by partners)
HD7850 2GB = $189 MSRP
GTX650Ti 1GB = $149 MSRP
Take 7850's overclocking into account and it ~ HD7870/GTX660/GTX580 in performance.
The only problem with cherry picking examples (I.e. price) and cherry picking performance ( assumed OC) is that you end up with a skewed data set. Example's from very recent reviews:
HT4U's HD 7850 review. Of the three 7850's, two - the Sapphire (16% overclock o/ref., 11% over shipping clock) and XFX (19% OC o/ref., 7% over shipping clock) were pretty average overclockers. The PC PCS+ produced respectable OC potential, but also retails at $205 after rebate.

MSI Power Edition, also produced a respectable OC... 22%-31% over reference (10.5%-18% over shipping clock)...but also retails for $220
Card vendors bin both AMD and Nvidia GPU's according to the end price they can extract, and now with the HD 7000 being a mature series, you're seeing multiple graduations of pricing and SKU's from each vendor. Expecting a massive OC from a budget reduced BoM card likely featuring the lowest percentile ASIC is unrealistic.
 
U guys should do these tests with vsync off. my hd5770 hits better frames in some of the games youve tested compared to higher end cards. vsync and shadows are fps rapists.

So wait, your saying any Graphics card can run at 60fps+ as long we all turn the graphics down?

Are you seriously suggesting that these tests should be modified so that the graphics settings are lower on the cheaper cards to make the fps the same? because I'm pretty sure that completely defeats the point of every single graphics card review out there...
 
Dividebyzero,

As usual you have decided to focus on 1 part of my post and missed the other points I made. I'll repost again since it appears you have missed my point entirely

Stock vs. Stock performance: GTX650Ti loses badly in price/performance to either HD7770 or HD7850.

Let's take a look at stock performance again. I'll use 5 websites to eliminate bias:

1) Computerbase: http://www.computerbase.de/artikel/grafikkarten/2012/test-nvidia-geforce-gtx-650-ti/6/

GTX650Ti is only 13% faster at 1080P than HD7770 but costs 25% more based on MSRP ($119 vs. $149). Market pricing for HD7770 is actually $105-115 on Newegg. GTX650Ti is very much overpriced in this comparison.

HD7850 1GB is 34% faster on average at 1080P than GTX650Ti. MSRP for HD7850 1GB is $169 ($159 with option partner rebates). Market prices for HD7850 1GB are $160-175 on Newegg. Again for 7-17% higher price of HD7850 1GB, you get 34% higher performance.

Let's see if other reviewers show a similar scenario:

2) TechPowerup has HD7850 32% faster than GTX650Ti (and 27% faster than factory preoverclocked 650Ti):
http://www.techpowerup.com/reviews/MSI/GTX_650_Ti_Power_Edition/28.html

3) Hardware.fr shows a 37% advantage for HD7850 against the 650Ti, while 650Ti only leads HD7770 by 12%.
http://www.hardware.fr/articles/879-19/recapitulatif-performances.html

4) Tom's Hardware's test is especially punishing as unlike TechSpot, they used MSAA and cranked settings to demanding levels. HD7850 is 42% faster than GTX650Ti is at high details, while GTX650Ti is just 9-10% faster than HD7770:
http://www.tomshardware.com/reviews/geforce-gtx-650-ti-benchmark-gk106,3318-19.html

Their testing is especially useful since we see that in GPU demanding games like
BF3, Crysis 2, Metro 2033, Dirt Showdown, SKYRIM, HD7850's minimum framerates are often at the same level as GTX650Ti's average! That's remarkable considering HD7850 only costs $20-30 more.

5) What about the factory pre-overclocked GTX650Ti 2GB version? It still performs awfully compared to HD7850 based on Bit-Tech's testing:
http://www.bit-tech.net/hardware/graphics/2012/10/09/nvidia-geforce-gtx-650-ti-review/5

In the end, GTX650Ti is much closer in performance to HD7770 than it is to HD7850 but is priced much closer to HD7850. This is terrible value for budget gamers. To make matters worse, the 2GB version of this card goes for $170+, putting it a hair from HD7850 2GB.

Now, we come to overclocking. With neutered ROPs and memory bus width, GTX650Ti benefits much less from overclocking than HD7850. We already know from testing across the web that HD7850 has no problem reaching GTX660/HD7870 level of performance at just 1100mhz.
http://www.anandtech.com/show/5625/...-7850-review-rounding-out-southern-islands/19

That means with overclocking, HD7850 will only extend its lead over than overclocked 650Ti further, making it an even better value.

As it stands, GTX650Ti is overpriced at $149. Stock vs. stock or OC vs. OC, it loses to HD7850 and rather dramatically by 30-40% in GPU demanding titles. Against, HD7770, it's also horrendously overpriced since HD7770 doesn't sell for $139 but more like $105-115 in retail.
 
Dividebyzero,
As usual you have decided to focus on 1 part of my post and missed the other points I made
Probably because that part of your post was the only part I took issue with.
Yes, the GTX 560Ti is overpriced - please quote my post that said otherwise. It seems as though you enjoy creating a straw man argument so you can present a wailing wall of text
If you take issue with what I wrote then argue the points I made. Don't try to refute points I haven't made
I'll repost again..
TL: DC
Before you go off half-cocked on the review stats, and you should consider the information that follows as supplemental to your reading- you might note that at least a couple of sites are presenting HD 7850 figures based on 2GB cards, and at least one other site is just duplicating the scores from the 2GB and applying it to the 1GB card. How do I know this, you may ask. Cutting down the frame buffer on Pitcairn seems to impact the memory controller setup of the card- noteably swapping out textures. Note TR's Skyrim findings as example:
skyrim-fps.png

skyrim-99th.png


My issue with your post concerned your assertion that overclocking increases are somehow a given with the HD 7850. They are not. Next time I suggest you RTFP, and not use it as a vehicle for your own agenda.
 
It is interesting that you still say the 660 gtx offers better value than the 7870 when I can find new 7870 cards (HIS is one) for 205.00. This is less than I have found any 660 gtx for and offers comparable performance across the board.

I liked the 660 gtx I bought and tested but it was not quite on par with the 7870 (imho).

It seems as though the reviews of the nVidia cards run here run the "value" comparison based on AMD prices that are unrealistically high for when the test was published
 
Back