• We've upgraded our forums. Please post any issues/requests in this thread.

MSI GeForce GTX 650 Ti 1024 MB

Joined
Mar 31, 2012
Messages
541 (0.26/day)
Likes
113
System Name SIGSEGV
Processor Intel | AMD
Motherboard INVALID| X399
Cooling 2 Heatpipes | EK
Memory 16 Gb DDR4 2400 | 32 GB DDR4
Video Card(s) 1060 | MI25 / Vega Frontier | Tesla K
Storage 1TB 7200/256 SSD PCIE | ~ TB
Display(s) 15,5" / 27" LEDTV | No Display
Case Black & Grey
Audio Device(s) Realtek
Power Supply Silverstone 850W | Unrated PSU
Software LinuxMint KDE | XFCE
Benchmark Scores i dont care about scores
#26
Also, additional stuff. I think tpu should add benchmarks only containing the 650 Ti (or whatever the card is) which is given as sample from nvidia or non-OC'd card so People wouldn't get confused about the result. Also, why is the 660 beating the 7870? I don't think it used to be like that.
true story :toast:
 
Joined
Jun 6, 2012
Messages
92 (0.05/day)
Likes
15
#28
Sorry wizzard, I didn't knew that. That's actually good mate :).

Power consumption... Hmm, that's a pretty good arguement. But from the experiences that I had, I rarely saw a guy who can't have a 7850 but can put a 650 Ti because of the power limitation. People that has power limitation for them would probably get a 7750. Which don't need an extra PCI-E cable.

I mean, take a look at the 7770. In terms of power it's pretty good, it's $30 less and most people that are in this segment won't find a huge difference in terms of performance. And, with $20 more you can get a much better card in terms of performance, which is the 7850. Yes, your power bills might be more, but I mean it's the risk! You can't have a 680 with low electricity bills. Don't wan't to pay high bills? Then I personally I'll get the 7770.

From my experience, someone's budget is often flexible, so why not pay more for a boost on performance?

I have to agree with nvidia's features. They're pretty good but not all people have to have it.
 
Joined
Oct 10, 2012
Messages
2 (0.00/day)
Likes
0
#29
Just guessing, but perhaps he used a different scene for the benchmark?
Yes in first case they used 12.7 beta drivers, and for this review they used 12.8. Amd drivers fail XD

Latest branch of the Nvidia drivers are adding some noticeable extra performance. Not sure, but perhaps that's the cause.

Hate to say it, but I was wrong and many other people were right: the card is clearly bottlenecked by memory. And because of it generalyy slow. Great power and noise though.

We can see it in the amazing OC results of this card (greater than core OC) compared to the Asus one. GTX 660's shader performance is only 25-30% higher, but actual gaming performance gain is over 50%, much much more than I expected.

BTW amazing performance gain across the board for Nvidia cards with 306.23 WHQL drivers. Anyone noticed that GTX660 is now just above HD7870, it was 5-10% slower with 304's.
If you pay attetion the 7870 gets less fps in many game, in connection with previous reviews

For example, in the review of 7870 powercolor ( http://www.techpowerup.com/reviews/Powercolor/HD_7870_PCS_Plus_Vortex_II/1.html )

Alan wake (before) 63.9 in this review 60.9 Difference = 3 fps (4.7% less)
Crysis 2 (before) 47.8 (now) 43.9 Difference = 3.9 fps (8.2% less)
Diablo III (before)145 (now) 98.2 Difference = 46.8 (32.7% less)
Dragon Age II (before) 44 (now) 33.5 Difference = 10.5 fps (23.9% less)
Starcraft (before) 137.7 (now) 119.6 Difference = 18.1 fps (13.1% less)
Skyrim (before) 70.3 (now) 65.4 Difference = 4.9 fps (7% less)

The difference is in the drivers, the catalyst 12.8 are very bad drivers. I can't find other explanation.


(sorry for my english)
 
Joined
Apr 19, 2011
Messages
1,725 (0.71/day)
Likes
203
Location
So. Cal.
#30
W1zzard doesn't review prices.
Really not in W1zzards "Performance/$" or the "Value and Conclusion" Right... Keep up the smoke!
the biggest jump in performance when compared to the card that it replaces (550 Ti)
That’s not saying much the bar was really Low! :rolleyes:
In the equation the only bad thing about it is the price and price doesn't make a product any worse or better.
So, you'd have no problem giving Nvidia $300?
GTX 650 ti consumes on average 63w, HD6870 consumes 119w and 560 Ti 148w. That's a huge difference, after 3 years a 560 Ti would end up costing 250 € more than the 650 Ti.
So "on average" you spend every day gaming 24/7? How do you pay for all that? On average a 650Ti is 3W lower than a 7850 at idle that's nifty. Although you ignore that most computers are set to sleep probably 75% of a day. At which point Radeons switch to an ultra-low-power state when the display goes to sleep and Kepler gets used to wipe the floor.

And, yes absolutely over older products it matters and improves with evolution, and 28Nm which makes that possible is a more costly this time around. While some would be influenced to upgrade, you can't live in a vacuum... You buy when/because you no longer are enjoying the experience you want. Spending money to only to save on power, while not completely improving your experience is never going to provide any payback if it really about gaming... its' just throwing good money after bad.

Anyone contemplating any of todays previous generation cards and especially Fermi is definitely better off with newer hardware. And, that brings us back to price… Nvidia can’t won’t lower this GTX650Ti while there are ton of GTX550Ti – GTX560 – GTX560Ti to flush out of the channel, and nobody smart should want them... So here we sit!
 
Joined
Apr 19, 2011
Messages
1,725 (0.71/day)
Likes
203
Location
So. Cal.
#31
The difference is in the drivers, the catalyst 12.8 are very bad drivers. I can't find other explanation.
Good catch... but really drivers there's no way in hell it's just drivers. Sure AMD/Nvidia would forgo 2-3% to fix a problem, but was there any complaints on Diablo III (32.7% less) or Dragon Age II (23.9% less) that AMD had to forfeit that much of a performance delta? Heck I could understand if AMD cut Diablo III, it's still plenty playable at 98FpS, but Dragon Age II down to 33FpS? Sorry there's something more. :confused:
 

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
17,057 (3.44/day)
Likes
17,962
Processor Core i7-4790K
Memory 16 GB
Video Card(s) GTX 1080
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 7
#32
but Dragon Age II down to 33FpS? Sorry there's something more.
i was wondering about that too, but it seems consistent across all cards. i retested dragon age ii just to be sure

starcraft and diablo have released new patches recently, maybe that affected the cards
 
Joined
Sep 11, 2009
Messages
2,680 (0.89/day)
Likes
693
Location
Reaching your left retina.
#33
Really not in W1zzards "Performance/$" or the "Value and Conclusion" Right... Keep up the smoke!
So? Again, is the product worse or better according to price? Nope. Should W1zzard review every card again, each time they get a price cut and change the score? More so since I think W1zz said that he uses lowest Newegg price to compare, he should rebench and adapt the score everytime there's a sale in Newegg?

Performance/$ and the conclusions is a very different thing to how good a card is. If it costed $50 you would be saying "Oh wow amazing card bla bla"? The card would be the same, performance, consumption and noise the same. So why should the score be different in that case than in this case? It shouldn't plain and simple.

That’s not saying much the bar was really Low! :rolleyes:
Not really. Not lower than other Fermi cards (or AMD cards) in respective segments. So no.

So, you'd have no problem giving Nvidia $300?
Who said that? I didn't. Read above.

So "on average" you spend every day gaming 24/7? How do you pay for all that? On average a 650Ti is 3W lower than a 7850 at idle that's nifty. Although you ignore that most computers are set to sleep probably 75% of a day. At which point Radeons switch to an ultra-low-power state when the display goes to sleep and Kepler gets used to wipe the floor.
Maybe you could read instead of imagine what I said. Because I already said that idle and shutdown times are included in the calculus:

I play and work on my PC and the result includes calclations for when it's idle and shut down
If it was on load 24/7 it would probably amount to 5 € ++ per watt per year. And I think that sleep mode is for lazy people. If I'm not going to be using my PC in a time, I shut it down.
 
Joined
Feb 18, 2011
Messages
1,240 (0.50/day)
Likes
503
#34
Alan wake (before) 63.9 in this review 60.9 Difference = 3 fps (4.7% less)
Crysis 2 (before) 47.8 (now) 43.9 Difference = 3.9 fps (8.2% less)
Diablo III (before)145 (now) 98.2 Difference = 46.8 (32.7% less)
Dragon Age II (before) 44 (now) 33.5 Difference = 10.5 fps (23.9% less)
Starcraft (before) 137.7 (now) 119.6 Difference = 18.1 fps (13.1% less)
Skyrim (before) 70.3 (now) 65.4 Difference = 4.9 fps (7% less)

The difference is in the drivers, the catalyst 12.8 are very bad drivers. I can't find other explanation.
It's a nice catch, you are probably right, well done. The old results would put AMD in a much better position.


(sorry for my english)
Don't worry about it, (mine is terrible too:B), there is no language barrier as long as we can understand each other, and let's hope that the native English speakers will forgive us.;)
 

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
17,057 (3.44/day)
Likes
17,962
Processor Core i7-4790K
Memory 16 GB
Video Card(s) GTX 1080
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 7
#35
the latest bench is running 100 mhz slower cpu clock, so less cooling is required, so i can sleep at night
how are nvidia results affected?
 

W1zzard

Administrator
Staff member
Joined
May 14, 2004
Messages
17,057 (3.44/day)
Likes
17,962
Processor Core i7-4790K
Memory 16 GB
Video Card(s) GTX 1080
Display(s) 30" 2560x1600 + 19" 1280x1024
Software Windows 7
#36
HD 7970 GHz Edition, Catalyst 12.8:
Code:
===== DragonAge2 =====
Frames per second: 62.2 FPS
Total Frames: 5619 

===== DragonAge2 =====
Frames per second: 49.8 FPS
Total Frames: 4499 

===== DragonAge2 =====
Frames per second: 45.4 FPS
Total Frames: 4103 

===== DragonAge2 =====
Frames per second: 35.7 FPS
Total Frames: 3225 
Benchmark completed.
Uninstalled 12.8, rebooted, installed 12.6, rebooted:

Code:
===== DragonAge2 =====
Frames per second: 101.7 FPS
Total Frames: 9188 

===== DragonAge2 =====
Frames per second: 75.5 FPS
Total Frames: 6815 

===== DragonAge2 =====
Frames per second: 65.1 FPS
Total Frames: 5881 

===== DragonAge2 =====
Frames per second: 46.4 FPS
Total Frames: 4189 
Benchmark completed.
 
Joined
Apr 19, 2011
Messages
1,725 (0.71/day)
Likes
203
Location
So. Cal.
#37
product worse or better according to price? Nope.
Just when they are in the real light of buyer in the market at that time. Today is today and if you're buying that's the only relevance. By Black Friday these could go the way of the GTX460 768Mb, but today is today while i think there's hope!

Not lower than other Fermi cards
Not true the GTX460 768Mb had it irrelevant even on the day it came out!

HD 7970 GHz Edition, Catalyst 12.8:
Ok, ouch I stand corrected... :toast:
 
Last edited:
Joined
Feb 14, 2012
Messages
1,682 (0.79/day)
Likes
578
System Name msdos
Processor 8088
Motherboard mainboard
Cooling passive
Memory 640KB + 384KB extended
Video Card(s) EGA
Storage 5.25"
Display(s) 80x25
Case plastic
Audio Device(s) modchip
Power Supply 45 watts
Mouse serial
Keyboard yes
Software disk commander
Benchmark Scores still running
#38
That's a huge difference, after 3 years a 560 Ti would end up costing 250 € more than the 650 Ti.
Well ... if you game 24/7 that is. So it's probably more like 25€ at 2.4 hours a day, which is still high probably. This is really the least of concerns. I'd mention hot rooms in the summer before electricity.
 
Joined
Sep 11, 2009
Messages
2,680 (0.89/day)
Likes
693
Location
Reaching your left retina.
#39
Well ... if you game 24/7 that is. So it's probably more like 25€ at 2.4 hours a day, which is still high probably.
The calculations were for MY case, but I've stated several times that it is NOT for 24/7 gaming (I don't know why everyone would think that, and why you don't read the previous posts). It's for a mix that includes 3 hours gaming, 8-10 hours working (3D modeling + testing on some real-time engine), a few hours of iddling and a little bit of "off-line" rendering on some nights. The actual time the GPU is loaded would amount to 6 hours or so, because 3D modeling, etc does not load it as much as games, so that's what I calculated anyway.

Simplifing my calculations a lot, it was something like:

For every 1w as seen in W1zzard's chart:
1w --> ~1.5w after accounting for 80% PSU efficiency and the power factor of the house, etc. Remember we are talking about how much it costs to you in the end so it's important to know exactly how much you'll endup paying for every watt on the GPU. The electricity company is not going to charge for clean DC watts provided by the PSU. Not even for the aparent consumption. They'll charge you for the actual consumption on their end.

1.5w * 365 days * 0.00025 €/w*h = 0.137 €/h

Now the only thing left is to calculate the hours of use. In my case the active time (heavy load on GPU) is roughly 6 hours so:

0.137 * 6 = 0.822 € per watt per year

This is when the GPU is fully loaded*, to this we need to add the spending when on idle. Which is less but it probably brings it close to 1€, so roughly 1€.

For someone who stresses the GPU only 2 hours it would be less. And the lucky pleople in areas where they pay 5 cents it will be a lot less too.

* Loaded as in playing a game, i.e Crysis, not 100% as in Furmark, that is, the chart that W1zz labels as "average consumption".

I'd mention hot rooms in the summer before electricity.
Spending on air conditioner derived from higher temps was included in the calculation I made 2 years ago. It's more than people would think.
 
Last edited:
Joined
Aug 3, 2008
Messages
1,644 (0.48/day)
Likes
344
System Name Black Killing Machine
Processor Delidded NO IHS mount Intel i7-4770k
Motherboard Gigabyte Z87X-OC
Cooling Swiftech DIR655 pump, Watercool HEATKILLER® GPU-X³ 79X0 Ni-Bl gpu, DT SNIPER CPU block, UT60 420 RAD
Memory Gskill Trident 2400mhz CL9 Samsung IC9
Video Card(s) SAPPHIRE HD7950 OC 950mhz Edition, VaporX HD7950
Storage 2xM4 Crucial 64GB RAID 0, 1 OCZ AGILILITY 3 60GB, WDCB 500GB x2RAID0, WD Green, Hitachi 1TB
Display(s) Samsung SyncMaster 226BW and 24" 120hz BenQ gamer
Case SilverStone Raven rv-02
Audio Device(s) Audio Technica ATH-AD900 headphones, Topping D2 DAC/Headphone AMP
Power Supply Seasonic Platinum 860
Software Windows 7 Pro 64bit
Benchmark Scores http://hwbot.org/user/sonda5/ Delidding is magic. http://valid.canardpc.com/2878462
#40
How did the triple voltage control over clocking go with MSI Afterburner?
 
Joined
Apr 19, 2011
Messages
1,725 (0.71/day)
Likes
203
Location
So. Cal.
#41
Now just a week away from Black Friday and Eggs got it for $135 -AR$15 with F/S much more substancial reduction that I had thought... for 6 weeks.
 
Joined
Feb 18, 2011
Messages
1,240 (0.50/day)
Likes
503
#42
Now just a week away from Black Friday and Eggs got it for $135 -AR$15 with F/S much more substancial reduction that I had thought... for 6 weeks.
This card is an overclocking beast when it comes to OC-ing the memory. I was asking around about this card in the last few days, and many who I know managed to get stable (24/7) 1700-1800Mhz-ish memory clocks with this card which is a huge huge speed increase, and it's exactly what you need with the 128bit wide bus to shrink the performance gap towards the 660.
I wonder why Wizzard got such a "low" memory OC result with his sample.