-
You heard it here first: Black Friday 2024 has arrived!
- Status
- Mar 7, 2003
- 12,012
- 0
- 40,790
Based on what yardstick?
40 FPS minimum is super playable, and if you want 60 FPS minimum turn down the settings.
It played every game we tested at 1080p and almost caught up with the 560/6870 without MSAA.
Who is going to play this at 720p on a 1080p monitor and destroy their image quality so they can use MSAA?
I totally disagree with you on this one, Ojas. 60 FPS minimum is a very high expectation.
cscott_it
Distinguished
- Jul 30, 2009
- 474
- 0
- 18,810
Most of the reviews I've been seeing are in line with this. AMD did a good job by dropping the price of the 7850. It seems like AMD is moving back towards the old ATI/Nvidia game. First one to the market with a new product charges the "new gen" premium until there is competition, drops the prices to make it a competitive offering, then uses the profits already made to cushion price drops to their advantage as the competition is rolling out their cards (since they already have the field).
From the murmurings I've heard, the 7850 price cut is an "unofficial" mandate.
From the murmurings I've heard, the 7850 price cut is an "unofficial" mandate.
luciferano
Honorable
- Sep 24, 2012
- 1,513
- 0
- 11,810
The 650 Ti is priced similarly to the 7850, a far superior card. The same is true for the 560. The 7770 has some highly factory overclocked models that can compete with the 560 in performance at a price of around $130-140, so they're recommendable if you don't want to spend $20-30 more for a 7850.
If you must have a Nvidia card, then the 650 Ti is okay, but AMD has superior and cheaper alternatives as well as similarly priced and far superior alternatives. I think that Nvidia users are better off paying a little more to get a GTX 660. It's a more balanced card than the 650 Ti.
nforce4max
Splendid
- Sep 9, 2009
- 8,337
- 0
- 31,460
Looking at another review it might be possible to squeeze out 1.2ghz on the core with a cooling mod to the power vrm but the card is extremely limited with its two phases for the gpu and one phase for the memory. Had it been 192bit and still $150 it would have been a much faster card for the price.
Overall for those who own a 68x0 or 460/560 256bit this card is not worth the sidegrade.
Overall for those who own a 68x0 or 460/560 256bit this card is not worth the sidegrade.
so... i was wondering... does this mean -
gtx 650ti is the most powerful gfx card with 128 bit memory bus,
gtx 660ti is the most powerful gfx card with 192 bit bus,
gtx 660 is the most powerful gfx card with single 6 pin pcie power connector?
i was hoping that the last one still belonged to radeon hd 7850....
in a different way, radeon hd 7850 seems to be currently the weakest 28nm gpu with 256 bit memory bus...
edit: i should detail further - i mean the reference models, not factory overclocked/customized. oc is not taken into consideration.
gtx 650ti is the most powerful gfx card with 128 bit memory bus,
gtx 660ti is the most powerful gfx card with 192 bit bus,
gtx 660 is the most powerful gfx card with single 6 pin pcie power connector?
i was hoping that the last one still belonged to radeon hd 7850....
in a different way, radeon hd 7850 seems to be currently the weakest 28nm gpu with 256 bit memory bus...
edit: i should detail further - i mean the reference models, not factory overclocked/customized. oc is not taken into consideration.
Ironslice
Honorable
- May 1, 2012
- 648
- 0
- 11,060
[citation][nom]De5_roy[/nom]so... i was wondering... does this mean -gtx 650ti is the most powerful gfx card with 128 bit memory bus,gtx 660ti is the most powerful gfx card with 192 bit bus,gtx 660 is the most powerful gfx card with single 6 pin pcie power connector? i was hoping that the last one still belonged to radeon hd 7850....in a different way, radeon hd 7850 seems to be currently the weakest 28nm gpu with 256 bit memory bus...edit: i should detail further - i mean the reference models, not factory overclocked/customized. oc is not taken into consideration.[/citation]
That's because nVidia's Kepler GPU's are much more powerful than AMD's GCN GPU's. All of nVidia's GPU's are being bottlenecked by the memory bus; some more than others, but it is still there. A fully unbottlenecked GK-104 is much more powerful than a Tahiti GPU, the same for the GK-106 and Pitcairn. The only reason these cards are competing is because nVidia bottlenecked them all with memory.
That's because nVidia's Kepler GPU's are much more powerful than AMD's GCN GPU's. All of nVidia's GPU's are being bottlenecked by the memory bus; some more than others, but it is still there. A fully unbottlenecked GK-104 is much more powerful than a Tahiti GPU, the same for the GK-106 and Pitcairn. The only reason these cards are competing is because nVidia bottlenecked them all with memory.
luciferano
Honorable
- Sep 24, 2012
- 1,513
- 0
- 11,810
[citation][nom]De5_roy[/nom]so... i was wondering... does this mean -gtx 650ti is the most powerful gfx card with 128 bit memory bus,gtx 660ti is the most powerful gfx card with 192 bit bus,gtx 660 is the most powerful gfx card with single 6 pin pcie power connector? i was hoping that the last one still belonged to radeon hd 7850....in a different way, radeon hd 7850 seems to be currently the weakest 28nm gpu with 256 bit memory bus...edit: i should detail further - i mean the reference models, not factory overclocked/customized. oc is not taken into consideration.[/citation]
Like Ironslice said, that's because they're all highly memory-bandwidth bottle-necked and the first two points aren't very important. They're like saying that the Fermi cards are the fastest per processing core per GPU frequency Hz, it doesn't matter. The 7850 can also overclock better than the 660 can, so the 660 is arguably the fastest stock card with only one six-pin PCIe power connector for additional power to the PCIe slot, but generally not the fastest overclocked card with only one six-pin PCIe power connector for additional power to the PCIe slot.
Like Ironslice said, that's because they're all highly memory-bandwidth bottle-necked and the first two points aren't very important. They're like saying that the Fermi cards are the fastest per processing core per GPU frequency Hz, it doesn't matter. The 7850 can also overclock better than the 660 can, so the 660 is arguably the fastest stock card with only one six-pin PCIe power connector for additional power to the PCIe slot, but generally not the fastest overclocked card with only one six-pin PCIe power connector for additional power to the PCIe slot.
technoholic
Distinguished
- Feb 27, 2008
- 800
- 0
- 19,160
Let's face it: Nvidia screwed up with Kepler. I'm not debating about the performance, they are all good performers and such. I am talking about wrong pricing...First 660Ti and now this
technoholic
Distinguished
- Feb 27, 2008
- 800
- 0
- 19,160
Comparing the results for BF 3 to that of Anand's shows the 7850 = 650 Ti, which is quite different to Toms, 7850 > 650 Ti. In Skyrim Anand shows the 6850 substantially lower than Toms.
Both use the same CPU, RAM and drivers - slight differences here and there, but not enough to warrant the differing results.
The settings they use differ quite a bit, so this must be the reason for the variation. Can someone in the know maybe explain why these different settings yield such different results?
Also, Metro 2033 doesn't seem to demonstrate the same differences, why not?
Both use the same CPU, RAM and drivers - slight differences here and there, but not enough to warrant the differing results.
The settings they use differ quite a bit, so this must be the reason for the variation. Can someone in the know maybe explain why these different settings yield such different results?
Also, Metro 2033 doesn't seem to demonstrate the same differences, why not?
"That's because nVidia's Kepler GPU's are much more powerful than AMD's GCN GPU's. All of nVidia's GPU's are being bottlenecked by the memory bus; some more than others, but it is still there. A fully unbottlenecked GK-104 is much more powerful than a Tahiti GPU, the same for the GK-106 and Pitcairn. The only reason these cards are competing is because nVidia bottlenecked them all with memory."
You should watch a fully unbottlenecked Tahiti (I mean non ROP starved, i.e. with 48 ROPs), it would completely destroy a GK104 (bottlenecked or not), it would be not even funny to watch. Think it this way, a Pitcairn GPU is like 25%-30% slower than GK104. Tahiti has 60%!! more shaders, yet it dont perform 60% faster. Why? cause while all functional units count increased ROPs stayed the same, they just cannot feed 2048 shaders
You should watch a fully unbottlenecked Tahiti (I mean non ROP starved, i.e. with 48 ROPs), it would completely destroy a GK104 (bottlenecked or not), it would be not even funny to watch. Think it this way, a Pitcairn GPU is like 25%-30% slower than GK104. Tahiti has 60%!! more shaders, yet it dont perform 60% faster. Why? cause while all functional units count increased ROPs stayed the same, they just cannot feed 2048 shaders
luciferano
Honorable
- Sep 24, 2012
- 1,513
- 0
- 11,810
[citation][nom]oomjcv[/nom]Comparing the results for BF 3 to that of Anand's shows the 7850 = 650 Ti, which is quite different to Toms, 7850 > 650 Ti. In Skyrim Anand shows the 6850 substantially lower than Toms.Both use the same CPU, RAM and drivers - slight differences here and there, but not enough to warrant the differing results.The settings they use differ quite a bit, so this must be the reason for the variation. Can someone in the know maybe explain why these different settings yield such different results? Also, Metro 2033 doesn't seem to demonstrate the same differences, why not?[/citation]
Anand's BF3 1920x1200 results were in the lower to middle 30FPS range, not that important IMO. They're second BF3 test had FXAA instead of te far superior MSAA, hence it worked out better for Nvidia in performance, but at the sacrifice of quality.
Anand's BF3 1920x1200 results were in the lower to middle 30FPS range, not that important IMO. They're second BF3 test had FXAA instead of te far superior MSAA, hence it worked out better for Nvidia in performance, but at the sacrifice of quality.
bawchicawawa
Distinguished
- Dec 27, 2011
- 350
- 0
- 18,810
[citation][nom]yialanliu[/nom]Awesome, finally something that mainstream budget users can afford from Nvidia, been waiting for ages![/citation]
for 10$ more you can get a 7850... Don't see why you would buy the 650 ti.
for 10$ more you can get a 7850... Don't see why you would buy the 650 ti.
luciferano
Honorable
- Sep 24, 2012
- 1,513
- 0
- 11,810
[citation][nom]GG112233[/nom]"That's because nVidia's Kepler GPU's are much more powerful than AMD's GCN GPU's. All of nVidia's GPU's are being bottlenecked by the memory bus; some more than others, but it is still there. A fully unbottlenecked GK-104 is much more powerful than a Tahiti GPU, the same for the GK-106 and Pitcairn. The only reason these cards are competing is because nVidia bottlenecked them all with memory."You should watch a fully unbottlenecked Tahiti (I mean non ROP starved, i.e. with 48 ROPs), it would completely destroy a GK104 (bottlenecked or not), it would be not even funny to watch. Think it this way, a Pitcairn GPU is like 25%-30% slower than GK104. Tahiti has 60%!! more shaders, yet it dont perform 60% faster. Why? cause while all functional units count increased ROPs stayed the same, they just cannot feed 2048 shaders[/citation]
I disagree. Increasing core count has always been an inefficient way of scaling performance when you get into high core counts and ROPs don't feed the cores, the cores feed the ROPs IIRC. That the 79xx cards have such incredible MSAA scaling implies that their ROPs are sufficient. I don't think that more ROPs would have helped.
I disagree. Increasing core count has always been an inefficient way of scaling performance when you get into high core counts and ROPs don't feed the cores, the cores feed the ROPs IIRC. That the 79xx cards have such incredible MSAA scaling implies that their ROPs are sufficient. I don't think that more ROPs would have helped.
If it meant to replace the GTX 460 1GB 256bit then why, AGAIN, didnt you benchmark it agaisnt it.
The GTX 460 is the card everyone bought and we are the ones looking to upgrade from it... I just dont understand why you cant keep that darn! card in your bechmark inventory and show how it does agaisnt 650TI 660 660TI
again, i find this very disappointing. i feel like you are being sponsored by nvidia and trying to trick us thinking that thing is a good card (when is clearly not)
The GTX 460 is the card everyone bought and we are the ones looking to upgrade from it... I just dont understand why you cant keep that darn! card in your bechmark inventory and show how it does agaisnt 650TI 660 660TI
again, i find this very disappointing. i feel like you are being sponsored by nvidia and trying to trick us thinking that thing is a good card (when is clearly not)
- Mar 7, 2003
- 12,012
- 0
- 40,790
I chose the 460 192-bit because it has been available longer, and it performs closely to the 560 SE which you can still get.
The 256-bit 460 has been gone for a long time. The 650 Ti isn't replacing something that hasn't existed in months.
If the 650 Ti is replacing anything it's the 550 Ti, and that card is too slow to bother with.
That's an interesting conclusion. How would the 460 256-bit be less of a 'trick' to make you believe the 650 Ti is good for the money?
Or are you suggesting Nvidia payed us to say "Are there any reasons to shy away from a $150 GeForce GTX 650 Ti? Only that there's a 1 GB Radeon HD 7850 selling for $30 more. We think that gamers with the extra money would do well to invest it in the notably-faster AMD card"
Your feelings don't commensurate with the evidence.
luciferano
Honorable
- Sep 24, 2012
- 1,513
- 0
- 11,810
A 460 256 bit would be between a 560 and the 460 192 bit, it's not hard to guess where it would be if you really want to. The 650 Ti is most certainly not replacing the 460 192 bit considering that it doesn't perform too differently. That'd be like saying that the 7770 replaced the 5830.
"I disagree. Increasing core count has always been an inefficient way of scaling performance when you get into high core counts and ROPs don't feed the cores, the cores feed the ROPs IIRC. That the 79xx cards have such incredible MSAA scaling implies that their ROPs are sufficient. I don't think that more ROPs would have helped."
Then tell me why cape verde to pitcairn scales almost perfectly, and not pitcairn to tahiti. Lack of resource balance is what really hurts tahiti in terms of performance and efficiency. Kepler does not do magic, GK104 is just more balanced than tahiti, yet I have to agree it might not be perfectly balanced as u said.
Then tell me why cape verde to pitcairn scales almost perfectly, and not pitcairn to tahiti. Lack of resource balance is what really hurts tahiti in terms of performance and efficiency. Kepler does not do magic, GK104 is just more balanced than tahiti, yet I have to agree it might not be perfectly balanced as u said.
luciferano
Honorable
- Sep 24, 2012
- 1,513
- 0
- 11,810
[citation][nom]GG1122333[/nom]"I disagree. Increasing core count has always been an inefficient way of scaling performance when you get into high core counts and ROPs don't feed the cores, the cores feed the ROPs IIRC. That the 79xx cards have such incredible MSAA scaling implies that their ROPs are sufficient. I don't think that more ROPs would have helped."Then tell me why cape verde to pitcairn scales almost perfectly, and not pitcairn to tahiti. Lack of resource balance is what really hurts tahiti in terms of performance and efficiency. Kepler does not do magic, GK104 is just more balanced than tahiti, yet I have to agree it might not be perfectly balanced as u said.[/citation]
I told you why and you didn't listen. As core count increases, the difference between core count jumps in performance gets smaller even for the same core count increase. It is like how when you overclock, power efficiency drops off as you need more and more voltage to get a stable overclock. This is a well-known phenomenon of scaling performance with core count increases.
GK104 is most certainly not as balanced as Tahiti. GCN is less power efficient in some cases (not all) because of the compute enhancements, not because of AMD's ROP count in Tahiti. AMD's consistently unnecessarily high voltage also plays a part in this.
EDIT: I'm not saying that you can't be right, just that you aren't necessarily right and shouldn't assume that you're right because there are other plausible explanations.
I told you why and you didn't listen. As core count increases, the difference between core count jumps in performance gets smaller even for the same core count increase. It is like how when you overclock, power efficiency drops off as you need more and more voltage to get a stable overclock. This is a well-known phenomenon of scaling performance with core count increases.
GK104 is most certainly not as balanced as Tahiti. GCN is less power efficient in some cases (not all) because of the compute enhancements, not because of AMD's ROP count in Tahiti. AMD's consistently unnecessarily high voltage also plays a part in this.
EDIT: I'm not saying that you can't be right, just that you aren't necessarily right and shouldn't assume that you're right because there are other plausible explanations.
arcticfury
Honorable
- Oct 9, 2012
- 8
- 0
- 10,510
650 Ti = Roughly $155-160 on NewEgg.
7850 = $180 *before* $20 rebate
Only game it performs on equal footing is Battlefield 3; everywhere else, once you hit 1980x1200/1080, the 650 Ti performs 30-50% worse than a stock 7850.
So, using the cheapest prices currently on NewEgg, [not including rebates on either card] the 650 Ti costs currently 14% less than the 7850 while under-performing by 30-50% at a standard 1980x1200 resolution.
Unless I absolutely could not spend an extra $20-25, why would I possibly buy a 650 Ti?
7850 = $180 *before* $20 rebate
Only game it performs on equal footing is Battlefield 3; everywhere else, once you hit 1980x1200/1080, the 650 Ti performs 30-50% worse than a stock 7850.
So, using the cheapest prices currently on NewEgg, [not including rebates on either card] the 650 Ti costs currently 14% less than the 7850 while under-performing by 30-50% at a standard 1980x1200 resolution.
Unless I absolutely could not spend an extra $20-25, why would I possibly buy a 650 Ti?
hehe i know i know i did pushed it with that last part of my comment but it was to make sure i get a response
anyway... i still think you should include the gtx 460 in your benchmark... The GTX 460 was good enough to be kept for at least 2 years.... and it was as popular as the 8800GT back in time so alot of gamers went with this card.
i just dont get why you never benchmark the new mainstream cards against it... you should be doing this for people who want to upgrade
Just look at the Steam videocard usage chart
http://store.steampowered.com/hwsurvey
The GTX 460 comes in third place (if you dont look at those Intel cards). and those with the 560ti wont upgrade now (most of them). SO basically, those new cards benchmark should be made for us first
You get what i mean??
anyway... i still think you should include the gtx 460 in your benchmark... The GTX 460 was good enough to be kept for at least 2 years.... and it was as popular as the 8800GT back in time so alot of gamers went with this card.
i just dont get why you never benchmark the new mainstream cards against it... you should be doing this for people who want to upgrade
Just look at the Steam videocard usage chart
http://store.steampowered.com/hwsurvey
The GTX 460 comes in third place (if you dont look at those Intel cards). and those with the 560ti wont upgrade now (most of them). SO basically, those new cards benchmark should be made for us first
You get what i mean??
luciferano
Honorable
- Sep 24, 2012
- 1,513
- 0
- 11,810
The 560 is almost exactly 25% faster in the average on the last page. The 460 256 bit would be right between there, almost exactly as fast on average as the 650 Ti. However, the 650 Ti has two outlier games, so without them, it would perform on average more like the 460 192 bit and the 460 256 bit would be slightly better. See that? Simple logic did almost everything that Tom's spending much more time on this would have done for you. I'm not mocking you, I'm just saying that it isn't difficult to figure out were the 460 192 bit sits in this.
- Status
TRENDING THREADS
-
-
Review AMD Ryzen 7 9800X3D Review: Devastating Gaming Performance- Started by Admin
- Replies: 202
-
Discussion What's your favourite video game you've been playing?- Started by amdfangirl
- Replies: 4K
-
-
Facebook
Twitter
Instagram